arXiv subjects

Zhe Yang

Publications and source records attributed to Zhe Yang.

At least 55 records · Page 3Linked to original sources

arXiv2025

LLaVA-Mini: Efficient Image and Video Large Multimodal Models with One Vision Token

The advent of real-time large multimodal models (LMMs) like GPT-4o has sparked considerable interest in efficient LMMs. LMM frameworks typically encode visual inputs into vision tokens (continuous representations) and integrate them and textual instructions into the context of large language models (LLMs), where large-scale parameters and numerous context tokens (predominantly vision tokens) result in substantial computational overhead. Previous efforts towards efficient LMMs always focus on replacing the LLM backbone with smaller models, while neglecting the crucial issue of token quantity. In this paper, we introduce LLaVA-Mini, an efficient LMM with minimal vision tokens. To achieve a high compression ratio of vision tokens while preserving visual information, we first analyze how LMMs understand vision tokens and find that most vision tokens only play a crucial role in the early layers of LLM backbone, where they mainly fuse visual information into text tokens. Building on this finding, LLaVA-Mini introduces modality pre-fusion to fuse visual information into text tokens in advance, thereby facilitating the extreme compression of vision tokens fed to LLM backbone into one token. LLaVA-Mini is a unified large multimodal model that can support the understanding of images, high-resolution images, and videos in an efficient manner. Experiments across 11 image-based and 7 video-based benchmarks demonstrate that LLaVA-Mini outperforms LLaVA-v1.5 with just 1 vision token instead of 576. Efficiency analyses reveal that LLaVA-Mini can reduce FLOPs by 77%, deliver low-latency responses within 40 milliseconds, and process over 10,000 frames of video on the GPU hardware with 24GB of memory.

cs.CV↗

arXiv2025

CiteCheck: Towards Accurate Citation Faithfulness Detection

Citation faithfulness detection is critical for enhancing retrieval-augmented generation (RAG) systems, yet large-scale Chinese datasets for this task are scarce. Existing methods face prohibitive costs due to the need for manually annotated negative samples. To address this, we introduce the first large-scale Chinese dataset CiteCheck for citation faithfulness detection, constructed via a cost-effective approach using two-stage manual annotation. This method balances positive and negative samples while significantly reducing annotation expenses. CiteCheck comprises training and test splits. Experiments demonstrate that: (1) the test samples are highly challenging, with even state-of-the-art LLMs failing to achieve high accuracy; and (2) training data augmented with LLM-generated negative samples enables smaller models to attain strong performance using parameter-efficient fine-tuning. CiteCheck provides a robust foundation for advancing citation faithfulness detection in Chinese RAG systems. The dataset is publicly available to facilitate research.

cs.CL↗

arXiv2025

Induced Interactions and Bipolarons in Spin-Orbit Coupled Bose-Einstein Condensates

Impurities immersed in a Bose-Einstein condensate (BEC) can interact indirectly through the exchange of Bogoliubov excitations. These impurities, which form dressed quasiparticles known as Bose polarons due to their interaction with the BEC, can pair up to form a bound state called bipolarons, via an induced interaction. Previous studies on induced interactions have primarily focused on cases with an isotropic excitation spectrum. In this work, we investigate the properties of induced interactions and bipolarons mediated by anisotropic Bogoliubov excitations using field theory. Taking a BEC with spin-orbit coupling as an example, we show that the induced interaction becomes anisotropic. Notably, a double-minima feature appears in the induced interaction in momentum space due to the exchange of roton excitations. Additionally, we calculate the binding energy and wave functions of these bipolarons induced by anisotropic interactions. Unlike previously studied bipolarons formed through the exchange of isotropic phonon excitations, we identify a new type of bipolarons whose wave functions feature a double-peak structure under strong impurity-boson interactions. Our work extends the theory of induced interactions from isotropic to anisotropic systems, and reveals the novel features in both the induced interactions and bipolarons arising from BEC with an unconventional excitation spectrum.

cond-mat.quant-gas↗

arXiv2024

Confidence v.s. Critique: A Decomposition of Self-Correction Capability for LLMs

Large Language Models (LLMs) can correct their self-generated responses, but a decline in accuracy after self-correction is also witnessed. To have a deeper understanding of self-correction, we endeavor to decompose, evaluate, and analyze the self-correction behaviors of LLMs. By enumerating and analyzing answer correctness before and after self-correction, we decompose the self-correction capability into confidence (being confident to correct answers) and critique (turning wrong answers to correct) capabilities, and propose two metrics from a probabilistic perspective to measure these 2 capabilities, along with another metric for overall self-correction capability evaluation. Based on our decomposition and evaluation metrics, we conduct extensive experiments and draw some empirical conclusions. For example, we find different models can exhibit distinct behaviors: some models are confident while others are more critical. We also find the trade-off between the two capabilities (i.e. improving one can lead to a decline in the other) when manipulating model self-correction behavior by prompts or in-context learning. Further, we find a simple yet efficient strategy to improve self-correction capability by transforming Supervision Fine-Tuning (SFT) data format, and our strategy outperforms vanilla SFT in both capabilities and achieves much higher accuracy after self-correction. Our code will be publicly available on GitHub.

cs.CL↗

arXiv2024

Omni-MATH: A Universal Olympiad Level Mathematic Benchmark For Large Language Models

Recent advancements in large language models (LLMs) have led to significant breakthroughs in mathematical reasoning capabilities. However, existing benchmarks like GSM8K or MATH are now being solved with high accuracy (e.g., OpenAI o1 achieves 94.8\% on MATH dataset), indicating their inadequacy for truly challenging these models. To bridge this gap, we propose a comprehensive and challenging benchmark specifically designed to assess LLMs' mathematical reasoning at the Olympiad level. Unlike existing Olympiad-related benchmarks, our dataset focuses exclusively on mathematics and comprises a vast collection of 4428 competition-level problems with rigorous human annotation. These problems are meticulously categorized into over 33 sub-domains and span more than 10 distinct difficulty levels, enabling a holistic assessment of model performance in Olympiad-mathematical reasoning. Furthermore, we conducted an in-depth analysis based on this benchmark. Our experimental results show that even the most advanced models, OpenAI o1-mini and OpenAI o1-preview, struggle with highly challenging Olympiad-level problems, with 60.54\% and 52.55\% accuracy, highlighting significant challenges in Olympiad-level mathematical reasoning.

cs.CL↗

arXiv2024

SPGL: Enhancing Session-based Recommendation with Single Positive Graph Learning

Session-based recommendation seeks to forecast the next item a user will be interested in, based on their interaction sequences. Due to limited interaction data, session-based recommendation faces the challenge of limited data availability. Traditional methods enhance feature learning by constructing complex models to generate positive and negative samples. This paper proposes a session-based recommendation model using Single Positive optimization loss and Graph Learning (SPGL) to deal with the problem of data sparsity, high model complexity and weak transferability. SPGL utilizes graph convolutional networks to generate global item representations and batch session representations, effectively capturing intrinsic relationships between items. The use of single positive optimization loss improves uniformity of item representations, thereby enhancing recommendation accuracy. In the intent extractor, SPGL considers the hop count of the adjacency matrix when constructing the directed global graph to fully integrate spatial information. It also takes into account the reverse positional information of items when constructing session representations to incorporate temporal information. Comparative experiments across three benchmark datasets, Tmall, RetailRocket and Diginetica, demonstrate the model's effectiveness. The source code can be accessed on https://github.com/liang-tian-tian/SPGL .

cs.IR↗

arXiv2024

Multi-Graph Co-Training for Capturing User Intent in Session-based Recommendation

Session-based recommendation focuses on predicting the next item a user will interact with based on sequences of anonymous user sessions. A significant challenge in this field is data sparsity due to the typically short-term interactions. Most existing methods rely heavily on users' current interactions, overlooking the wealth of auxiliary information available. To address this, we propose a novel model, the Multi-Graph Co-Training model (MGCOT), which leverages not only the current session graph but also similar session graphs and a global item relation graph. This approach allows for a more comprehensive exploration of intrinsic relationships and better captures user intent from multiple views, enabling session representations to complement each other. Additionally, MGCOT employs multi-head attention mechanisms to effectively capture relevant session intent and uses contrastive learning to form accurate and robust session representations. Extensive experiments on three datasets demonstrate that MGCOT significantly enhances the performance of session-based recommendations, particularly on the Diginetica dataset, achieving improvements up to 2.00% in P@20 and 10.70% in MRR@20. Resources have been made publicly available in our GitHub repository https://github.com/liang-tian-tian/MGCOT.

cs.IR↗

arXiv2024

Hyperbolic-constraint Point Cloud Reconstruction from Single RGB-D Images

Reconstructing desired objects and scenes has long been a primary goal in 3D computer vision. Single-view point cloud reconstruction has become a popular technique due to its low cost and accurate results. However, single-view reconstruction methods often rely on expensive CAD models and complex geometric priors. Effectively utilizing prior knowledge about the data remains a challenge. In this paper, we introduce hyperbolic space to 3D point cloud reconstruction, enabling the model to represent and understand complex hierarchical structures in point clouds with low distortion. We build upon previous methods by proposing a hyperbolic Chamfer distance and a regularized triplet loss to enhance the relationship between partial and complete point clouds. Additionally, we design adaptive boundary conditions to improve the model's understanding and reconstruction of 3D structures. Our model outperforms most existing models, and ablation studies demonstrate the significance of our model and its components. Experimental results show that our method significantly improves feature extraction capabilities. Our model achieves outstanding performance in 3D reconstruction tasks.

cs.CV↗

arXiv2024

TimeGPT in Load Forecasting: A Large Time Series Model Perspective

Machine learning models have made significant progress in load forecasting, but their forecast accuracy is limited in cases where historical load data is scarce. Inspired by the outstanding performance of large language models (LLMs) in computer vision and natural language processing, this paper aims to discuss the potential of large time series models in load forecasting with scarce historical data. Specifically, the large time series model is constructed as a time series generative pre-trained transformer (TimeGPT), which is trained on massive and diverse time series datasets consisting of 100 billion data points (e.g., finance, transportation, banking, web traffic, weather, energy, healthcare, etc.). Then, the scarce historical load data is used to fine-tune the TimeGPT, which helps it to adapt to the data distribution and characteristics associated with load forecasting. Simulation results show that TimeGPT outperforms the benchmarks (e.g., popular machine learning models and statistical models) for load forecasting on several real datasets with scarce training samples, particularly for short look-ahead times. However, it cannot be guaranteed that TimeGPT is always superior to benchmarks for load forecasting with scarce data, since the performance of TimeGPT may be affected by the distribution differences between the load data and the training data. In practical applications, we can divide the historical data into a training set and a validation set, and then use the validation set loss to decide whether TimeGPT is the best choice for a specific dataset.

cs.LG↗

arXiv2024

ChatHTTPFuzz: Large Language Model-Assisted IoT HTTP Fuzzing

Internet of Things (IoT) devices offer convenience through web interfaces, web VPNs, and other web-based services, all relying on the HTTP protocol. However, these externally exposed HTTP services resent significant security risks. Although fuzzing has shown some effectiveness in identifying vulnerabilities in IoT HTTP services, most state-of-the-art tools still rely on random mutation trategies, leading to difficulties in accurately understanding the HTTP protocol's structure and generating many invalid test cases. Furthermore, These fuzzers rely on a limited set of initial seeds for testing. While this approach initiates testing, the limited number and diversity of seeds hinder comprehensive coverage of complex scenarios in IoT HTTP services. In this paper, we investigate and find that large language models (LLMs) excel in parsing HTTP protocol data and analyzing code logic. Based on these findings, we propose a novel LLM-guided IoT HTTP fuzzing method, ChatHTTPFuzz, which automatically parses protocol fields and analyzes service code logic to generate protocol-compliant test cases. Specifically, we use LLMs to label fields in HTTP protocol data, creating seed templates. Second, The LLM analyzes service code to guide the generation of additional packets aligned with the code logic, enriching the seed templates and their field values. Finally, we design an enhanced Thompson sampling algorithm based on the exploration balance factor and mutation potential factor to schedule seed templates. We evaluate ChatHTTPFuzz on 14 different real-world IoT devices. It finds more vulnerabilities than SNIPUZZ, BOOFUZZ, and MUTINY. ChatHTTPFuzz has discovered 103 vulnerabilities, of which 68 are unique, and 23 have been assigned CVEs.

cs.CR↗

arXiv2024

The X-ray re-brightening of GRB afterglow revisited: a possible signature from activity of the central engine

Long-duration gamma-ray bursts (GRBs) are thought to be from core collapse of massive stars, and a rapidly spinning magnetar or black hole may be formed as the central engine. The extended emission in the prompt emission, flares and plateaus in X-ray afterglow, are proposed to be as the signature of central engine re-activity. However, the directly evidence from observations of identifying the central engines remain an open question. In this paper, we systemically search for long-duration GRBs that consist of bumps in X-ray afterglow detected by Swift/XRT, and find that the peak time of the X-ray bumps exhibit bimodal distribution (defined as early and late bumps) with division line at $t=7190$ s. Although we cannot rule out that such a bimodality arises from selection effects. We proposed that the long-duration GRBs with an early (or late) bumps may be originated from the fall-back accretion onto a new-born magnetar (or black hole). By adopting MCMC method to fit the early (or late) bumps of X-ray afterglow with the fall-back accretion of magnetar (or black hole), it is found that the initial surface magnetic filed and period of magnetars for most early bumps are clustered around $5.88\times10^{13}$ G and $1.04$ ms, respectively. Meanwhile, the derived accretion mass of black hole for late bumps is range of $[4\times10^{-4}, 1.8\times10^{-2}]~M_{\odot}$, and the typical fall-back radius is distributed range of $[1.04, 4.23]\times 10^{11}$ cm which is consistent with the typical radius of a Wolf-Rayet star. However, we also find that the fall-back accretion magnetar model is disfavored by the late bumps, but the fall-back accretion of black hole model can not be ruled out to interpret the early bumps of X-ray afterglow.

astro-ph.HE↗

arXiv2024

Towards a Unified View of Preference Learning for Large Language Models: A Survey

Large Language Models (LLMs) exhibit remarkably powerful capabilities. One of the crucial factors to achieve success is aligning the LLM's output with human preferences. This alignment process often requires only a small amount of data to efficiently enhance the LLM's performance. While effective, research in this area spans multiple domains, and the methods involved are relatively complex to understand. The relationships between different methods have been under-explored, limiting the development of the preference alignment. In light of this, we break down the existing popular alignment strategies into different components and provide a unified framework to study the current alignment strategies, thereby establishing connections among them. In this survey, we decompose all the strategies in preference learning into four components: model, data, feedback, and algorithm. This unified view offers an in-depth understanding of existing alignment algorithms and also opens up possibilities to synergize the strengths of different strategies. Furthermore, we present detailed working examples of prevalent existing algorithms to facilitate a comprehensive understanding for the readers. Finally, based on our unified perspective, we explore the challenges and future research directions for aligning large language models with human preferences.

cs.CL↗

arXiv2024

SceneDreamer360: Text-Driven 3D-Consistent Scene Generation with Panoramic Gaussian Splatting

Text-driven 3D scene generation has seen significant advancements recently. However, most existing methods generate single-view images using generative models and then stitch them together in 3D space. This independent generation for each view often results in spatial inconsistency and implausibility in the 3D scenes. To address this challenge, we proposed a novel text-driven 3D-consistent scene generation model: SceneDreamer360. Our proposed method leverages a text-driven panoramic image generation model as a prior for 3D scene generation and employs 3D Gaussian Splatting (3DGS) to ensure consistency across multi-view panoramic images. Specifically, SceneDreamer360 enhances the fine-tuned Panfusion generator with a three-stage panoramic enhancement, enabling the generation of high-resolution, detail-rich panoramic images. During the 3D scene construction, a novel point cloud fusion initialization method is used, producing higher quality and spatially consistent point clouds. Our extensive experiments demonstrate that compared to other methods, SceneDreamer360 with its panoramic image generation and 3DGS can produce higher quality, spatially consistent, and visually appealing 3D scenes from any text prompt. Our codes are available at \url{https://github.com/liwrui/SceneDreamer360}.

cs.CV↗

arXiv2024

SHMamba: Structured Hyperbolic State Space Model for Audio-Visual Question Answering

The Audio-Visual Question Answering (AVQA) task holds significant potential for applications. Compared to traditional unimodal approaches, the multi-modal input of AVQA makes feature extraction and fusion processes more challenging. Euclidean space is difficult to effectively represent multi-dimensional relationships of data. Especially when extracting and processing data with a tree structure or hierarchical structure, Euclidean space is not suitable as an embedding space. Additionally, the self-attention mechanism in Transformers is effective in capturing the dynamic relationships between elements in a sequence. However, the self-attention mechanism's limitations in window modeling and quadratic computational complexity reduce its effectiveness in modeling long sequences. To address these limitations, we propose SHMamba: Structured Hyperbolic State Space Model to integrate the advantages of hyperbolic geometry and state space models. Specifically, SHMamba leverages the intrinsic properties of hyperbolic space to represent hierarchical structures and complex relationships in audio-visual data. Meanwhile, the state space model captures dynamic changes over time by globally modeling the entire sequence. Furthermore, we introduce an adaptive curvature hyperbolic alignment module and a cross fusion block to enhance the understanding of hierarchical structures and the dynamic exchange of cross-modal information, respectively. Extensive experiments demonstrate that SHMamba outperforms previous methods with fewer parameters and computational costs. Our learnable parameters are reduced by 78.12\%, while the average performance improves by 2.53\%. Experiments show that our method demonstrates superiority among all current major methods and is more suitable for practical application scenarios.

cs.AI↗

arXiv2024

Diagnosing Emergent Isotropy in Anisotropic Holographic Systems using Quantum Information Measures

This study presents a comprehensive investigation of anisotropy in a holographic p-wave superconductor model, revealing novel insights into the behavior of quantum information measures in strongly coupled systems. Through rigorous semi-analytical methods, we uncover the existence of an isotropic point emerging at a critical temperature $T_{II}$, marking a significant transition in the system's anisotropic properties. We offer a systematic analysis of the mechanisms driving anisotropy and isotropy transitions, finding that this phenomenon is unique to the p-wave model and absent in other anisotropic systems like anisotropic axion models with metal-insulator transitions. We propose that the explicit component dependence of the vector field manifesting anisotropy is the key driver of the emergent isotropy. Our analysis of holographic entanglement entropy (HEE), entanglement wedge cross-section (EWCS), and butterfly velocity demonstrates their distinct sensitivities to bulk anisotropy, with EWCS and butterfly velocity emerging as superior probes for detecting the isotropic point.

hep-th↗

arXiv2024

Can Large Language Models Always Solve Easy Problems if They Can Solve Harder Ones?

Large language models (LLMs) have demonstrated impressive capabilities, but still suffer from inconsistency issues (e.g. LLMs can react differently to disturbances like rephrasing or inconsequential order change). In addition to these inconsistencies, we also observe that LLMs, while capable of solving hard problems, can paradoxically fail at easier ones. To evaluate this hard-to-easy inconsistency, we develop the ConsisEval benchmark, where each entry comprises a pair of questions with a strict order of difficulty. Furthermore, we introduce the concept of consistency score to quantitatively measure this inconsistency and analyze the potential for improvement in consistency by relative consistency score. Based on comprehensive experiments across a variety of existing models, we find: (1) GPT-4 achieves the highest consistency score of 92.2\% but is still inconsistent to specific questions due to distraction by redundant information, misinterpretation of questions, etc.; (2) models with stronger capabilities typically exhibit higher consistency, but exceptions also exist; (3) hard data enhances consistency for both fine-tuning and in-context learning. Our data and code will be publicly available on GitHub.

cs.CL↗

arXiv2024

Unlocking Efficiency in Large Language Model Inference: A Comprehensive Survey of Speculative Decoding

To mitigate the high inference latency stemming from autoregressive decoding in Large Language Models (LLMs), Speculative Decoding has emerged as a novel decoding paradigm for LLM inference. In each decoding step, this method first drafts several future tokens efficiently and then verifies them in parallel. Unlike autoregressive decoding, Speculative Decoding facilitates the simultaneous decoding of multiple tokens per step, thereby accelerating inference. This paper presents a comprehensive overview and analysis of this promising decoding paradigm. We begin by providing a formal definition and formulation of Speculative Decoding. Then, we organize in-depth discussions on its key facets, such as drafter selection and verification strategies. Furthermore, we present a comparative analysis of leading methods under third-party testing environments. We aim for this work to serve as a catalyst for further research on Speculative Decoding, ultimately contributing to more efficient LLM inference.

cs.CL↗

arXiv2024

Explainable Modeling for Wind Power Forecasting: A Glass-Box Approach with High Accuracy

Machine learning models (e.g., neural networks) achieve high accuracy in wind power forecasting, but they are usually regarded as black boxes that lack interpretability. To address this issue, the paper proposes a glass-box approach that combines high accuracy with transparency for wind power forecasting. Specifically, the core is to sum up the feature effects by constructing shape functions, which effectively map the intricate non-linear relationships between wind power output and input features. Furthermore, the forecasting model is enriched by incorporating interaction terms that adeptly capture interdependencies and synergies among the input features. The additive nature of the proposed glass-box approach ensures its interpretability. Simulation results show that the proposed glass-box approach effectively interprets the results of wind power forecasting from both global and instance perspectives. Besides, it outperforms most benchmark models and exhibits comparable performance to the best-performing neural networks. This dual strength of transparency and high accuracy positions the proposed glass-box approach as a compelling choice for reliable wind power forecasting.

cs.LG↗