arXiv subjects

Wei Lu

Publications and source records attributed to Wei Lu.

At least 181 records · Page 10Linked to original sources

arXiv2024

SailCompass: Towards Reproducible and Robust Evaluation for Southeast Asian Languages

In this paper, we introduce SailCompass, a reproducible and robust evaluation benchmark for assessing Large Language Models (LLMs) on Southeast Asian Languages (SEA). SailCompass encompasses three main SEA languages, eight primary tasks including 14 datasets covering three task types (generation, multiple-choice questions, and classification). To improve the robustness of the evaluation approach, we explore different prompt configurations for multiple-choice questions and leverage calibrations to improve the faithfulness of classification tasks. With SailCompass, we derive the following findings: (1) SEA-specialized LLMs still outperform general LLMs, although the gap has narrowed; (2) A balanced language distribution is important for developing better SEA-specialized LLMs; (3) Advanced prompting techniques (e.g., calibration, perplexity-based ranking) are necessary to better utilize LLMs. All datasets and evaluation scripts are public.

cs.CL↗

arXiv2024

The dual codes of two families of BCH codes

In this paper, we present an infinite family of MDS codes over $\mathbb{F}_{2^s}$ and two infinite families of almost MDS codes over $\mathbb{F}_{p^s}$ for any prime $p$, by investigating the parameters of the dual codes of two families of BCH codes. Notably, these almost MDS codes include two infinite families of near MDS codes over $\mathbb{F}_{3^s}$, resolving a conjecture posed by Geng et al. in 2022. Furthermore, we demonstrate that both of these almost AMDS codes and their dual codes hold infinite families of $3$-designs over $\mathbb{F}_{p^s}$ for any prime $p$. Additionally, we study the subfield subcodes of these families of MDS and near MDS codes, and provide several binary, ternary, and quaternary codes with best known parameters.

cs.IT↗

arXiv2024

Fine-tuning large language models for domain adaptation: Exploration of training strategies, scaling, model merging and synergistic capabilities

The advancement of Large Language Models (LLMs) for domain applications in fields such as materials science and engineering depends on the development of fine-tuning strategies that adapt models for specialized, technical capabilities. In this work, we explore the effects of Continued Pretraining (CPT), Supervised Fine-Tuning (SFT), and various preference-based optimization approaches, including Direct Preference Optimization (DPO) and Odds Ratio Preference Optimization (ORPO), on fine-tuned LLM performance. Our analysis shows how these strategies influence model outcomes and reveals that the merging of multiple fine-tuned models can lead to the emergence of capabilities that surpass the individual contributions of the parent models. We find that model merging leads to new functionalities that neither parent model could achieve alone, leading to improved performance in domain-specific assessments. Experiments with different model architectures are presented, including Llama 3.1 8B and Mistral 7B models, where similar behaviors are observed. Exploring whether the results hold also for much smaller models, we use a tiny LLM with 1.7 billion parameters and show that very small LLMs do not necessarily feature emergent capabilities under model merging, suggesting that model scaling may be a key component. In open-ended yet consistent chat conversations between a human and AI models, our assessment reveals detailed insights into how different model variants perform and show that the smallest model achieves a high intelligence score across key criteria including reasoning depth, creativity, clarity, and quantitative precision. Other experiments include the development of image generation prompts based on disparate biological material design concepts, to create new microstructures, architectural concepts, and urban design based on biological materials-inspired construction principles.

cs.CL↗

arXiv2024

Predicting Scientific Impact Through Diffusion, Conformity, and Contribution Disentanglement

The scientific impact of academic papers is influenced by intricate factors such as dynamic popularity and inherent contribution. Existing models typically rely on static graphs for citation count estimation, failing to differentiate among its sources. In contrast, we propose distinguishing effects derived from various factors and predicting citation increments as estimated potential impacts within the dynamic context. In this research, we introduce a novel model, DPPDCC, which Disentangles the Potential impacts of Papers into Diffusion, Conformity, and Contribution values. It encodes temporal and structural features within dynamic heterogeneous graphs derived from the citation networks and applies various auxiliary tasks for disentanglement. By emphasizing comparative and co-cited/citing information and aggregating snapshots evolutionarily, DPPDCC captures knowledge flow within the citation network. Afterwards, popularity is outlined by contrasting augmented graphs to extract the essence of citation diffusion and predicting citation accumulation bins for quantitative conformity modeling. Orthogonal constraints ensure distinct modeling of each perspective, preserving the contribution value. To gauge generalization across publication times and replicate the realistic dynamic context, we partition data based on specific time points and retain all samples without strict filtering. Extensive experiments on three datasets validate DPPDCC's superiority over baselines for papers published previously, freshly, and immediately, with further analyses confirming its robustness. Our codes and supplementary materials can be found at https://github.com/ECNU-Text-Computing/DPPDCC.

cs.SI↗

arXiv2024

Adaptive Two-Stage Cloud Resource Scaling via Hierarchical Multi-Indicator Forecasting and Bayesian Decision-Making

The surging demand for cloud computing resources, driven by the rapid growth of sophisticated large-scale models and data centers, underscores the critical importance of efficient and adaptive resource allocation. As major tech enterprises deploy massive infrastructures with thousands of GPUs, existing cloud platforms still struggle with low resource utilization due to key challenges: capturing hierarchical indicator structures, modeling non-Gaussian distributions, and decision-making under uncertainty. To address these challenges, we propose HRAMONY, an adaptive Hierarchical Attention-based Resource Modeling and Decision-Making System. HARMONY combines hierarchical multi-indicator distribution forecasting and uncertainty-aware Bayesian decision-making. It introduces a novel hierarchical attention mechanism that comprehensively models complex inter-indicator dependencies, enabling accurate predictions that can adapt to evolving environment states. By transforming Gaussian projections into adaptive non-Gaussian distributions via Normalizing Flows. Crucially, HARMONY leverages the full predictive distributions in an adaptive Bayesian process, proactively incorporating uncertainties to optimize resource allocation while robustly meeting SLA constraints under varying conditions. Extensive evaluations across four large-scale cloud datasets demonstrate HARMONY's state-of-the-art performance, significantly outperforming nine established methods. A month-long real-world deployment validated HARMONY's substantial practical impact, realizing over 35,000 GPU hours in savings and translating to $100K+ in cost reduction, showcasing its remarkable economic value through adaptive, uncertainty-aware scaling. Our code is available at https://github.com/Floating-LY/HARMONY1.

cs.LG↗

arXiv2024

Every Part Matters: Integrity Verification of Scientific Figures Based on Multimodal Large Language Models

This paper tackles a key issue in the interpretation of scientific figures: the fine-grained alignment of text and figures. It advances beyond prior research that primarily dealt with straightforward, data-driven visualizations such as bar and pie charts and only offered a basic understanding of diagrams through captioning and classification. We introduce a novel task, Figure Integrity Verification, designed to evaluate the precision of technologies in aligning textual knowledge with visual elements in scientific figures. To support this, we develop a semi-automated method for constructing a large-scale dataset, Figure-seg, specifically designed for this task. Additionally, we propose an innovative framework, Every Part Matters (EPM), which leverages Multimodal Large Language Models (MLLMs) to not only incrementally improve the alignment and verification of text-figure integrity but also enhance integrity through analogical reasoning. Our comprehensive experiments show that these innovations substantially improve upon existing methods, allowing for more precise and thorough analysis of complex scientific figures. This progress not only enhances our understanding of multimodal technologies but also stimulates further research and practical applications across fields requiring the accurate interpretation of complex visual data.

cs.CL↗

arXiv2024

Self-Training with Direct Preference Optimization Improves Chain-of-Thought Reasoning

Effective training of language models (LMs) for mathematical reasoning tasks demands high-quality supervised fine-tuning data. Besides obtaining annotations from human experts, a common alternative is sampling from larger and more powerful LMs. However, this knowledge distillation approach can be costly and unstable, particularly when relying on closed-source, proprietary LMs like GPT-4, whose behaviors are often unpredictable. In this work, we demonstrate that the reasoning abilities of small-scale LMs can be enhanced through self-training, a process where models learn from their own outputs. We also show that the conventional self-training can be further augmented by a preference learning algorithm called Direct Preference Optimization (DPO). By integrating DPO into self-training, we leverage preference data to guide LMs towards more accurate and diverse chain-of-thought reasoning. We evaluate our method across various mathematical reasoning tasks using different base models. Our experiments show that this approach not only improves LMs' reasoning performance but also offers a more cost-effective and scalable solution compared to relying on large proprietary LMs.

cs.CL↗

arXiv2024

Coarse-to-Fine Proposal Refinement Framework for Audio Temporal Forgery Detection and Localization

Recently, a novel form of audio partial forgery has posed challenges to its forensics, requiring advanced countermeasures to detect subtle forgery manipulations within long-duration audio. However, existing countermeasures still serve a classification purpose and fail to perform meaningful analysis of the start and end timestamps of partial forgery segments. To address this challenge, we introduce a novel coarse-to-fine proposal refinement framework (CFPRF) that incorporates a frame-level detection network (FDN) and a proposal refinement network (PRN) for audio temporal forgery detection and localization. Specifically, the FDN aims to mine informative inconsistency cues between real and fake frames to obtain discriminative features that are beneficial for roughly indicating forgery regions. The PRN is responsible for predicting confidence scores and regression offsets to refine the coarse-grained proposals derived from the FDN. To learn robust discriminative features, we devise a difference-aware feature learning (DAFL) module guided by contrastive representation learning to enlarge the sensitive differences between different frames induced by minor manipulations. We further design a boundary-aware feature enhancement (BAFE) module to capture the contextual information of multiple transition boundaries and guide the interaction between boundary information and temporal features via a cross-attention mechanism. Extensive experiments show that our CFPRF achieves state-of-the-art performance on various datasets, including LAV-DF, ASVS2019PS, and HAD.

cs.MM↗

arXiv2024

Black-Box Opinion Manipulation Attacks to Retrieval-Augmented Generation of Large Language Models

Retrieval-Augmented Generation (RAG) is applied to solve hallucination problems and real-time constraints of large language models, but it also induces vulnerabilities against retrieval corruption attacks. Existing research mainly explores the unreliability of RAG in white-box and closed-domain QA tasks. In this paper, we aim to reveal the vulnerabilities of Retrieval-Enhanced Generative (RAG) models when faced with black-box attacks for opinion manipulation. We explore the impact of such attacks on user cognition and decision-making, providing new insight to enhance the reliability and security of RAG models. We manipulate the ranking results of the retrieval model in RAG with instruction and use these results as data to train a surrogate model. By employing adversarial retrieval attack methods to the surrogate model, black-box transfer attacks on RAG are further realized. Experiments conducted on opinion datasets across multiple topics show that the proposed attack strategy can significantly alter the opinion polarity of the content generated by RAG. This demonstrates the model's vulnerability and, more importantly, reveals the potential negative impact on user cognition and decision-making, making it easier to mislead users into accepting incorrect or biased information.

cs.CL↗

arXiv2024

Low-Resource Multi-Granularity Academic Function Recognition Based on Multiple Prompt Knowledge

Fine-tuning pre-trained language models (PLMs), e.g., SciBERT, generally requires large numbers of annotated data to achieve state-of-the-art performance on a range of NLP tasks in the scientific domain. However, obtaining the fine-tune data for scientific NLP task is still challenging and expensive. Inspired by recent advancement in prompt learning, in this paper, we propose the Mix Prompt Tuning (MPT), which is a semi-supervised method to alleviate the dependence on annotated data and improve the performance of multi-granularity academic function recognition tasks with a small number of labeled examples. Specifically, the proposed method provides multi-perspective representations by combining manual prompt templates with automatically learned continuous prompt templates to help the given academic function recognition task take full advantage of knowledge in PLMs. Based on these prompt templates and the fine-tuned PLM, a large number of pseudo labels are assigned to the unlabeled examples. Finally, we fine-tune the PLM using the pseudo training set. We evaluate our method on three academic function recognition tasks of different granularity including the citation function, the abstract sentence function, and the keyword function, with datasets from computer science domain and biomedical domain. Extensive experiments demonstrate the effectiveness of our method and statistically significant improvements against strong baselines. In particular, it achieves an average increase of 5% in Macro-F1 score compared with fine-tuning, and 6% in Macro-F1 score compared with other semi-supervised method under low-resource settings. In addition, MPT is a general method that can be easily applied to other low-resource scientific classification tasks.

cs.CL↗

arXiv2024

Let's Learn Step by Step: Enhancing In-Context Learning Ability with Curriculum Learning

Demonstration ordering, which is an important strategy for in-context learning (ICL), can significantly affects the performance of large language models (LLMs). However, most of the current approaches of ordering require high computational costs to introduce the priori knowledge. In this paper, inspired by the human learning process, we propose a simple but effective demonstration ordering method for ICL, named the few-shot In-Context Curriculum Learning (ICCL). The ICCL implies gradually increasing the complexity of prompt demonstrations during the inference process. The difficulty can be assessed by human experts or LLMs-driven metrics, such as perplexity. Then we design extensive experiments to discuss the effectiveness of the ICCL at both corpus-level and instance-level. Moreover, we also investigate the formation mechanism of LLM's ICCL capability. Experimental results demonstrate that ICCL, developed during the instruction-tuning stage, is effective for representative open-source LLMs. To facilitate further research and applications by other scholars, we make the code publicly available.

cs.CL↗

arXiv2024

Ultra-low noise laser and optical frequency comb-based timing system for the Black Hole Explorer (BHEX) mission

In this effort, we demonstrate the performance of a highly stable time reference for the proposed Black Hole Explorer (BHEX) mission, a space-based extension to the Event Horizon Telescope (EHT) Very Long Baseline Interferometry (VLBI) project. This precision timing system is based on the use of a space-qualified, ultra-low noise laser developed as part of the Laser Interferometer Space Antenna (LISA) mission as the timing reference, and an optical frequency comb to transfer the stability of this laser to the microwave regime for instrumentation use. We describe the implementation of this system and experimental setup to characterize the stability performance. We present the results of this experiment that demonstrate the performance of this system meets requirements for the BHEX mission.

astro-ph.IM↗

arXiv2024

TinyLlama: An Open-Source Small Language Model

We present TinyLlama, a compact 1.1B language model pretrained on around 1 trillion tokens for approximately 3 epochs. Building on the architecture and tokenizer of Llama 2, TinyLlama leverages various advances contributed by the open-source community (e.g., FlashAttention and Lit-GPT), achieving better computational efficiency. Despite its relatively small size, TinyLlama demonstrates remarkable performance in a series of downstream tasks. It significantly outperforms existing open-source language models with comparable sizes. Our model checkpoints and code are publicly available on GitHub at https://github.com/jzhang38/TinyLlama.

cs.CL↗

arXiv2024

The weight hierarchies of three classes of linear codes

Studying the generalized Hamming weights of linear codes is a significant research area within coding theory, as it provides valuable structural information about the codes and plays a crucial role in determining their performance in various applications. However, determining the generalized Hamming weights of linear codes, particularly their weight hierarchy, is generally a challenging task. In this paper, we focus on investigating the generalized Hamming weights of three classes of linear codes over finite fields. These codes are constructed by different defining sets. By analysing the intersections between the definition sets and the duals of all $r$-dimensional subspaces, we get the inequalities on the sizes of these intersections. Then constructing subspaces that reach the upper bounds of these inequalities, we successfully determine the complete weight hierarchies of these codes.

cs.IT↗

arXiv2024

Beam Shaping Based on Axisymmetric Aspheric Mirrors

Flat-top beam, known for its ability to generate a consistently even irradiation area, holds vast utility in many fields of scientific and industrial applications. In this paper, a reflective laser beam shaping method based on two axisymmetric aspheric mirrors (AAMs), a polarizing beam splitter (PBS) and two quarter wave plates (QWPs) is proposed to transform Gaussian beam into flat-top beam. Compared to alternative beam shaping methods, the method using AAMs demonstrates distinct advantages on notably high energy efficiency and unique capability to generate parallel beams. Thanks to its relative simplicities of design, manufacture and tunability, AAMs-shaping further enhances its appeal in applied research scenarios.

physics.optics↗

arXiv2024

G2LTraj: A Global-to-Local Generation Approach for Trajectory Prediction

Predicting future trajectories of traffic agents accurately holds substantial importance in various applications such as autonomous driving. Previous methods commonly infer all future steps of an agent either recursively or simultaneously. However, the recursive strategy suffers from the accumulated error, while the simultaneous strategy overlooks the constraints among future steps, resulting in kinematically infeasible predictions. To address these issues, in this paper, we propose G2LTraj, a plug-and-play global-to-local generation approach for trajectory prediction. Specifically, we generate a series of global key steps that uniformly cover the entire future time range. Subsequently, the local intermediate steps between the adjacent key steps are recursively filled in. In this way, we prevent the accumulated error from propagating beyond the adjacent key steps. Moreover, to boost the kinematical feasibility, we not only introduce the spatial constraints among key steps but also strengthen the temporal constraints among the intermediate steps. Finally, to ensure the optimal granularity of key steps, we design a selectable granularity strategy that caters to each predicted trajectory. Our G2LTraj significantly improves the performance of seven existing trajectory predictors across the ETH, UCY and nuScenes datasets. Experimental results demonstrate its effectiveness. Code will be available at https://github.com/Zhanwei-Z/G2LTraj.

cs.CV↗

arXiv2024

Ultrafast Electron Diffraction with MeV Electron Source from a Laser Wakefield Accelerator

MeV ultrafast electron diffraction (UED) is a widely used technology for ultrafast structural dynamic studies of matters in numerous areas. The development of laser wakefield accelerator (LWFA) envisions great potential of advanced all-optical electron source based on LWFA in UED applications. We experimentally demonstrated that an LWFA-based device with a miniaturized permanent magnet beamline can generate and manipulate electron beams suitable for UED. In the beam transmission, the LWFA electron beams with intrinsic short duration stretch due to energy spread and then are compressed by a following double bend achromat. The optimized double bend achromat can make the beamline isochronous such that the arrival time jitter induced by the shot-to-shot energy fluctuation can be eliminated, and allow the advantage of the natural laser-beam synchronization for LWFAs to emerge. With the energy filtering, the beam energy spread can be reduced to 3% (FWHM) while a sufficient amount of charge (11.9 fC) per bunch for diffraction is retained. Start-to-end simulations showed that the bunch length reaches ~30 fs (rms) with the same experimental configuration. Clear single-shot and multi-shot diffraction patterns of single-crystalline gold samples are obtained and the derived lattice constant agrees excellently with the real value. Our proof-of-principle experiments open the door to the detection of ultrafast structural dynamics using MeV LWFA beams, and pave the way for the UED applications with sub-10-fs temporal resolution.

physics.acc-ph↗

arXiv2024

Enhance Robustness of Language Models Against Variation Attack through Graph Integration

The widespread use of pre-trained language models (PLMs) in natural language processing (NLP) has greatly improved performance outcomes. However, these models' vulnerability to adversarial attacks (e.g., camouflaged hints from drug dealers), particularly in the Chinese language with its rich character diversity/variation and complex structures, hatches vital apprehension. In this study, we propose a novel method, CHinese vAriatioN Graph Enhancement (CHANGE), to increase the robustness of PLMs against character variation attacks in Chinese content. CHANGE presents a novel approach for incorporating a Chinese character variation graph into the PLMs. Through designing different supplementary tasks utilizing the graph structure, CHANGE essentially enhances PLMs' interpretation of adversarially manipulated text. Experiments conducted in a multitude of NLP tasks show that CHANGE outperforms current language models in combating against adversarial attacks and serves as a valuable contribution to robust language model research. These findings contribute to the groundwork on robust language models and highlight the substantial potential of graph-guided pre-training strategies for real-world applications.

cs.CL↗