Large language models (LLMs) have been widely integrated into information retrieval to advance traditional techniques. However, effectively enabling LLMs to seek accurate knowledge in complex tasks remains a challenge due to the complexity of multi-hop queries as well as the irrelevant retrieved content. To address these limitations, we propose EXSEARCH, an agentic search framework, where the LLM learns to retrieve useful information as the reasoning unfolds through a self-incentivized process. At each step, the LLM decides what to retrieve (thinking), triggers an external retriever (search), and extracts fine-grained evidence (recording) to support next-step reasoning. To enable LLM with this capability, EXSEARCH adopts a Generalized Expectation-Maximization algorithm. In the E-step, the LLM generates multiple search trajectories and assigns an importance weight to each; the M-step trains the LLM on them with a re-weighted loss function. This creates a self-incentivized loop, where the LLM iteratively learns from its own generated data, progressively improving itself for search. We further theoretically analyze this training process, establishing convergence guarantees. Extensive experiments on four knowledge-intensive benchmarks show that EXSEARCH substantially outperforms baselines, e.g., +7.8% improvement on exact match score. Motivated by these promising results, we introduce EXSEARCH-Zoo, an extension that extends our method to broader scenarios, to facilitate future work.
AI agents that leverage Large Language Models (LLMs) are increasingly becoming core building blocks of modern software systems. A wide range of frameworks is now available to support the specification of such applications. These frameworks enable the definition of agent setups using natural language prompting, which specifies the roles, goals, and tools assigned to the various agents involved. Within such setups, agent behavior is non-deterministic for any given input, highlighting the critical need for robust debugging and observability tools. In this work, we explore the use of process and causal discovery applied to agent execution trajectories as a means of enhancing developer observability. This approach aids in monitoring and understanding the emergent variability in agent behavior. Additionally, we complement this with LLM-based static analysis techniques to distinguish between intended and unintended behavioral variability. We argue that such instrumentation is essential for giving developers greater control over evolving specifications and for identifying aspects of functionality that may require more precise and explicit definitions.
As large language models (LLMs) become integrated into sensitive workflows, concerns grow over their potential to leak confidential information. We propose TrojanStego, a novel threat model in which an adversary fine-tunes an LLM to embed sensitive context information into natural-looking outputs via linguistic steganography, without requiring explicit control over inference inputs. We introduce a taxonomy outlining risk factors for compromised LLMs, and use it to evaluate the risk profile of the threat. To implement TrojanStego, we propose a practical encoding scheme based on vocabulary partitioning learnable by LLMs via fine-tuning. Experimental results show that compromised models reliably transmit 32-bit secrets with 87% accuracy on held-out prompts, reaching over 97% accuracy using majority voting across three generations. Further, they maintain high utility, can evade human detection, and preserve coherence. These results highlight a new class of LLM data exfiltration attacks that are passive, covert, practical, and dangerous.
We present REARANK, a large language model (LLM)-based listwise reasoning reranking agent. REARANK explicitly reasons before reranking, significantly improving both performance and interpretability. Leveraging reinforcement learning and data augmentation, REARANK achieves substantial improvements over baseline models across popular information retrieval benchmarks, notably requiring only 179 annotated samples. Built on top of Qwen2.5-7B, our REARANK-7B demonstrates performance comparable to GPT-4 on both in-domain and out-of-domain benchmarks and even surpasses GPT-4 on reasoning-intensive BRIGHT benchmarks. These results underscore the effectiveness of our approach and highlight how reinforcement learning can enhance LLM reasoning capabilities in reranking.
Autonomous agents, which perceive environments and take actions to achieve goals, have become increasingly feasible with the advancements in large language models (LLMs). However, current powerful agents often depend on sophisticated prompt engineering combined with closed-source LLMs like GPT-4. Although training open-source LLMs using expert trajectories from teacher models has yielded some improvements in agent capabilities, this approach still faces limitations such as performance plateauing and error propagation. To mitigate these challenges, we propose STeP, a novel method for improving LLM-based agent training. We synthesize self-reflected trajectories that include reflections and corrections of error steps, which enhance the effectiveness of LLM agents in learning from teacher models, enabling them to become agents capable of self-reflecting and correcting. We also introduce partial masking strategy that prevents the LLM from internalizing incorrect or suboptimal steps. Experiments demonstrate that our method improves agent performance across three representative tasks: ALFWorld, WebShop, and SciWorld. For the open-source model LLaMA2-7B-Chat, when trained using self-reflected trajectories constructed with Qwen1.5-110B-Chat as the teacher model, it achieves comprehensive improvements with less training data compared to agents trained exclusively on expert trajectories.
Web agents powered by Large Language Models (LLMs) show promise for next-generation AI, but their limited reasoning in uncertain, dynamic web environments hinders robust deployment. In this paper, we identify key reasoning skills essential for effective web agents, i.e., reflection & lookahead, branching, and rollback, and curate trajectory data that exemplifies these abilities by reconstructing the agent's (inference-time) reasoning algorithms into chain-of-thought rationales. We conduct experiments in the agent self-improving benchmark, OpenWebVoyager, and demonstrate that distilling salient reasoning patterns into the backbone LLM via simple fine-tuning can substantially enhance its performance. Our approach yields significant improvements across multiple benchmarks, including WebVoyager, Mind2web-live, and SimpleQA (web search), highlighting the potential of targeted reasoning skill enhancement for web agents.
Large language models (LLMs) are revolutionizing education, with LLM-based agents playing a key role in simulating student behavior. A major challenge in student simulation is modeling the diverse learning patterns of students at various cognitive levels. However, current LLMs, typically trained as ``helpful assistants'', target at generating perfect responses. As a result, they struggle to simulate students with diverse cognitive abilities, as they often produce overly advanced answers, missing the natural imperfections that characterize student learning and resulting in unrealistic simulations. To address this issue, we propose a training-free framework for student simulation. We begin by constructing a cognitive prototype for each student using a knowledge graph, which captures their understanding of concepts from past learning records. This prototype is then mapped to new tasks to predict student performance. Next, we simulate student solutions based on these predictions and iteratively refine them using a beam search method to better replicate realistic mistakes. To validate our approach, we construct the \texttt{Student\_100} dataset, consisting of $100$ students working on Python programming and $5,000$ learning records. Experimental results show that our method consistently outperforms baseline models, achieving $100\%$ improvement in simulation accuracy.
Recent advancements in AI agents have demonstrated their growing potential to drive and support scientific discovery. In this work, we introduce MLR-Bench, a comprehensive benchmark for evaluating AI agents on open-ended machine learning research. MLR-Bench includes three key components: (1) 201 research tasks sourced from NeurIPS, ICLR, and ICML workshops covering diverse ML topics; (2) MLR-Judge, an automated evaluation framework combining LLM-based reviewers with carefully designed review rubrics to assess research quality; and (3) MLR-Agent, a modular agent scaffold capable of completing research tasks through four stages: idea generation, proposal formulation, experimentation, and paper writing. Our framework supports both stepwise assessment across these distinct research stages, and end-to-end evaluation of the final research paper. We then use MLR-Bench to evaluate six frontier LLMs and an advanced coding agent, finding that while LLMs are effective at generating coherent ideas and well-structured papers, current coding agents frequently (e.g., in 80% of the cases) produce fabricated or invalidated experimental results--posing a major barrier to scientific reliability. We validate MLR-Judge through human evaluation, showing high agreement with expert reviewers, supporting its potential as a scalable tool for research evaluation. We open-source MLR-Bench to help the community benchmark, diagnose, and improve AI research agents toward trustworthy and transparent scientific discovery.
Zero-Shot Composed Image Retrieval (ZS-CIR) aims to retrieve target images
given a compositional query, consisting of a reference image and a modifying
text-without relying on annotated training data. Existing approaches often
generate a synthetic target text using large language models (LLMs) to serve as
an intermediate anchor between the compositional query and the target image.
Models are then trained to align the compositional query with the generated
text, and separately align images with their corresponding texts using
contrastive learning. However, this reliance on intermediate text introduces
error propagation, as inaccuracies in query-to-text and text-to-image mappings
accumulate, ultimately degrading retrieval performance. To address these
problems, we propose a novel framework by employing a Multimodal Reasoning
Agent (MRA) for ZS-CIR. MRA eliminates the dependence on textual intermediaries
by directly constructing triplets,
Large Language Models (LLMs) are increasingly used for decision making in embodied agents, yet existing safety evaluations often rely on coarse success rates and domain-specific setups, making it difficult to diagnose why and where these models fail. This obscures our understanding of embodied safety and limits the selective deployment of LLMs in high-risk physical environments. We introduce SAFEL, the framework for systematically evaluating the physical safety of LLMs in embodied decision making. SAFEL assesses two key competencies: (1) rejecting unsafe commands via the Command Refusal Test, and (2) generating safe and executable plans via the Plan Safety Test. Critically, the latter is decomposed into functional modules, goal interpretation, transition modeling, action sequencing, enabling fine-grained diagnosis of safety failures. To support this framework, we introduce EMBODYGUARD, a PDDL-grounded benchmark containing 942 LLM-generated scenarios covering both overtly malicious and contextually hazardous instructions. Evaluation across 13 state-of-the-art LLMs reveals that while models often reject clearly unsafe commands, they struggle to anticipate and mitigate subtle, situational risks. Our results highlight critical limitations in current LLMs and provide a foundation for more targeted, modular improvements in safe embodied reasoning.
Although LLMs demonstrate proficiency in several text-based reasoning and planning tasks, their implementation in robotics control is constrained by significant deficiencies: (1) LLM agents are designed to work mainly with textual inputs rather than visual conditions; (2) Current multimodal agents treat LLMs as static planners, which separates their reasoning from environment dynamics, resulting in actions that do not take domain-specific knowledge into account; and (3) LLMs are not designed to learn from visual interactions, which makes it harder for them to make better policies for specific domains. In this paper, we introduce EMAC+, an Embodied Multimodal Agent that collaboratively integrates LLM and VLM via a bidirectional training paradigm. Unlike existing methods, EMAC+ dynamically refines high-level textual plans generated by an LLM using real-time feedback from a VLM executing low-level visual control tasks. We address critical limitations of previous models by enabling the LLM to internalize visual environment dynamics directly through interactive experience, rather than relying solely on static symbolic mappings. Extensive experimental evaluations on ALFWorld and RT-1 benchmarks demonstrate that EMAC+ achieves superior task performance, robustness against noisy observations, and efficient learning. We also conduct thorough ablation studies and provide detailed analyses of success and failure cases.
Large Language Models (LLMs) have extended their impact beyond Natural Language Processing, substantially fostering the development of interdisciplinary research. Recently, various LLM-based agents have been developed to assist scientific discovery progress across multiple aspects and domains. Among these, computer-using agents, capable of interacting with operating systems as humans do, are paving the way to automated scientific problem-solving and addressing routines in researchers' workflows. Recognizing the transformative potential of these agents, we introduce ScienceBoard, which encompasses two complementary contributions: (i) a realistic, multi-domain environment featuring dynamic and visually rich scientific workflows with integrated professional software, where agents can autonomously interact via different interfaces to accelerate complex research tasks and experiments; and (ii) a challenging benchmark of 169 high-quality, rigorously validated real-world tasks curated by humans, spanning scientific-discovery workflows in domains such as biochemistry, astronomy, and geoinformatics. Extensive evaluations of agents with state-of-the-art backbones (e.g., GPT-4o, Claude 3.7, UI-TARS) show that, despite some promising results, they still fall short of reliably assisting scientists in complex workflows, achieving only a 15% overall success rate. In-depth analysis further provides valuable insights for addressing current agent limitations and more effective design principles, paving the way to build more capable agents for scientific discovery. Our code, environment, and benchmark are at https://qiushisun.github.io/ScienceBoard-Home/.
Recent trends are emerging in the use of Large Language Models (LLMs) as autonomous agents that take actions based on the content of the user text prompts. We intend to apply these concepts to the field of Control in space, enabling LLMs to play a significant role in the decision-making process for autonomous satellite operations. As a first step towards this goal, we have developed a pure LLM-based solution for the Kerbal Space Program Differential Games (KSPDG) challenge, a public software design competition where participants create autonomous agents for maneuvering satellites involved in non-cooperative space operations, running on the KSP game engine. Our approach leverages prompt engineering, few-shot prompting, and fine-tuning techniques to create an effective LLM-based agent that ranked 2nd in the competition. To the best of our knowledge, this work pioneers the integration of LLM agents into space research. The project comprises several open repositories to facilitate replication and further research. The codebase is accessible on \href{https://github.com/ARCLab-MIT/kspdg}{GitHub}, while the trained models and datasets are available on \href{https://huggingface.co/OhhTuRnz}{Hugging Face}. Additionally, experiment tracking and detailed results can be reviewed on \href{https://wandb.ai/carrusk/huggingface}{Weights \& Biases
Large Language Models (LLMs) have shown promise in software engineering tasks, but evaluating their effectiveness in vulnerability detection is challenging due to the lack of high-quality datasets. Most existing datasets are limited to function-level labels, ignoring finer-grained vulnerability patterns and crucial contextual information. Also, poor data quality such as mislabeling, inconsistent annotations, and duplicates can lead to inflated performance and weak generalization. Moreover, by including only the functions, these datasets miss broader program context, like data/control dependencies and interprocedural interactions, that are essential for accurately understanding real-world security flaws. Without this context, detection models are evaluated under unrealistic assumptions. To address these limitations, this paper introduces SecVulEval, a benchmark designed to support fine-grained evaluation of LLMs and other detection methods with rich contextual information. SecVulEval focuses on real-world C/C++ vulnerabilities at the statement level. This granularity enables more precise evaluation of a model's ability to localize vulnerabilities, beyond simple binary classification at the function level. By incorporating rich contextual information, SecVulEval sets a new standard for vulnerability detection benchmarks in realistic scenarios. This benchmark includes 25,440 function samples covering 5,867 unique CVEs in C/C++ projects from 1999 to 2024. We evaluated the SOTA LLMs with a multi-agent-based approach. The evaluation on our dataset shows that the models are still far from accurately predicting vulnerable statements in a given function. The best-performing Claude-3.7-Sonnet model achieves 23.83% F1-score for detecting vulnerable statements with correct reasoning. Finally, we analyze the LLM outputs and provide insights into their behavior in vulnerability detection for C/C++.
Large language models (LLMs) have demonstrated remarkable capabilities across diverse tasks, but optimizing LLM-based agentic systems remains challenging due to the vast search space of agent configurations, prompting strategies, and communication patterns. Existing approaches often rely on heuristic-based tuning or exhaustive evaluation, which can be computationally expensive and suboptimal. This paper proposes Agentic Predictor, a lightweight predictor for efficient agentic workflow evaluation. Agentic Predictor is equipped with a multi-view workflow encoding technique that leverages multi-view representation learning of agentic systems by incorporating code architecture, textual prompts, and interaction graph features. To achieve high predictive accuracy while significantly reducing the number of required workflow evaluations for training a predictor, Agentic Predictor employs cross-domain unsupervised pretraining. By learning to approximate task success rates, Agentic Predictor enables fast and accurate selection of optimal agentic workflow configurations for a given task, significantly reducing the need for expensive trial-and-error evaluations. Experiments on a carefully curated benchmark spanning three domains show that our predictor outperforms state-of-the-art methods in both predictive accuracy and workflow utility, highlighting the potential of performance predictors in streamlining the design of LLM-based agentic workflows.
While showing sophisticated reasoning abilities, large language models (LLMs) still struggle with long-horizon decision-making tasks due to deficient exploration and long-term credit assignment, especially in sparse-reward scenarios. Inspired by the divide-and-conquer principle, we propose an innovative framework **GLIDER** (**G**rounding **L**anguage Models as Eff**I**cient **D**ecision-Making Agents via Offline Hi**E**rarchical **R**einforcement Learning) that introduces a parameter-efficient and generally applicable hierarchy to LLM policies. We develop a scheme where the low-level controller is supervised with abstract, step-by-step plans that are learned and instructed by the high-level policy. This design decomposes complicated problems into a series of coherent chain-of-thought reasoning sub-tasks, providing flexible temporal abstraction to significantly enhance exploration and learning for long-horizon tasks. Furthermore, GLIDER facilitates fast online adaptation to non-stationary environments owing to the strong transferability of its task-agnostic low-level skills. Experiments on ScienceWorld and ALFWorld benchmarks show that GLIDER achieves consistent performance gains, along with enhanced generalization capabilities.
The increasing number of academic papers poses significant challenges for researchers to efficiently acquire key details. While retrieval augmented generation (RAG) shows great promise in large language model (LLM) based automated question answering, previous works often isolate neural and symbolic retrieval despite their complementary strengths. Moreover, conventional single-view chunking neglects the rich structure and layout of PDFs, e.g., sections and tables. In this work, we propose NeuSym-RAG, a hybrid neural symbolic retrieval framework which combines both paradigms in an interactive process. By leveraging multi-view chunking and schema-based parsing, NeuSym-RAG organizes semi-structured PDF content into both the relational database and vectorstore, enabling LLM agents to iteratively gather context until sufficient to generate answers. Experiments on three full PDF-based QA datasets, including a self-annotated one AIRQA-REAL, show that NeuSym-RAG stably defeats both the vector-based RAG and various structured baselines, highlighting its capacity to unify both retrieval schemes and utilize multiple views. Code and data are publicly available at https://github.com/X-LANCE/NeuSym-RAG.
Coding with hardware description languages (HDLs) such as Verilog is a time-intensive and laborious task. With the rapid advancement of large language models (LLMs), there is increasing interest in applying LLMs to assist with HDL coding. Recent efforts have demonstrated the potential of LLMs in translating natural language to traditional HDL Verilog. Chisel, a next-generation HDL based on Scala, introduces higher-level abstractions, facilitating more concise, maintainable, and scalable hardware designs. However, the potential of using LLMs for Chisel code generation remains largely unexplored. This work proposes ReChisel, an LLM-based agentic system designed to enhance the effectiveness of Chisel code generation. ReChisel incorporates a reflection mechanism to iteratively refine the quality of generated code using feedback from compilation and simulation processes, and introduces an escape mechanism to break free from non-progress loops. Experiments demonstrate that ReChisel significantly improves the success rate of Chisel code generation, achieving performance comparable to state-of-the-art LLM-based agentic systems for Verilog code generation.
Planning represents a fundamental capability of intelligent agents, requiring comprehensive environmental understanding, rigorous logical reasoning, and effective sequential decision-making. While Large Language Models (LLMs) have demonstrated remarkable performance on certain planning tasks, their broader application in this domain warrants systematic investigation. This paper presents a comprehensive review of LLM-based planning. Specifically, this survey is structured as follows: First, we establish the theoretical foundations by introducing essential definitions and categories about automated planning. Next, we provide a detailed taxonomy and analysis of contemporary LLM-based planning methodologies, categorizing them into three principal approaches: 1) External Module Augmented Methods that combine LLMs with additional components for planning, 2) Finetuning-based Methods that involve using trajectory data and feedback signals to adjust LLMs in order to improve their planning abilities, and 3) Searching-based Methods that break down complex tasks into simpler components, navigate the planning space, or enhance decoding strategies to find the best solutions. Subsequently, we systematically summarize existing evaluation frameworks, including benchmark datasets, evaluation metrics and performance comparisons between representative planning methods. Finally, we discuss the underlying mechanisms enabling LLM-based planning and outline promising research directions for this rapidly evolving field. We hope this survey will serve as a valuable resource to inspire innovation and drive progress in this field.
This paper proposes FieldWorkArena, a benchmark for agentic AI targeting real-world field work. With the recent increase in demand for agentic AI, they are required to monitor and report safety and health incidents, as well as manufacturing-related incidents, that may occur in real-world work environments. Existing agentic AI benchmarks have been limited to evaluating web tasks and are insufficient for evaluating agents in real-world work environments, where complexity increases significantly. In this paper, we define a new action space that agentic AI should possess for real world work environment benchmarks and improve the evaluation function from previous methods to assess the performance of agentic AI in diverse real-world tasks. The dataset consists of videos captured on-site and documents actually used in factories and warehouses, and tasks were created based on interviews with on-site workers and managers. Evaluation results confirmed that performance evaluation considering the characteristics of Multimodal LLM (MLLM) such as GPT-4o is feasible. Additionally, the effectiveness and limitations of the proposed new evaluation method were identified. The complete dataset (HuggingFace) and evaluation program (GitHub) can be downloaded from the following website: https://en-documents.research.global.fujitsu.com/fieldworkarena/.
Large language models (LLMs) have demonstrated excellent capabilities in the field of biomedical question answering, but their application in real-world clinical consultations still faces core challenges. Existing systems rely on a one-way information transmission mode where patients must fully describe their symptoms in a single round, leading to nonspecific diagnostic recommendations when complaints are vague. Traditional multi-turn dialogue methods based on supervised learning are constrained by static data-driven paradigms, lacking generalizability and struggling to intelligently extract key clinical information. To address these limitations, we propose DoctorAgent-RL, a reinforcement learning (RL)-based multi-agent collaborative framework that models medical consultations as a dynamic decision-making process under uncertainty. The doctor agent continuously optimizes its questioning strategy within the RL framework through multi-turn interactions with the patient agent, dynamically adjusting its information-gathering path based on comprehensive rewards from the Consultation Evaluator. This RL fine-tuning mechanism enables LLMs to autonomously develop interaction strategies aligned with clinical reasoning logic, rather than superficially imitating patterns in existing dialogue data. Notably, we constructed MTMedDialog, the first English multi-turn medical consultation dataset capable of simulating patient interactions. Experiments demonstrate that DoctorAgent-RL outperforms existing models in both multi-turn reasoning capability and final diagnostic performance, demonstrating practical value in assisting clinical consultations. https://github.com/JarvisUSTC/DoctorAgent-RL
The emergence of agentic recommender systems powered by Large Language Models (LLMs) represents a paradigm shift in personalized recommendations, leveraging LLMs' advanced reasoning and role-playing capabilities to enable autonomous, adaptive decision-making. Unlike traditional recommendation approaches, agentic recommender systems can dynamically gather and interpret user-item interactions from complex environments, generating robust recommendation strategies that generalize across diverse scenarios. However, the field currently lacks standardized evaluation protocols to systematically assess these methods. To address this critical gap, we propose: (1) an interactive textual recommendation simulator incorporating rich user and item metadata and three typical evaluation scenarios (classic, evolving-interest, and cold-start recommendation tasks); (2) a unified modular framework for developing and studying agentic recommender systems; and (3) the first comprehensive benchmark comparing 10 classical and agentic recommendation methods. Our findings demonstrate the superiority of agentic systems and establish actionable design guidelines for their core components. The benchmark environment has been rigorously validated through an open challenge and remains publicly available with a continuously maintained leaderboard~\footnote[2]{https://tsinghua-fib-lab.github.io/AgentSocietyChallenge/pages/overview.html}, fostering ongoing community engagement and reproducible research. The benchmark is available at: \hyperlink{https://huggingface.co/datasets/SGJQovo/AgentRecBench}{https://huggingface.co/datasets/SGJQovo/AgentRecBench}.
Large language models (LLMs) have achieved remarkable results across diverse downstream tasks, but their monolithic nature restricts scalability and efficiency in complex problem-solving. While recent research explores multi-agent collaboration among LLMs, most approaches rely on static organizational structures that struggle to adapt as task complexity and agent numbers grow, resulting in coordination overhead and inefficiencies. To this end, we propose a puppeteer-style paradigm for LLM-based multi-agent collaboration, where a centralized orchestrator ("puppeteer") dynamically directs agents ("puppets") in response to evolving task states. This orchestrator is trained via reinforcement learning to adaptively sequence and prioritize agents, enabling flexible and evolvable collective reasoning. Experiments on closed- and open-domain scenarios show that this method achieves superior performance with reduced computational costs. Analyses further reveal that the key improvements consistently stem from the emergence of more compact, cyclic reasoning structures under the orchestrator's evolution.
This study presents the LLM-Agent-Controller, a multi-agent large language model (LLM) system developed to address a wide range of problems in control engineering (Control Theory). The system integrates a central controller agent with multiple specialized auxiliary agents, responsible for tasks such as controller design, model representation, control analysis, time-domain response, and simulation. A supervisor oversees high-level decision-making and workflow coordination, enhancing the system's reliability and efficiency. The LLM-Agent-Controller incorporates advanced capabilities, including Retrieval-Augmented Generation (RAG), Chain-of-Thought reasoning, self-criticism and correction, efficient memory handling, and user-friendly natural language communication. It is designed to function without requiring users to have prior knowledge of Control Theory, enabling them to input problems in plain language and receive complete, real-time solutions. To evaluate the system, we propose new performance metrics assessing both individual agents and the system as a whole. We test five categories of Control Theory problems and benchmark performance across three advanced LLMs. Additionally, we conduct a comprehensive qualitative conversational analysis covering all key services. Results show that the LLM-Agent-Controller successfully solved 83% of general tasks, with individual agents achieving an average success rate of 87%. Performance improved with more advanced LLMs. This research demonstrates the potential of multi-agent LLM architectures to solve complex, domain-specific problems. By integrating specialized agents, supervisory control, and advanced reasoning, the LLM-Agent-Controller offers a scalable, robust, and accessible solution framework that can be extended to various technical domains.
Large language models (LLMs) are reaching expert-level accuracy on medical diagnosis questions, yet their mistakes and the biases behind them pose life-critical risks. Bias linked to race, sex, and socioeconomic status is already well known, but a consistent and automatic testbed for measuring it is missing. To fill this gap, this paper presents AMQA -- an Adversarial Medical Question-Answering dataset -- built for automated, large-scale bias evaluation of LLMs in medical QA. AMQA includes 4,806 medical QA pairs sourced from the United States Medical Licensing Examination (USMLE) dataset, generated using a multi-agent framework to create diverse adversarial descriptions and question pairs. Using AMQA, we benchmark five representative LLMs and find surprisingly substantial disparities: even GPT-4.1, the least biased model tested, answers privileged-group questions over 10 percentage points more accurately than unprivileged ones. Compared with the existing benchmark CPV, AMQA reveals 15% larger accuracy gaps on average between privileged and unprivileged groups. Our dataset and code are publicly available at https://github.com/XY-Showing/AMQA to support reproducible research and advance trustworthy, bias-aware medical AI.
Large Language Models (LLMs) have recently been widely adopted in conversational agents. However, the increasingly long interactions between users and agents accumulate extensive dialogue records, making it difficult for LLMs with limited context windows to maintain a coherent long-term dialogue memory and deliver personalized responses. While retrieval-augmented memory systems have emerged to address this issue, existing methods often depend on single-granularity memory segmentation and retrieval. This approach falls short in capturing deep memory connections, leading to partial retrieval of useful information or substantial noise, resulting in suboptimal performance. To tackle these limits, we propose MemGAS, a framework that enhances memory consolidation by constructing multi-granularity association, adaptive selection, and retrieval. MemGAS is based on multi-granularity memory units and employs Gaussian Mixture Models to cluster and associate new memories with historical ones. An entropy-based router adaptively selects optimal granularity by evaluating query relevance distributions and balancing information completeness and noise. Retrieved memories are further refined via LLM-based filtering. Experiments on four long-term memory benchmarks demonstrate that MemGAS outperforms state-of-the-art methods on both question answer and retrieval tasks, achieving superior performance across different query types and top-K settings.
The Linux kernel is a critical system, serving as the foundation for numerous systems. Bugs in the Linux kernel can cause serious consequences, affecting billions of users. Fault localization (FL), which aims at identifying the buggy code elements in software, plays an essential role in software quality assurance. While recent LLM agents have achieved promising accuracy in FL on recent benchmarks like SWE-bench, it remains unclear how well these methods perform in the Linux kernel, where FL is much more challenging due to the large-scale code base, limited observability, and diverse impact factors. In this paper, we introduce LinuxFLBench, a FL benchmark constructed from real-world Linux kernel bugs. We conduct an empirical study to assess the performance of state-of-the-art LLM agents on the Linux kernel. Our initial results reveal that existing agents struggle with this task, achieving a best top-1 accuracy of only 41.6% at file level. To address this challenge, we propose LinuxFL$^+$, an enhancement framework designed to improve FL effectiveness of LLM agents for the Linux kernel. LinuxFL$^+$ substantially improves the FL accuracy of all studied agents (e.g., 7.2% - 11.2% accuracy increase) with minimal costs. Data and code are available at https://github.com/FudanSELab/LinuxFLBench.
Traffic signal control (TSC) is a core challenge in urban mobility, where real-time decisions must balance efficiency and safety. Existing methods - ranging from rule-based heuristics to reinforcement learning (RL) - often struggle to generalize to complex, dynamic, and safety-critical scenarios. We introduce VLMLight, a novel TSC framework that integrates vision-language meta-control with dual-branch reasoning. At the core of VLMLight is the first image-based traffic simulator that enables multi-view visual perception at intersections, allowing policies to reason over rich cues such as vehicle type, motion, and spatial density. A large language model (LLM) serves as a safety-prioritized meta-controller, selecting between a fast RL policy for routine traffic and a structured reasoning branch for critical cases. In the latter, multiple LLM agents collaborate to assess traffic phases, prioritize emergency vehicles, and verify rule compliance. Experiments show that VLMLight reduces waiting times for emergency vehicles by up to 65% over RL-only systems, while preserving real-time performance in standard conditions with less than 1% degradation. VLMLight offers a scalable, interpretable, and safety-aware solution for next-generation traffic signal control.
Large language models (LLMs) have shown remarkable performance across diverse reasoning and generation tasks, and are increasingly deployed as agents in dynamic environments such as code generation and recommendation systems. However, many real-world applications, such as high-frequency trading and real-time competitive gaming, require decisions under strict latency constraints, where faster responses directly translate into higher rewards. Despite the importance of this latency quality trade off, it remains underexplored in the context of LLM based agents. In this work, we present the first systematic study of this trade off in real time decision making tasks. To support our investigation, we introduce two new benchmarks: HFTBench, a high frequency trading simulation, and StreetFighter, a competitive gaming platform. Our analysis reveals that optimal latency quality balance varies by task, and that sacrificing quality for lower latency can significantly enhance downstream performance. To address this, we propose FPX, an adaptive framework that dynamically selects model size and quantization level based on real time demands. Our method achieves the best performance on both benchmarks, improving win rate by up to 80% in Street Fighter and boosting daily yield by up to 26.52% in trading, underscoring the need for latency aware evaluation and deployment strategies for LLM based agents. These results demonstrate the critical importance of latency aware evaluation and deployment strategies for real world LLM based agents. Our benchmarks are available at Latency Sensitive Benchmarks.
LLM-as-Judge has emerged as a scalable alternative to human evaluation, enabling large language models (LLMs) to provide reward signals in trainings. While recent work has explored multi-agent extensions such as multi-agent debate and meta-judging to enhance evaluation quality, the question of how intrinsic biases manifest in these settings remains underexplored. In this study, we conduct a systematic analysis of four diverse bias types: position bias, verbosity bias, chain-of-thought bias, and bandwagon bias. We evaluate these biases across two widely adopted multi-agent LLM-as-Judge frameworks: Multi-Agent-Debate and LLM-as-Meta-Judge. Our results show that debate framework amplifies biases sharply after the initial debate, and this increased bias is sustained in subsequent rounds, while meta-judge approaches exhibit greater resistance. We further investigate the incorporation of PINE, a leading single-agent debiasing method, as a bias-free agent within these systems. The results reveal that this bias-free agent effectively reduces biases in debate settings but provides less benefit in meta-judge scenarios. Our work provides a comprehensive study of bias behavior in multi-agent LLM-as-Judge systems and highlights the need for targeted bias mitigation strategies in collaborative evaluation settings.