arXiv subjects

Xiangyang Li

Publications and source records attributed to Xiangyang Li.

At least 37 records · Page 2Linked to original sources

arXiv2025

Game-TARS: Pretrained Foundation Models for Scalable Generalist Multimodal Game Agents

We present Game-TARS, a generalist game agent trained with a unified, scalable action space anchored to human-aligned native keyboard-mouse inputs. Unlike API- or GUI-based approaches, this paradigm enables large-scale continual pre-training across heterogeneous domains, including OS, web, and simulation games. Game-TARS is pre-trained on over 500B tokens with diverse trajectories and multimodal data. Key techniques include a decaying continual loss to reduce causal confusion and an efficient Sparse-Thinking strategy that balances reasoning depth and inference cost. Experiments show that Game-TARS achieves about 2 times the success rate over the previous sota model on open-world Minecraft tasks, is close to the generality of fresh humans in unseen web 3d games, and outperforms GPT-5, Gemini-2.5-Pro, and Claude-4-Sonnet in FPS benchmarks. Scaling results on training-time and test-time confirm that the unified action space sustains improvements when scaled to cross-game and multimodal data. Our results demonstrate that simple, scalable action representations combined with large-scale pre-training provide a promising path toward generalist agents with broad computer-use abilities.

cs.AI↗

arXiv2025

Humanity's Last Code Exam: Can Advanced LLMs Conquer Human's Hardest Code Competition?

Code generation is a core capability of large language models (LLMs), yet mainstream benchmarks (e.g., APPs and LiveCodeBench) contain questions with medium-level difficulty and pose no challenge to advanced LLMs. To better reflected the advanced reasoning and code generation ability, We introduce Humanity's Last Code Exam (HLCE), comprising 235 most challenging problems from the International Collegiate Programming Contest (ICPC World Finals) and the International Olympiad in Informatics (IOI) spanning 2010 - 2024. As part of HLCE, we design a harmonized online-offline sandbox that guarantees fully reproducible evaluation. Through our comprehensive evaluation, we observe that even the strongest reasoning LLMs: o4-mini(high) and Gemini-2.5 Pro, achieve pass@1 rates of only 15.9% and 11.4%, respectively. Meanwhile, we propose a novel "self-recognition" task to measure LLMs' awareness of their own capabilities. Results indicate that LLMs' self-recognition abilities are not proportionally correlated with their code generation performance. Finally, our empirical validation of test-time scaling laws reveals that current advanced LLMs have substantial room for improvement on complex programming tasks. We expect HLCE to become a milestone challenge for code generation and to catalyze advances in high-performance reasoning and human-AI collaborative programming. Our code and dataset are also public available(https://github.com/Humanity-s-Last-Code-Exam/HLCE).

cs.SE↗

arXiv2025

ATGen: Adversarial Reinforcement Learning for Test Case Generation

Large Language Models (LLMs) excel at code generation, yet their outputs often contain subtle bugs, for which effective test cases are a critical bottleneck. Existing test generation methods, whether based on prompting or supervised fine-tuning, rely on static datasets. This imposes a ``fixed-difficulty ceiling'', fundamentally limiting their ability to uncover novel or more complex bugs beyond their training scope. To overcome this, we introduce ATGen, a framework that trains a test case generator via adversarial reinforcement learning. ATGen pits a test generator against an adversarial code generator that continuously crafts harder bugs to evade the current policy. This dynamic loop creates a curriculum of increasing difficulty challenging current policy. The test generator is optimized via Reinforcement Learning (RL) to jointly maximize ``Output Accuracy'' and ``Attack Success'', enabling it to learn a progressively stronger policy that breaks the fixed-difficulty ceiling of static training. Extensive experiments demonstrate that ATGen significantly outperforms state-of-the-art baselines. We further validate its practical utility, showing it serves as both a more effective filter for Best-of-N inference and a higher-quality reward source for training code generation models. Our work establishes a new, dynamic paradigm for improving the reliability of LLM-generated code.

cs.SE↗

arXiv2025

MTRec: Learning to Align with User Preferences via Mental Reward Models

Recommendation models are predominantly trained using implicit user feedback, since explicit feedback is often costly to obtain. However, implicit feedback, such as clicks, does not always reflect users' real preferences. For example, a user might click on a news article because of its attractive headline, but end up feeling uncomfortable after reading the content. In the absence of explicit feedback, such erroneous implicit signals may severely mislead recommender systems. In this paper, we propose MTRec, a novel sequential recommendation framework designed to align with real user preferences by uncovering their internal satisfaction on recommended items. Specifically, we introduce a mental reward model to quantify user satisfaction and propose a distributional inverse reinforcement learning approach to learn it. The learned mental reward model is then used to guide recommendation models to better align with users' real preferences. Our experiments show that MTRec brings significant improvements to a variety of recommendation models. We also deploy MTRec on an industrial short video platform and observe a 7 percent increase in average user viewing time.

cs.IR↗

arXiv2025

Purest Quantum State Identification

Quantum noise constitutes a fundamental obstacle to realizing practical quantum technologies. To address the pivotal challenge of identifying quantum systems least affected by noise, we introduce the purest quantum state identification, which can be used to improve the accuracy of quantum computation and communication. We formulate a rigorous paradigm for identifying the purest quantum state among $K$ unknown $n$-qubit quantum states using total $N$ quantum state copies. For incoherent strategies, we derive the first adaptive algorithm achieving error probability $\exp\left(- Ω\left(\frac{N H_1}{\log(K) 2^n }\right) \right)$, fundamentally improving quantum property learning through measurement optimization. By developing a coherent measurement protocol with error bound $\exp\left(- Ω\left(\frac{N H_2}{\log(K) }\right) \right)$, we demonstrate a significant separation from incoherent strategies, formally quantifying the power of quantum memory and coherent measurement. Furthermore, we establish a lower bound by demonstrating that all strategies with fixed two-outcome incoherent POVM must suffer error probability exceeding $ \exp\left( - O\left(\frac{NH_1}{2^n}\right)\right)$. This research advances the characterization of quantum noise through efficient learning frameworks. Our results establish theoretical foundations for noise-adaptive quantum property learning while delivering practical protocols for enhancing the reliability of quantum hardware.

quant-ph↗

arXiv2025

A Dynamic Service Offloading Algorithm Based on Lyapunov Optimization in Edge Computing

This study investigates the trade-off between system stability and offloading cost in collaborative edge computing. While collaborative offloading among multiple edge servers enhances resource utilization, existing methods often overlook the role of queue stability in overall system performance. To address this, a multi-hop data transmission model is developed, along with a cost model that captures both energy consumption and delay. A time-varying queue model is then introduced to maintain system stability. Based on Lyapunov optimization, a dynamic offloading algorithm (LDSO) is proposed to minimize offloading cost while ensuring long-term stability. Theoretical analysis and experimental results verify that the proposed LDSO achieves significant improvements in both cost efficiency and system stability compared to the state-of-the-art.

cs.NI↗

arXiv2025

LLM4MSR: An LLM-Enhanced Paradigm for Multi-Scenario Recommendation

As the demand for more personalized recommendation grows and a dramatic boom in commercial scenarios arises, the study on multi-scenario recommendation (MSR) has attracted much attention, which uses the data from all scenarios to simultaneously improve their recommendation performance. However, existing methods tend to integrate insufficient scenario knowledge and neglect learning personalized cross-scenario preferences, thus leading to sub-optimal performance. Meanwhile, though large language model (LLM) has shown great capability of reasoning and capturing semantic information, the high inference latency and high computation cost of tuning hinder its implementation in industrial recommender systems. To fill these gaps, we propose an LLM-enhanced paradigm LLM4MSR in this work. Specifically, we first leverage LLM to uncover multi-level knowledge from the designed scenario- and user-level prompt without fine-tuning the LLM, then adopt hierarchical meta networks to generate multi-level meta layers to explicitly improve the scenario-aware and personalized recommendation capability. Our experiments on KuaiSAR-small, KuaiSAR, and Amazon datasets validate significant advantages of LLM4MSR: (i) the effectiveness and compatibility with different multi-scenario backbone models, (ii) high efficiency and deployability on industrial recommender systems, and (iii) improved interpretability. The implemented code and data is available to ease reproduction.

cs.IR↗

arXiv2025

DERMARK: A Dynamic, Efficient and Robust Multi-bit Watermark for Large Language Models

As large language models (LLMs) grow more powerful, concerns over copyright infringement of LLM-generated texts have intensified. LLM watermarking has been proposed to trace unauthorized redistribution or resale of generated content by embedding identifiers within the text. Existing approaches primarily rely on one-bit watermarking, which only verifies whether a text was generated by a specific LLM. In contrast, multi-bit watermarking encodes richer information, enabling the identification of the specific LLM and user involved in generated or distributed content. However, current multi-bit methods directly embed the watermark into the text without considering its watermark capacity, which can result in failures, especially in low-entropy texts. In this paper, we analyze that the watermark embedding follows a normal distribution. We then derive a formal inequality to optimally segment the text for watermark embedding. Building upon this, we propose DERMARK, a dynamic, efficient, and robust multi-bit watermarking method that divides the text into variable-length segments for each watermark bit during the inference. Moreover, DERMARK incurs negligible overhead since no additional intermediate matrices are generated and achieves robustness against text editing by minimizing watermark extraction loss. Experiments demonstrate that, compared to SOTA, on average, our method reduces the number of tokens required per embedded bit by 25\%, reduces watermark embedding time by 50\%, and maintains high robustness against text modifications and watermark erasure attacks.

cs.CR↗

arXiv2025

CoIR: A Comprehensive Benchmark for Code Information Retrieval Models

Despite the substantial success of Information Retrieval (IR) in various NLP tasks, most IR systems predominantly handle queries and corpora in natural language, neglecting the domain of code retrieval. Code retrieval is critically important yet remains under-explored, with existing methods and benchmarks inadequately representing the diversity of code in various domains and tasks. Addressing this gap, we present COIR (Code Information Retrieval Benchmark), a robust and comprehensive benchmark specifically designed to assess code retrieval capabilities. COIR comprises ten meticulously curated code datasets, spanning eight distinctive retrieval tasks across seven diverse domains. We first discuss the construction of COIR and its diverse dataset composition. Further, we evaluate nine widely used retrieval models using COIR, uncovering significant difficulties in performing code retrieval tasks even with state-of-the-art systems. To facilitate easy adoption and integration within existing research workflows, COIR has been developed as a user-friendly Python framework, readily installable via pip. It shares same data schema as other popular benchmarks like MTEB and BEIR, enabling seamless cross-benchmark evaluations. Through COIR, we aim to invigorate research in the code retrieval domain, providing a versatile benchmarking tool that encourages further development and exploration of code retrieval systems. https://github.com/CoIR-team/coir.

cs.IR↗

arXiv2025

CrossLinear: Plug-and-Play Cross-Correlation Embedding for Time Series Forecasting with Exogenous Variables

Time series forecasting with exogenous variables is a critical emerging paradigm that presents unique challenges in modeling dependencies between variables. Traditional models often struggle to differentiate between endogenous and exogenous variables, leading to inefficiencies and overfitting. In this paper, we introduce CrossLinear, a novel Linear-based forecasting model that addresses these challenges by incorporating a plug-and-play cross-correlation embedding module. This lightweight module captures the dependencies between variables with minimal computational cost and seamlessly integrates into existing neural networks. Specifically, it captures time-invariant and direct variable dependencies while disregarding time-varying or indirect dependencies, thereby mitigating the risk of overfitting in dependency modeling and contributing to consistent performance improvements. Furthermore, CrossLinear employs patch-wise processing and a global linear head to effectively capture both short-term and long-term temporal dependencies, further improving its forecasting precision. Extensive experiments on 12 real-world datasets demonstrate that CrossLinear achieves superior performance in both short-term and long-term forecasting tasks. The ablation study underscores the effectiveness of the cross-correlation embedding module. Additionally, the generalizability of this module makes it a valuable plug-in for various forecasting tasks across different domains. Codes are available at https://github.com/mumiao2000/CrossLinear.

cs.LG↗

arXiv2025

KnowPath: Knowledge-enhanced Reasoning via LLM-generated Inference Paths over Knowledge Graphs

Large language models (LLMs) have demonstrated remarkable capabilities in various complex tasks, yet they still suffer from hallucinations. By incorporating and exploring external knowledge, such as knowledge graphs(KGs), LLM's ability to provide factual answers has been enhanced. This approach carries significant practical implications. However, existing methods suffer from three key limitations: insufficient mining of LLMs' internal knowledge, constrained generation of interpretable reasoning paths, and unclear fusion of internal and external knowledge. Therefore, we propose KnowPath, a knowledge-enhanced large model framework driven by the collaboration of internal and external knowledge. It relies on the internal knowledge of the LLM to guide the exploration of interpretable directed subgraphs in external knowledge graphs, better integrating the two knowledge sources for more accurate reasoning. Extensive experiments on multiple real-world datasets demonstrate the effectiveness of KnowPath. Our code and data are available at https://github.com/tize-72/KnowPath.

cs.AI↗

arXiv2025

TeleEval-OS: Performance evaluations of large language models for operations scheduling

The rapid advancement of large language models (LLMs) has significantly propelled progress in artificial intelligence, demonstrating substantial application potential across multiple specialized domains. Telecommunications operation scheduling (OS) is a critical aspect of the telecommunications industry, involving the coordinated management of networks, services, risks, and human resources to optimize production scheduling and ensure unified service control. However, the inherent complexity and domain-specific nature of OS tasks, coupled with the absence of comprehensive evaluation benchmarks, have hindered thorough exploration of LLMs' application potential in this critical field. To address this research gap, we propose the first Telecommunications Operation Scheduling Evaluation Benchmark (TeleEval-OS). Specifically, this benchmark comprises 15 datasets across 13 subtasks, comprehensively simulating four key operational stages: intelligent ticket creation, intelligent ticket handling, intelligent ticket closure, and intelligent evaluation. To systematically assess the performance of LLMs on tasks of varying complexity, we categorize their capabilities in telecommunications operation scheduling into four hierarchical levels, arranged in ascending order of difficulty: basic NLP, knowledge Q&A, report generation, and report analysis. On TeleEval-OS, we leverage zero-shot and few-shot evaluation methods to comprehensively assess 10 open-source LLMs (e.g., DeepSeek-V3) and 4 closed-source LLMs (e.g., GPT-4o) across diverse scenarios. Experimental results demonstrate that open-source LLMs can outperform closed-source LLMs in specific scenarios, highlighting their significant potential and value in the field of telecommunications operation scheduling.

cs.CL↗

arXiv2025

An Investigation of Large Language Models and Their Vulnerabilities in Spam Detection

Spam messages continue to present significant challenges to digital users, cluttering inboxes and posing security risks. Traditional spam detection methods, including rules-based, collaborative, and machine learning approaches, struggle to keep up with the rapidly evolving tactics employed by spammers. This project studies new spam detection systems that leverage Large Language Models (LLMs) fine-tuned with spam datasets. More importantly, we want to understand how LLM-based spam detection systems perform under adversarial attacks that purposefully modify spam emails and data poisoning attacks that exploit the differences between the training data and the massages in detection, to which traditional machine learning models are shown to be vulnerable. This experimentation employs two LLM models of GPT2 and BERT and three spam datasets of Enron, LingSpam, and SMSspamCollection for extensive training and testing tasks. The results show that, while they can function as effective spam filters, the LLM models are susceptible to the adversarial and data poisoning attacks. This research provides very useful insights for future applications of LLM models for information security.

cs.CR↗

arXiv2025

A Survey of Personalization: From RAG to Agent

Personalization has become an essential capability in modern AI systems, enabling customized interactions that align with individual user preferences, contexts, and goals. Recent research has increasingly concentrated on Retrieval-Augmented Generation (RAG) frameworks and their evolution into more advanced agent-based architectures within personalized settings to enhance user satisfaction. Building on this foundation, this survey systematically examines personalization across the three core stages of RAG: pre-retrieval, retrieval, and generation. Beyond RAG, we further extend its capabilities into the realm of Personalized LLM-based Agents, which enhance traditional RAG systems with agentic functionalities, including user understanding, personalized planning and execution, and dynamic generation. For both personalization in RAG and agent-based personalization, we provide formal definitions, conduct a comprehensive review of recent literature, and summarize key datasets and evaluation metrics. Additionally, we discuss fundamental challenges, limitations, and promising research directions in this evolving field. Relevant papers and resources are continuously updated at https://github.com/Applied-Machine-Learning-Lab/Awesome-Personalized-RAG-Agent.

cs.IR↗

arXiv2025

MoonCast: High-Quality Zero-Shot Podcast Generation

Recent advances in text-to-speech synthesis have achieved notable success in generating high-quality short utterances for individual speakers. However, these systems still face challenges when extending their capabilities to long, multi-speaker, and spontaneous dialogues, typical of real-world scenarios such as podcasts. These limitations arise from two primary challenges: 1) long speech: podcasts typically span several minutes, exceeding the upper limit of most existing work; 2) spontaneity: podcasts are marked by their spontaneous, oral nature, which sharply contrasts with formal, written contexts; existing works often fall short in capturing this spontaneity. In this paper, we propose MoonCast, a solution for high-quality zero-shot podcast generation, aiming to synthesize natural podcast-style speech from text-only sources (e.g., stories, technical reports, news in TXT, PDF, or Web URL formats) using the voices of unseen speakers. To generate long audio, we adopt a long-context language model-based audio modeling approach utilizing large-scale long-context speech data. To enhance spontaneity, we utilize a podcast generation module to generate scripts with spontaneous details, which have been empirically shown to be as crucial as the text-to-speech modeling itself. Experiments demonstrate that MoonCast outperforms baselines, with particularly notable improvements in spontaneity and coherence.

eess.AS↗

arXiv2025

SampleLLM: Optimizing Tabular Data Synthesis in Recommendations

Tabular data synthesis is crucial in machine learning, yet existing general methods-primarily based on statistical or deep learning models-are highly data-dependent and often fall short in recommender systems. This limitation arises from their difficulty in capturing complex distributions and understanding feature relationships from sparse and limited data, along with their inability to grasp semantic feature relations. Recently, Large Language Models (LLMs) have shown potential in generating synthetic data samples through few-shot learning and semantic understanding. However, they often suffer from inconsistent distribution and lack of diversity due to their inherent distribution disparity with the target dataset. To address these challenges and enhance tabular data synthesis for recommendation tasks, we propose a novel two-stage framework named SampleLLM to improve the quality of LLM-based tabular data synthesis for recommendations by ensuring better distribution alignment. In the first stage, SampleLLM employs LLMs with Chain-of-Thought prompts and diverse exemplars to generate data that closely aligns with the target dataset distribution, even when input samples are limited. The second stage uses an advanced feature attribution-based importance sampling method to refine feature relationships within the synthesized data, reducing any distribution biases introduced by the LLM. Experimental results on three recommendation datasets, two general datasets, and online deployment illustrate that SampleLLM significantly surpasses existing methods for recommendation tasks and holds promise for a broader range of tabular data scenarios.

cs.IR↗

arXiv2025

A-VL: Adaptive Attention for Large Vision-Language Models

The Large Vision-Language Model (LVLM) integrates computer vision and natural language processing techniques, offering substantial application potential. However, these models demand extensive resources during inference. Adaptive attention techniques can dynamically reduce computational redundancy and thus improve efficiency. Although current adaptive attention methods significantly reduce the memory requirements of Transformer-based language models, they are not tailored for LVLMs. We observe that LVLMs generate responses from both remote image tokens and local text tokens, and different modalities have different attention patterns. This observation inspires us to manage the attention for each modality separately. Specifically, for visual input, we store the cache of potentially useful information but only compute the most critical parts. For language input, we care more about local information. Based on our observation and analysis of vision-language attention patterns, we develop A-VL, a plug-and-play adaptive attention tailored for LVLM inference. Extensive evaluations on three vision-language tasks and five datasets show the effectiveness of our designs. Our approach A-VL outperforms existing adaptive attention methods in reducing memory usage and computational load without compromising performance.

cs.AI↗

arXiv2025

LLM4Rerank: LLM-based Auto-Reranking Framework for Recommendations

Reranking is a critical component in recommender systems, playing an essential role in refining the output of recommendation algorithms. Traditional reranking models have focused predominantly on accuracy, but modern applications demand consideration of additional criteria such as diversity and fairness. Existing reranking approaches often fail to harmonize these diverse criteria effectively at the model level. Moreover, these models frequently encounter challenges with scalability and personalization due to their complexity and the varying significance of different reranking criteria in diverse scenarios. In response, we introduce a comprehensive reranking framework enhanced by LLM, designed to seamlessly integrate various reranking criteria while maintaining scalability and facilitating personalized recommendations. This framework employs a fully connected graph structure, allowing the LLM to simultaneously consider multiple aspects such as accuracy, diversity, and fairness through a coherent Chain-of-Thought (CoT) process. A customizable input mechanism is also integrated, enabling the tuning of the language model's focus to meet specific reranking needs. We validate our approach using three popular public datasets, where our framework demonstrates superior performance over existing state-of-the-art reranking models in balancing multiple criteria.

cs.IR↗