arXiv subjects

Meng Li

Publications and source records attributed to Meng Li.

At least 181 records · Page 10Linked to original sources

arXiv2025

Optimal Transport-Based Token Weighting scheme for Enhanced Preference Optimization

Direct Preference Optimization (DPO) has emerged as a promising framework for aligning Large Language Models (LLMs) with human preferences by directly optimizing the log-likelihood difference between chosen and rejected responses. However, existing methods assign equal importance to all tokens in the response, while humans focus on more meaningful parts. This leads to suboptimal preference optimization, as irrelevant or noisy tokens disproportionately influence DPO loss. To address this limitation, we propose \textbf{O}ptimal \textbf{T}ransport-based token weighting scheme for enhancing direct \textbf{P}reference \textbf{O}ptimization (OTPO). By emphasizing semantically meaningful token pairs and de-emphasizing less relevant ones, our method introduces a context-aware token weighting scheme that yields a more contrastive reward difference estimate. This adaptive weighting enhances reward stability, improves interpretability, and ensures that preference optimization focuses on meaningful differences between responses. Extensive experiments have validated OTPO's effectiveness in improving instruction-following ability across various settings\footnote{Code is available at https://github.com/Mimasss2/OTPO.}.

cs.CL↗

arXiv2025

FPQVAR: Floating Point Quantization for Visual Autoregressive Model with FPGA Hardware Co-design

Visual autoregressive (VAR) modeling has marked a paradigm shift in image generation from next-token prediction to next-scale prediction. VAR predicts a set of tokens at each step from coarse to fine scale, leading to better image quality and faster inference speed compared to existing diffusion models. However, the large parameter size and computation cost hinder its deployment on edge devices. To reduce the memory and computation cost, we propose FPQVAR, an efficient post-training floating-point (FP) quantization framework for VAR featuring algorithm and hardware co-design. At the algorithm level, we first identify the challenges of quantizing VAR. To address them, we propose Dual Format Quantization for the highly imbalanced input activation. We further propose Group-wise Hadamard Transformation and GHT-Aware Learnable Transformation to address the time-varying outlier channels. At the hardware level, we design the first low-bit FP quantizer and multiplier with lookup tables on FPGA and propose the first FPGA-based VAR accelerator featuring low-bit FP computation and an elaborate two-level pipeline. Extensive experiments show that compared to the state-of-the-art quantization method, our proposed FPQVAR significantly improves Fréchet Inception Distance (FID) from 10.83 to 3.58, Inception Score (IS) from 175.9 to 241.5 under 4-bit quantization. FPQVAR also significantly improves the performance of 6-bit quantized VAR, bringing it on par with the FP16 model. Our accelerator on AMD-Xilinx VCK190 FPGA achieves a throughput of 1.1 image/s, which is 3.1x higher than the integer-based accelerator. It also demonstrates 3.6x and 2.8x higher energy efficiency compared to the integer-based accelerator and GPU baseline, respectively.

cs.CV↗

arXiv2025

A Global Commuting Origin-Destination Flow Dataset for Urban Sustainable Development

Commuting Origin-Destination (OD) flows capture movements of people from residences to workplaces, representing the predominant form of intra-city mobility and serving as a critical reference for understanding urban dynamics and supporting sustainable policies. However, acquiring such data requires costly, time-consuming censuses. In this study, we introduce a commuting OD flow dataset for cities around the world, spanning 6 continents, 179 countries, and 1,625 cities, providing unprecedented coverage of dynamics under diverse urban environments. Specifically, we collected fine-grained demographic data, satellite imagery, and points of interest~(POIs) for each city as foundational inputs to characterize the functional roles of urban regions. Leveraging these, a deep generative model is employed to capture the complex relationships between urban geospatial features and human mobility, enabling the generation of commuting OD flows between urban regions. Comprehensively, validation shows that the spatial distributions of the generated flows closely align with real-world observations. We believe this dataset offers a valuable resource for advancing sustainable urban development research in urban science, data science, transportation engineering, and related fields.

cs.OH↗

arXiv2025

Degradation-Aware Feature Perturbation for All-in-One Image Restoration

All-in-one image restoration aims to recover clear images from various degradation types and levels with a unified model. Nonetheless, the significant variations among degradation types present challenges for training a universal model, often resulting in task interference, where the gradient update directions of different tasks may diverge due to shared parameters. To address this issue, motivated by the routing strategy, we propose DFPIR, a novel all-in-one image restorer that introduces Degradation-aware Feature Perturbations(DFP) to adjust the feature space to align with the unified parameter space. In this paper, the feature perturbations primarily include channel-wise perturbations and attention-wise perturbations. Specifically, channel-wise perturbations are implemented by shuffling the channels in high-dimensional space guided by degradation types, while attention-wise perturbations are achieved through selective masking in the attention space. To achieve these goals, we propose a Degradation-Guided Perturbation Block (DGPB) to implement these two functions, positioned between the encoding and decoding stages of the encoder-decoder architecture. Extensive experimental results demonstrate that DFPIR achieves state-of-the-art performance on several all-in-one image restoration tasks including image denoising, image dehazing, image deraining, motion deblurring, and low-light image enhancement. Our codes are available at https://github.com/TxpHome/DFPIR.

cs.CV↗

arXiv2025

Projection method for quasiperiodic elliptic equations and application to quasiperiodic homogenization

In this study, we address the challenge of solving elliptic equations with quasiperiodic coefficients. To achieve accurate and efficient computation, we introduce the projection method, which enables the embedding of quasiperiodic systems into higher-dimensional periodic systems. To enhance the computational efficiency, we propose a compressed storage strategy for the stiffness matrix by its multi-level block circulant structure, significantly reducing memory requirements. Furthermore, we design a diagonal preconditioner to efficiently solve the resulting high-dimensional linear system by reducing the condition number of the stiffness matrix. These techniques collectively contribute to the computational effectiveness of our proposed approach. Convergence analysis shows the polynomial accuracy of the proposed method. We demonstrate the effectiveness and accuracy of our approach through a series of numerical examples. Moreover, we apply our method to achieve a highly accurate computation of the homogenized coefficients for a quasiperiodic multiscale elliptic equation.

math.NA↗

arXiv2025

Error analysis for temporal second-order finite element approximations of axisymmetric mean curvature flow of genus-1 surfaces

Existing studies on the convergence of numerical methods for curvature flows primarily focus on first-order temporal schemes. In this paper, we establish a novel error analysis for parametric finite element approximations of genus-1 axisymmetric mean curvature flow, formulated using two classical second-order time-stepping methods: the Crank-Nicolson method and the BDF2 method. Our results establish optimal error bounds in both the L^2-norm and H^1-norm, along with a superconvergence result in the H^1-norm for each fully discrete approximation. Finally, we perform convergence experiments to validate the theoretical findings and present numerical simulations for various genus-1 surfaces. Through a series of comparative experiments, we also demonstrate that the methods proposed in this paper exhibit significant mesh advantages.

math.NA↗

arXiv2025

Holistic Capability Preservation: Towards Compact Yet Comprehensive Reasoning Models

This technical report presents Ring-Lite-Distill, a lightweight reasoning model derived from our open-source Mixture-of-Experts (MoE) Large Language Models (LLMs) Ling-Lite. This study demonstrates that through meticulous high-quality data curation and ingenious training paradigms, the compact MoE model Ling-Lite can be further trained to achieve exceptional reasoning capabilities, while maintaining its parameter-efficient architecture with only 2.75 billion activated parameters, establishing an efficient lightweight reasoning architecture. In particular, in constructing this model, we have not merely focused on enhancing advanced reasoning capabilities, exemplified by high-difficulty mathematical problem solving, but rather aimed to develop a reasoning model with more comprehensive competency coverage. Our approach ensures coverage across reasoning tasks of varying difficulty levels while preserving generic capabilities, such as instruction following, tool use, and knowledge retention. We show that, Ring-Lite-Distill's reasoning ability reaches a level comparable to DeepSeek-R1-Distill-Qwen-7B, while its general capabilities significantly surpass those of DeepSeek-R1-Distill-Qwen-7B. The models are accessible at https://huggingface.co/inclusionAI

cs.LG↗

arXiv2025

UniCAIM: A Unified CAM/CIM Architecture with Static-Dynamic KV Cache Pruning for Efficient Long-Context LLM Inference

Transformer-based large language models (LLMs) have achieved impressive performance in various natural language processing (NLP) applications. However, the high memory and computation cost induced by the KV cache limits the inference efficiency, especially for long input sequences. Compute-in-memory (CIM)-based accelerators have been proposed for LLM acceleration with KV cache pruning. However, as existing accelerators only support static pruning with a fixed pattern or dynamic pruning with primitive implementations, they suffer from either high accuracy degradation or low efficiency. In this paper, we propose a ferroelectric FET (FeFET)-based unified content addressable memory (CAM) and CIM architecture, dubbed as UniCAIM. UniCAIM features simultaneous support for static and dynamic pruning with 3 computation modes: 1) in the CAM mode, UniCAIM enables approximate similarity measurement in O(1) time for dynamic KV cache pruning with high energy efficiency; 2) in the charge-domain CIM mode, static pruning can be supported based on accumulative similarity score, which is much more flexible compared to fixed patterns; 3) in the current-domain mode, exact attention computation can be conducted with a subset of selected KV cache. We further propose a novel CAM/CIM cell design that leverages the multi-level characteristics of FeFETs for signed multibit storage of the KV cache and in-place attention computation. With extensive experimental results, we demonstrate UniCAIM can reduce the area-energy-delay product (AEDP) by 8.2-831x over the state-ofthe-art CIM-based LLM accelerators at the circuit level, along with high accuracy comparable with dense attention at the application level, showing its great potential for efficient long-context LLM inference.

cs.AR↗

arXiv2025

HybriMoE: Hybrid CPU-GPU Scheduling and Cache Management for Efficient MoE Inference

The Mixture of Experts (MoE) architecture has demonstrated significant advantages as it enables to increase the model capacity without a proportional increase in computation. However, the large MoE model size still introduces substantial memory demands, which usually requires expert offloading on resource-constrained platforms and incurs significant overhead. Hybrid CPU-GPU inference has been proposed to leverage CPU computation to reduce expert loading overhead but faces major challenges: on one hand, the expert activation patterns of MoE models are highly unstable, rendering the fixed mapping strategies in existing works inefficient; on the other hand, the hybrid CPU-GPU schedule for MoE is inherently complex due to the diverse expert sizes, structures, uneven workload distribution, etc. To address these challenges, in this paper, we propose HybriMoE, a hybrid CPU-GPU inference framework that improves resource utilization through a novel CPU-GPU scheduling and cache management system. HybriMoE introduces (i) a dynamic intra-layer scheduling strategy to balance workloads across CPU and GPU, (ii) an impact-driven inter-layer prefetching algorithm, and (iii) a score-based caching algorithm to mitigate expert activation instability. We implement HybriMoE on top of the kTransformers framework and evaluate it on three widely used MoE-based LLMs. Experimental results demonstrate that HybriMoE achieves an average speedup of 1.33$\times$ in the prefill stage and 1.70$\times$ in the decode stage compared to state-of-the-art hybrid MoE inference framework. Our code is available at: https://github.com/PKU-SEC-Lab/HybriMoE.

cs.LG↗

arXiv2025

ReaLM: Reliable and Efficient Large Language Model Inference with Statistical Algorithm-Based Fault Tolerance

The demand for efficient large language model (LLM) inference has propelled the development of dedicated accelerators. As accelerators are vulnerable to hardware faults due to aging, variation, etc, existing accelerator designs often reserve a large voltage margin or leverage algorithm-based fault tolerance (ABFT) techniques to ensure LLM inference correctness. However, previous methods often overlook the inherent fault tolerance of LLMs, leading to high computation and energy overhead. To enable reliable yet efficient LLM inference, in this paper, we propose a novel algorithm/circuit co-design framework, dubbed ReaLM. For the first time, we systematically characterize the fault tolerance of LLMs by performing a large-scale error injection study of representative LLMs and natural language understanding tasks. Then, we propose a statistical ABFT algorithm that fully leverages the error robustness to minimize error recovery as much as possible. We also customize the error detection circuits to enable a low-cost online collection of error statistics. Extensive experiments show that with only 1.42% circuit area and 1.79% power overhead, our ReaLM can reduce perplexity degradation from 18.54 to 0.29. Compared to existing methods, ReaLM consistently reduces recovery costs across different operating voltages and improves energy efficiency by up to 35.83% without compromising LLM performance. Our error injection code is available at https://github.com/PKU-SEC-Lab/ReaLM_DAC25/

cs.AR↗

arXiv2025

Bayesian Covariate-Dependent Graph Learning with a Dual Group Spike-and-Slab Prior

Covariate-dependent graph learning has gained increasing interest in the graphical modeling literature for the analysis of heterogeneous data. This task, however, poses challenges to modeling, computational efficiency, and interpretability. The parameter of interest can be naturally represented as a three-dimensional array with elements that can be grouped according to two directions, corresponding to node level and covariate level, respectively. In this article, we propose a novel dual group spike-and-slab prior that enables multi-level selection at covariate-level and node-level, as well as individual (local) level sparsity. We introduce a nested strategy with specific choices to address distinct challenges posed by the various grouping directions. For posterior inference, we develop a tuning-free Gibbs sampler for all parameters, which mitigates the difficulties of parameter tuning often encountered in high-dimensional graphical models and facilitates routine implementation. Through simulation studies, we demonstrate that the proposed model outperforms existing methods in its accuracy of graph recovery. We show the practical utility of our model via an application to microbiome data where we seek to better understand the interactions among microbes as well as how these are affected by relevant covariates.

stat.ME↗

arXiv2025

Empower Vision Applications with LoRA LMM

Large Multimodal Models (LMMs) have shown significant progress in various complex vision tasks with the solid linguistic and reasoning capacity inherited from large language models (LMMs). Low-rank adaptation (LoRA) offers a promising method to integrate external knowledge into LMMs, compensating for their limitations on domain-specific tasks. However, the existing LoRA model serving is excessively computationally expensive and causes extremely high latency. In this paper, we present an end-to-end solution that empowers diverse vision tasks and enriches vision applications with LoRA LMMs. Our system, VaLoRA, enables accurate and efficient vision tasks by 1) an accuracy-aware LoRA adapter generation approach that generates LoRA adapters rich in domain-specific knowledge to meet application-specific accuracy requirements, 2) an adaptive-tiling LoRA adapters batching operator that efficiently computes concurrent heterogeneous LoRA adapters, and 3) a flexible LoRA adapter orchestration mechanism that manages application requests and LoRA adapters to achieve the lowest average response latency. We prototype VaLoRA on five popular vision tasks on three LMMs. Experiment results reveal that VaLoRA improves 24-62% of the accuracy compared to the original LMMs and reduces 20-89% of the latency compared to the state-of-the-art LoRA model serving systems.

cs.CV↗

arXiv2025

A sharp-interface approach for simulating solid-state dewetting of thin films with double-bubble structure

We develop a sharp-interface model for solid-state dewetting of double-bubble thin films using an energy variational approach based on a newly proposed interfacial energy. This model characterizes the dynamic evolution of interfaces in double-bubble thin films, a process primarily governed by surface diffusion and junction/contact points migration, and fundamentally distinct from the behavior observed in a single thin film. Subsequently, a structure-preserving parametric finite element approximation is developed for the sharp-interface model, which can preserve both area conservation and energy stability. Extensive numerical experiments are presented to demonstrate the convergence, structure-preserving properties, and superior mesh quality of the proposed method. Additionally, we investigate several specific evolution processes, including the equilibrium shapes of double-bubble thin films and the pinch-off dynamics of long islands.

math.NA↗

arXiv2025

Zero-Knowledge Proof-based Verifiable Decentralized Machine Learning in Communication Network: A Comprehensive Survey

Over recent decades, machine learning has significantly advanced network communication, enabling improved decision-making, user behavior analysis, and fault detection. Decentralized approaches, where participants exchange computation results instead of raw private data, mitigate these risks but introduce challenges related to trust and verifiability. A critical issue arises: How can one ensure the integrity and validity of computation results shared by other participants? Existing survey articles predominantly address security and privacy concerns in decentralized machine learning, whereas this survey uniquely highlights the emerging issue of verifiability. Recognizing the critical role of zero-knowledge proofs in ensuring verifiability, we present a comprehensive review of Zero-Knowledge Proof-based Verifiable Machine Learning (ZKP-VML). To clarify the research problem, we present a definition of ZKP-VML consisting of four algorithms, along with several corresponding key security properties. Besides, we provide an overview of the current research landscape by systematically organizing the research timeline and categorizing existing schemes based on their security properties. Furthermore, through an in-depth analysis of each existing scheme, we summarize their technical contributions and optimization strategies, aiming to uncover common design principles underlying ZKP-VML schemes. Building on the reviews and analysis presented, we identify current research challenges and suggest future research directions. To the best of our knowledge, this is the most comprehensive survey to date on verifiable decentralized machine learning and ZKP-VML.

cs.LG↗

arXiv2025

SCALES: Boost Binary Neural Network for Image Super-Resolution with Efficient Scalings

Deep neural networks for image super-resolution (SR) have demonstrated superior performance. However, the large memory and computation consumption hinders their deployment on resource-constrained devices. Binary neural networks (BNNs), which quantize the floating point weights and activations to 1-bit can significantly reduce the cost. Although BNNs for image classification have made great progress these days, existing BNNs for SR still suffer from a large performance gap between the FP SR networks. To this end, we observe the activation distribution in SR networks and find much larger pixel-to-pixel, channel-to-channel, layer-to-layer, and image-to-image variation in the activation distribution than image classification networks. However, existing BNNs for SR fail to capture these variations that contain rich information for image reconstruction, leading to inferior performance. To address this problem, we propose SCALES, a binarization method for SR networks that consists of the layer-wise scaling factor, the spatial re-scaling method, and the channel-wise re-scaling method, capturing the layer-wise, pixel-wise, and channel-wise variations efficiently in an input-dependent manner. We evaluate our method across different network architectures and datasets. For CNN-based SR networks, our binarization method SCALES outperforms the prior art method by 0.2dB with fewer parameters and operations. With SCALES, we achieve the first accurate binary Transformer-based SR network, improving PSNR by more than 1dB compared to the baseline method.

cs.CV↗

arXiv2025

DeepSeek-V3 Technical Report

We present DeepSeek-V3, a strong Mixture-of-Experts (MoE) language model with 671B total parameters with 37B activated for each token. To achieve efficient inference and cost-effective training, DeepSeek-V3 adopts Multi-head Latent Attention (MLA) and DeepSeekMoE architectures, which were thoroughly validated in DeepSeek-V2. Furthermore, DeepSeek-V3 pioneers an auxiliary-loss-free strategy for load balancing and sets a multi-token prediction training objective for stronger performance. We pre-train DeepSeek-V3 on 14.8 trillion diverse and high-quality tokens, followed by Supervised Fine-Tuning and Reinforcement Learning stages to fully harness its capabilities. Comprehensive evaluations reveal that DeepSeek-V3 outperforms other open-source models and achieves performance comparable to leading closed-source models. Despite its excellent performance, DeepSeek-V3 requires only 2.788M H800 GPU hours for its full training. In addition, its training process is remarkably stable. Throughout the entire training process, we did not experience any irrecoverable loss spikes or perform any rollbacks. The model checkpoints are available at https://github.com/deepseek-ai/DeepSeek-V3.

cs.CL↗

arXiv2025

Classification of Major Depressive Disorder Using Vertex-Wise Brain Sulcal Depth, Curvature, and Thickness with a Deep and a Shallow Learning Model

Major depressive disorder (MDD) is a complex psychiatric disorder that affects the lives of hundreds of millions of individuals around the globe. Even today, researchers debate if morphological alterations in the brain are linked to MDD, likely due to the heterogeneity of this disorder. The application of deep learning tools to neuroimaging data, capable of capturing complex non-linear patterns, has the potential to provide diagnostic and predictive biomarkers for MDD. However, previous attempts to demarcate MDD patients and healthy controls (HC) based on segmented cortical features via linear machine learning approaches have reported low accuracies. Here, we used globally representative data from the ENIGMA-MDD working group containing 7,012 participants from 30 sites (N=2,772 MDD and N=4,240 HC), which allows a comprehensive analysis with generalizable results. Based on the hypothesis that integration of vertex-wise cortical features can improve classification performance, we evaluated the classification of a DenseNet and a Support Vector Machine (SVM), with the expectation that the former would outperform the latter. We found that both classifiers exhibited close to chance performance (balanced accuracy DenseNet: 51%; SVM: 53%), when estimated on unseen sites. Slightly higher classification performance (balanced accuracy DenseNet: 58%; SVM: 55%) was found when the cross-validation folds contained subjects from all sites, indicating site effect. In conclusion, the integration of vertex-wise morphometric features and the use of the non-linear classifier did not lead to the differentiability between MDD and HC. Our results support the notion that MDD classification on this combination of such features and classifiers is unfeasible. Perhaps more sophisticated integration of multimodal information may lead to a higher performance in this diagnostic task.

q-bio.QM↗

arXiv2025

Structure-preserving parametric finite element methods for anisotropic surface diffusion flow with minimal deformation formulation

High mesh quality plays a crucial role in maintaining the stability of solutions in geometric flow problems. Duan and Li [Duan & Li, SIAM J. Sci. Comput. 46 (1) (2024) A587-A608] applied the minimal deformation (MD) formulation to propose an artificial tangential velocity determined by harmonic mapping to improve mesh quality. In this work, we extend the method to anisotropic surface diffusion flows, which, similar to isotropic curvature flow, also preserves excellent mesh quality. Furthermore, developing a numerical algorithm for the flow with MD formulation that guarantees volume conservation and energy stability remains a challenging task. We, in this paper, successfully construct several structure-preserving algorithms, including first-order and high-order temporal discretization methods. Extensive numerical experiments show that our methods effectively preserve mesh quality for anisotropic SDFs, ensuring high-order temporal accuracy, volume conservation or/and energy stability.

math.NA↗