arXiv subjects

Yuan Yuan

Publications and source records attributed to Yuan Yuan.

At least 109 records · Page 6Linked to original sources

arXiv2025

TurnaboutLLM: A Deductive Reasoning Benchmark from Detective Games

This paper introduces TurnaboutLLM, a novel framework and dataset for evaluating the deductive reasoning abilities of Large Language Models (LLMs) by leveraging the interactive gameplay of detective games Ace Attorney and Danganronpa. The framework tasks LLMs with identifying contradictions between testimonies and evidences within long narrative contexts, a challenging task due to the large answer space and diverse reasoning types presented by its questions. We evaluate twelve state-of-the-art LLMs on the dataset, hinting at limitations of popular strategies for enhancing deductive reasoning such as extensive thinking and Chain-of-Thought prompting. The results also suggest varying effects of context size, the number of reasoning step and answer space size on model performance. Overall, TurnaboutLLM presents a substantial challenge for LLMs' deductive reasoning abilities in complex, narrative-rich environments.

cs.CL↗

arXiv2025

ClusterRCA: An End-to-End Approach for Network Fault Localization and Classification for HPC System

Network failure diagnosis is challenging yet critical for high-performance computing (HPC) systems. Existing methods cannot be directly applied to HPC scenarios due to data heterogeneity and lack of accuracy. This paper proposes a novel framework, called ClusterRCA, to localize culprit nodes and determine failure types by leveraging multimodal data. ClusterRCA extracts features from topologically connected network interface controller (NIC) pairs to analyze the diverse, multimodal data in HPC systems. To accurately localize culprit nodes and determine failure types, ClusterRCA combines classifier-based and graph-based approaches. A failure graph is constructed based on the output of the state classifier, and then it performs a customized random walk on the graph to localize the root cause. Experiments on datasets collected by a top-tier global HPC device vendor show ClusterRCA achieves high accuracy in diagnosing network failure for HPC systems. ClusterRCA also maintains robust performance across different application scenarios.

cs.DC↗

arXiv2025

PE-MA: Parameter-Efficient Co-Evolution of Multi-Agent Systems

Multi-Agent Systems have recently emerged as a promising paradigm for collaborative reasoning and solving complex tasks. However, the design of collaborative learning algorithms in multi-agent systems faces several challenges, including high communication overhead and insufficient agent-level personalization. In this paper, we propose PE-MA (Parameter-Efficient Multi-Agent Co-Evolution), a novel collaboration framework that supports efficient, scalable, and personalized co-evolution in multi-agent systems. In PE-MA, each agent maintains a lightweight personalized adapter to support agent-specific behavior, while a shared adapter is collaboratively optimized across neighboring agents. This design balances global coordination with local adaptation under heterogeneous environments. We achieve an asymptotically optimal convergence rate of O( 1/(NK)^(1/2) ), where N is the number of agents and K the local update steps.

cs.MA↗

arXiv2025

Crystalline electric field excitations in Weyl semimetal \textit{R}AlSi (\textit{R} = Ce, Pr and Nd)

The rare earth intermetallic system \textit{R}Al\textit{X} (\textit{R} = rare earth elements, \textit{X} = Si and Ge) is known to be a promising candidate of magnetic Weyl semimetal. Due to the complex interactions between the rare earth elements and surrounding atoms, as well as hybridization with itinerant electrons, this family likely possesses highly intriguing and novel magnetic structures and thus exhibits dynamic behaviors. We systematically probe polycrystalline samples of \textit{R}AlSi (\textit{R} = La, Ce, Pr and Nd) combining inelastic neutron scattering (INS), heat capacity and magnetic susceptibility measurements. The INS measurements identify well-resolved crystalline electric field (CEF) excitations at 19.2 and 24.9 meV in CeAlSi, at 5.4 meV in PrAlSi, and at 2.5 and 4.2 meV in NdAlSi. We analyzed the INS data using the corresponding CEF models and determined the CEF parameters and ground state wave functions of \textit{R}AlSi (\textit{R} = Ce, Pr and Nd). Our results suggest strong single-ion anisotropy in their ground states: $|\pm3/2\rangle$ (94.5\%) in CeAlSi, $|\pm3\rangle$ (99.2\%) in PrAlSi, and $|\pm9/2\rangle$ (76.2\%) in NdAlSi. Notably, the weaker anisotropy and strong exchange interactions in NdAlSi promote competing magnetic orders and CEF splitting at low temperature, contrasting with the robust CEF levels in magnetic states of CeAlSi and PrAlSi.

cond-mat.str-el↗

arXiv2025

From Hard Refusals to Safe-Completions: Toward Output-Centric Safety Training

Large Language Models used in ChatGPT have traditionally been trained to learn a refusal boundary: depending on the user's intent, the model is taught to either fully comply or outright refuse. While this is a strong mitigation for explicitly malicious prompts, focusing safety training on refusals can lead to brittleness for prompts with obscured user intent. Binary refusal boundaries are especially ill-suited for dual-use cases (such as biology or cybersecurity), where a user request can be answered safely at a high level, but in some cases can lead to malicious uplift if sufficiently detailed or actionable. As an alternative, we propose safe-completions: a safety-training approach that centers on the safety of the assistant's output, rather than a binary classification of the user's intent. Safe-completions seek to maximize helpfulness within the safety policy's constraints. We incorporated this approach into GPT-5 and find that across both production comparisons and internally controlled experiments, safe-completion training improves safety (especially on dual-use prompts), reduces the severity of residual safety failures, and substantially increases model helpfulness.

cs.CY↗

arXiv2025

Dynamic Coupling of Infiltration-Soil Moisture Feedback:Emergent Vegetation Patterns in a Water-Vegetation Model

We present a modified water-vegetation model to investigate the mechanistic relationship between infiltration-soil moisture feedback and vegetation pattern in arid/semi-arid ecosystems. Employing Turing pattern formation theory, we drive conditions for diffusion-induced instability and analyze spatiotemporal dynamics near Turing-Hopf bifurcation points. Our key findings include: (i) The system exhibits rich dynamics including multiple stable equilibria, supercritical/subcritical Hopf bifurcations, bubble loops of limit cycles and homoclinic bifurcations. (ii) The system admits Turing-Hopf bifurcation. Using normal form theory, we establish the existence of quasiperiodic solutions and mixedmode oscillations near critical thresholds, providing a mathematical framework for predicting nonlinear ecological regime shifts. (iii) Soil moisture feedbacks govern critical transitions between three distinct ecosystem states: uniform vegetation covering, self-organized spatial patterns (labyrinth/gapped vegetation), and bare soil state, which demonstrates that soil moisture thresholds control the final state selection in this system.

math.DS↗

arXiv2025

CoCoEvo: Co-Evolution of Programs and Test Cases to Enhance Code Generation

Large Language Models (LLMs) have shown remarkable performance in automated code generation. However, existing approaches often rely heavily on pre-defined test cases, which become impractical in scenarios where such cases are unavailable. While prior works explore filtering techniques between programs and test cases, they overlook the refinement of test cases. To address this limitation, we introduce CoCoEvo, a novel LLM-based co-evolution framework that simultaneously evolves programs and test cases. CoCoEvo eliminates the dependency on pre-defined test cases by generating both programs and test cases directly from natural language problem descriptions and function headers. The framework employs specialized evolutionary operators, including LLM-based crossover and mutation operators for program evolution, along with an additional test case generation operator for test case evolution. Additionally, we propose optimization strategies such as a crossover rate scheduler to balance exploration and convergence, and a multi-objective optimization method for test case selection. Experimental results on multiple state-of-the-art LLMs demonstrate that CoCoEvo surpasses existing methods, achieving state-of-the-art performance in automated code generation and testing. These results underscore the potential of co-evolutionary techniques in advancing the field of automated programming.

cs.SE↗

arXiv2025

DistrAttention: An Efficient and Flexible Self-Attention Mechanism on Modern GPUs

The Transformer architecture has revolutionized deep learning, delivering the state-of-the-art performance in areas such as natural language processing, computer vision, and time series prediction. However, its core component, self-attention, has the quadratic time complexity relative to input sequence length, which hinders the scalability of Transformers. The exsiting approaches on optimizing self-attention either discard full-contextual information or lack of flexibility. In this work, we design DistrAttention, an effcient and flexible self-attention mechanism with the full context. DistrAttention achieves this by grouping data on the embedding dimensionality, usually referred to as $d$. We realize DistrAttention with a lightweight sampling and fusion method that exploits locality-sensitive hashing to group similar data. A block-wise grouping framework is further designed to limit the errors introduced by locality sensitive hashing. By optimizing the selection of block sizes, DistrAttention could be easily integrated with FlashAttention-2, gaining high-performance on modern GPUs. We evaluate DistrAttention with extensive experiments. The results show that our method is 37% faster than FlashAttention-2 on calculating self-attention. In ViT inference, DistrAttention is the fastest and the most accurate among approximate self-attention mechanisms. In Llama3-1B, DistrAttention still achieves the lowest inference time with only 1% accuray loss.

cs.LG↗

arXiv2025

MobiWorld: World Models for Mobile Wireless Network

Accurate modeling and simulation of mobile networks are essential for enabling intelligent and cost-effective network optimization. In this paper, we propose MobiWorld, a generative world model designed to support high-fidelity and flexible environment simulation for mobile network planning and optimization. Unlike traditional predictive models constrained by limited generalization capabilities, MobiWorld exhibits strong universality by integrating heterogeneous data sources, including sensors, mobile devices, and base stations, as well as multimodal data types such as sequences and images. It is capable of generating both network element-level observations (e.g., traffic load, user distribution) and system-level performance indicators (e.g., throughput, energy consumption) to support a wide range of planning and optimization tasks. Built upon advanced diffusion models, MobiWorld offers powerful controllable generation capabilities by modeling the joint distribution between mobile network data and diverse conditional factors including spatio temporal contexts, user behaviors, and optimization policies. This enables accurate simulation of dynamic network states under varying policy configurations, providing optimization agents with precise environmental feedback and facilitating effective decision-making without relying on costly real-network interactions. We demonstrate the effectiveness of MobiWorld in a collaborative energy-saving scenario, where an agent uses observations and rewards generated by MobiWorld to optimize base station sleep and user offloading policies. Experimental results show that MobiWorld exhibits strong controllable generation performance and outperforms traditional methods in energy optimization.

cs.NI↗

arXiv2025

Towards Exception Safety Code Generation with Intermediate Representation Agents Framework

Large Language Models (LLMs) often struggle with robust exception handling in generated code, leading to fragile programs that are prone to runtime errors. We propose Seeker, a novel multi-agent framework that enforces exception safety in LLM generated code through an Intermediate Representation (IR) approach. Seeker decomposes exception handling into five specialized agents: Scanner, Detector, Predator, Ranker, and Handler that collaboratively analyze code, detect fragile segments, retrieve best practice exception strategies, and inject robust handling code. We also introduce Common Exception Enumeration (CEE), a comprehensive knowledge base derived from official documentation, technical practices, and real world code, to standardize exception handling strategies. Seeker also incorporates a Deep Retrieval-Augmented Generation (Deep RAG) algorithm to efficiently navigate the exception inheritance hierarchy, cutting down search overhead by 93% while improving accuracy in identifying relevant exceptions. We evaluate Seeker on 15 open source Java projects and multiple benchmarks. Seeker outperforms state of the art baselines, improving exception handling precision by up to 37% and overall code robustness by 38% as measured by expert code review. It significantly closes the gap between LLM and human developers in exception management, achieving a 28% success rate on real world issue fixes (SWE bench) versus 19% by prior methods. Our framework preserves functional correctness of code while proactively handling errors, demonstrating a practical, generalizable solution for safer code generation. In this paper, we discuss the novelty of using intermediate representation and multi-agent collaboration for exception handling, and outline how Seeker can be extended to other programming languages and complex software engineering tasks, aligning LLM-generated code with industrial standard.

cs.SE↗

arXiv2025

Multilingual Retrieval Augmented Generation for Culturally-Sensitive Tasks: A Benchmark for Cross-lingual Robustness

The paradigm of retrieval-augmented generated (RAG) helps mitigate hallucinations of large language models (LLMs). However, RAG also introduces biases contained within the retrieved documents. These biases can be amplified in scenarios which are multilingual and culturally-sensitive, such as territorial disputes. We thus introduce BordIRLines, a dataset of territorial disputes paired with retrieved Wikipedia documents, across 49 languages. We evaluate the cross-lingual robustness of this RAG setting by formalizing several modes for multilingual retrieval. Our experiments on several LLMs show that incorporating perspectives from diverse languages can in fact improve robustness; retrieving multilingual documents best improves response consistency and decreases geopolitical bias over RAG with purely in-language documents. We also consider how RAG responses utilize presented documents, finding a much wider variance in the linguistic distribution of response citations, when querying in low-resource languages. Our further analyses investigate the various aspects of a cross-lingual RAG pipeline, from retrieval to document contents. We release our benchmark and code to support continued research towards equitable information access across languages at https://huggingface.co/datasets/borderlines/bordirlines.

cs.CL↗

arXiv2025

Enhancing Vector Quantization with Distributional Matching: A Theoretical and Empirical Study

The success of autoregressive models largely depends on the effectiveness of vector quantization, a technique that discretizes continuous features by mapping them to the nearest code vectors within a learnable codebook. Two critical issues in existing vector quantization methods are training instability and codebook collapse. Training instability arises from the gradient discrepancy introduced by the straight-through estimator, especially in the presence of significant quantization errors, while codebook collapse occurs when only a small subset of code vectors are utilized during training. A closer examination of these issues reveals that they are primarily driven by a mismatch between the distributions of the features and code vectors, leading to unrepresentative code vectors and significant data information loss during compression. To address this, we employ the Wasserstein distance to align these two distributions, achieving near 100\% codebook utilization and significantly reducing the quantization error. Both empirical and theoretical analyses validate the effectiveness of the proposed approach.

cs.CV↗

arXiv2025

How Well Do Large Language Models Serve as End-to-End Secure Code Agents for Python?

The rapid advancement of large language models (LLMs) such as GPT-4 has revolutionized the landscape of software engineering, positioning these models at the core of modern development practices. As we anticipate these models to evolve into the primary and trustworthy tools used in software development, ensuring the security of the code they produce becomes paramount. How well can LLMs serve as end-to-end secure code producers? This paper presents a systematic investigation into LLMs' inherent potential to generate code with fewer vulnerabilities. Specifically, We studied GPT-3.5 and GPT-4's capability to identify and repair vulnerabilities in the code generated by four popular LLMs including themselves (GPT-3.5, GPT-4, Code Llama, and CodeGeeX2). By manually or automatically reviewing 4,900 pieces of code, our study reveals that: (1) large language models lack awareness of scenario-relevant security risks, which leads to the generation of over 75% vulnerable code on the SecurityEval benchmark; (2) LLMs such as GPT-3.5 and GPT-4 are unable to precisely identify vulnerabilities in the code they generated; (3) GPT-3.5 and GPT-4 can achieve 33.2%~59.6% success rates in repairing the insecure code produced by the 4 LLMs, but they both perform poorly when repairing self-produced code, indicating self-repair "blind spots". To address the limitation of a single round of repair, we developed a lightweight tool that prompts LLMs to construct safer source code through an iterative repair procedure based on the insights gained from our study. Experiments show that assisted by semantic analysis engines, our tool significantly improves the success rates of repair to 65.9%~85.5%.

cs.SE↗

arXiv2025

A Survey of Physics-Informed AI for Complex Urban Systems

Urban systems are typical examples of complex systems, where the integration of physics-based modeling with artificial intelligence (AI) presents a promising paradigm for enhancing predictive accuracy, interpretability, and decision-making. In this context, AI excels at capturing complex, nonlinear relationships, while physics-based models ensure consistency with real-world laws and provide interpretable insights. We provide a comprehensive review of physics-informed AI methods in urban applications. The proposed taxonomy categorizes existing approaches into three paradigms - Physics-Integrated AI, Physics-AI Hybrid Ensemble, and AI-Integrated Physics - and further details seven representative methods. This classification clarifies the varying degrees and directions of physics-AI integration, guiding the selection and development of appropriate methods based on application needs and data availability. We systematically examine their applications across eight key urban domains: energy, environment, economy, transportation, information, public services, emergency management, and the urban system as a whole. Our analysis highlights how these methodologies leverage physical laws and data-driven models to address urban challenges, enhancing system reliability, efficiency, and adaptability. By synthesizing existing methodologies and their urban applications, we identify critical gaps and outline future research directions, paving the way toward next-generation intelligent urban system modeling.

physics.soc-ph↗

arXiv2025

TIME: TabPFN-Integrated Multimodal Engine for Robust Tabular-Image Learning

Tabular-image multimodal learning, which integrates structured tabular data with imaging data, holds great promise for a variety of tasks, especially in medical applications. Yet, two key challenges remain: (1) the lack of a standardized, pretrained representation for tabular data, as is commonly available in vision and language domains; and (2) the difficulty of handling missing values in the tabular modality, which are common in real-world medical datasets. To address these issues, we propose the TabPFN-Integrated Multimodal Engine (TIME), a novel multimodal framework that builds on the recently introduced tabular foundation model, TabPFN. TIME leverages TabPFN as a frozen tabular encoder to generate robust, strong embeddings that are naturally resilient to missing data, and combines them with image features from pretrained vision backbones. We explore a range of fusion strategies and tabular encoders, and evaluate our approach on both natural and medical datasets. Extensive experiments demonstrate that TIME consistently outperforms competitive baselines across both complete and incomplete tabular inputs, underscoring its practical value in real-world multimodal learning scenarios.

cs.CV↗

arXiv2025

MRGRP: Empowering Courier Route Prediction in Food Delivery Service with Multi-Relational Graph

Instant food delivery has become one of the most popular web services worldwide due to its convenience in daily life. A fundamental challenge is accurately predicting courier routes to optimize task dispatch and improve delivery efficiency. This enhances satisfaction for couriers and users and increases platform profitability. The current heuristic prediction method uses only limited human-selected task features and ignores couriers preferences, causing suboptimal results. Additionally, existing learning-based methods do not fully capture the diverse factors influencing courier decisions or the complex relationships among them. To address this, we propose a Multi-Relational Graph-based Route Prediction (MRGRP) method that models fine-grained correlations among tasks affecting courier decisions for accurate prediction. We encode spatial and temporal proximity, along with pickup-delivery relationships, into a multi-relational graph and design a GraphFormer architecture to capture these complex connections. We also introduce a route decoder that leverages courier information and dynamic distance and time contexts for prediction, using existing route solutions as references to improve outcomes. Experiments show our model achieves state-of-the-art route prediction on offline data from cities of various sizes. Deployed on the Meituan Turing platform, it surpasses the current heuristic algorithm, reaching a high route prediction accuracy of 0.819, essential for courier and user satisfaction in instant food delivery.

cs.AI↗

arXiv2025

A Survey of Large Language Model-Powered Spatial Intelligence Across Scales: Advances in Embodied Agents, Smart Cities, and Earth Science

Over the past year, the development of large language models (LLMs) has brought spatial intelligence into focus, with much attention on vision-based embodied intelligence. However, spatial intelligence spans a broader range of disciplines and scales, from navigation and urban planning to remote sensing and earth science. What are the differences and connections between spatial intelligence across these fields? In this paper, we first review human spatial cognition and its implications for spatial intelligence in LLMs. We then examine spatial memory, knowledge representations, and abstract reasoning in LLMs, highlighting their roles and connections. Finally, we analyze spatial intelligence across scales -- from embodied to urban and global levels -- following a framework that progresses from spatial memory and understanding to spatial reasoning and intelligence. Through this survey, we aim to provide insights into interdisciplinary spatial intelligence research and inspire future studies.

cs.AI↗

arXiv2025

UniFlow: A Foundation Model for Unified Urban Spatio-Temporal Flow Prediction

Urban spatio-temporal flow prediction, encompassing traffic flows and crowd flows, is crucial for optimizing city infrastructure and managing traffic and emergency responses. Traditional approaches have relied on separate models tailored to either grid-based data, representing cities as uniform cells, or graph-based data, modeling cities as networks of nodes and edges. In this paper, we build UniFlow, a foundational model for general urban flow prediction that unifies both grid-based and graphbased data. We first design a multi-view spatio-temporal patching mechanism to standardize different data into a consistent sequential format and then introduce a spatio-temporal transformer architecture to capture complex correlations and dynamics. To leverage shared spatio-temporal patterns across different data types and facilitate effective cross-learning, we propose SpatioTemporal Memory Retrieval Augmentation (ST-MRA). By creating structured memory modules to store shared spatio-temporal patterns, ST-MRA enhances predictions through adaptive memory retrieval. Extensive experiments demonstrate that UniFlow outperforms existing models in both grid-based and graph-based flow prediction, excelling particularly in scenarios with limited data availability, showcasing its superior performance and broad applicability. The datasets and code implementation have been released on https://github.com/YuanYuan98/UniFlow.

cs.LG↗