Showing 1–50 of 181 results for author: Zhong, M

Agentic Tool Use in Large Language Models

Authors: Jinchao Hu, Meizhi Zhong, Kehai Chen, Xuefeng Bai, Min Zhang

Abstract: Large language models are increasingly being deployed as autonomous agents yet their real world effectiveness depends on reliable tools for information retrieval, computation and external action. Existing studies remain fragmented across tasks, tool types, and training settings, lacking a unified view of how tool-use methods differ and evolve. This paper organizes the literature into three paradig… ▽ More Large language models are increasingly being deployed as autonomous agents yet their real world effectiveness depends on reliable tools for information retrieval, computation and external action. Existing studies remain fragmented across tasks, tool types, and training settings, lacking a unified view of how tool-use methods differ and evolve. This paper organizes the literature into three paradigms: prompting as plug-and-play, supervised tool learning and reward-driven tool policy learning, analyzes their methods, strengths and failure modes, reviews the evaluation landscape and highlights key challenges, aiming to address this fragmentation and provide a more structured evolutionary view of agentic tool use. △ Less

Submitted 1 April, 2026; originally announced April 2026.

DIVE: Scaling Diversity in Agentic Task Synthesis for Generalizable Tool Use

Authors: Aili Chen, Chi Zhang, Junteng Liu, Jiangjie Chen, Chengyu Du, Yunji Li, Ming Zhong, Qin Wang, Zhengmao Zhu, Jiayuan Song, Ke Ji, Junxian He, Pengyu Zhao, Yanghua Xiao

Abstract: Recent work synthesizes agentic tasks for post-training tool-using LLMs, yet robust generalization under shifts in tasks and toolsets remains an open challenge. We trace this brittleness to insufficient diversity in synthesized tasks. Scaling diversity is difficult because training requires tasks to remain executable and verifiable, while generalization demands coverage of diverse tool types, tool… ▽ More Recent work synthesizes agentic tasks for post-training tool-using LLMs, yet robust generalization under shifts in tasks and toolsets remains an open challenge. We trace this brittleness to insufficient diversity in synthesized tasks. Scaling diversity is difficult because training requires tasks to remain executable and verifiable, while generalization demands coverage of diverse tool types, toolset combinations, and heterogeneous tool-use patterns. We propose DIVE, an evidence-driven recipe that inverts synthesis order, executing diverse, real-world tools first and reverse-deriving tasks strictly entailed by the resulting traces, thereby providing grounding by construction. DIVE scales structural diversity along two controllable axes, tool-pool coverage and per-task toolset variety, and an Evidence Collection--Task Derivation loop further induces rich multi-step tool-use patterns across 373 tools in five domains. Training Qwen3-8B on DIVE data (48k SFT + 3.2k RL) improves by +22 average points across 9 OOD benchmarks and outperforms the strongest 8B baseline by +68. Remarkably, controlled scaling analysis reveals that diversity scaling consistently outperforms quantity scaling for OOD generalization, even with 4x less data. △ Less

Submitted 10 March, 2026; originally announced March 2026.

GAST: Gradient-aligned Sparse Tuning of Large Language Models with Data-layer Selection

Authors: Kai Yao, Zhenghan Song, Kaixin Wu, Mingjie Zhong, Danzhao Cheng, Zhaorui Tan, Yixin Ji, Penglei Gao

Abstract: Parameter-Efficient Fine-Tuning (PEFT) has become a key strategy for adapting large language models, with recent advances in sparse tuning reducing overhead by selectively updating key parameters or subsets of data. Existing approaches generally focus on two distinct paradigms: layer-selective methods aiming to fine-tune critical layers to minimize computational load, and data-selective methods ai… ▽ More Parameter-Efficient Fine-Tuning (PEFT) has become a key strategy for adapting large language models, with recent advances in sparse tuning reducing overhead by selectively updating key parameters or subsets of data. Existing approaches generally focus on two distinct paradigms: layer-selective methods aiming to fine-tune critical layers to minimize computational load, and data-selective methods aiming to select effective training subsets to boost training. However, current methods typically overlook the fact that different data points contribute varying degrees to distinct model layers, and they often discard potentially valuable information from data perceived as of low quality. To address these limitations, we propose Gradient-aligned Sparse Tuning (GAST), an innovative method that simultaneously performs selective fine-tuning at both data and layer dimensions as integral components of a unified optimization strategy. GAST specifically targets redundancy in information by employing a layer-sparse strategy that adaptively selects the most impactful data points for each layer, providing a more comprehensive and sophisticated solution than approaches restricted to a single dimension. Experiments demonstrate that GAST consistently outperforms baseline methods, establishing a promising direction for future research in PEFT strategies. △ Less

Submitted 10 March, 2026; originally announced March 2026.

MiroFlow: Towards High-Performance and Robust Open-Source Agent Framework for General Deep Research Tasks

Authors: Shiqian Su, Sen Xing, Xuan Dong, Muyan Zhong, Bin Wang, Xizhou Zhu, Yuntao Chen, Wenhai Wang, Yue Deng, Pengxiang Zhu, Ziyuan Liu, Tiantong Li, Jiaheng Yu, Zhe Chen, Lidong Bing, Jifeng Dai

Abstract: Despite the remarkable progress of large language models (LLMs), the capabilities of standalone LLMs have begun to plateau when tackling real-world, complex tasks that require interaction with external tools and dynamic environments. Although recent agent frameworks aim to enhance model autonomy through tool integration and external interaction, they still suffer from naive workflows, unstable per… ▽ More Despite the remarkable progress of large language models (LLMs), the capabilities of standalone LLMs have begun to plateau when tackling real-world, complex tasks that require interaction with external tools and dynamic environments. Although recent agent frameworks aim to enhance model autonomy through tool integration and external interaction, they still suffer from naive workflows, unstable performance, limited support across diverse benchmarks and tasks, and heavy reliance on costly commercial APIs. In this work, we propose a high-performance and robust open-source agent framework, termed MiroFlow, which incorporates an agent graph for flexible orchestration, an optional deep reasoning mode to enhance performance, and a robust workflow execution to ensure stable and reproducible performance. Extensive experiments demonstrate that MiroFlow consistently achieves state-of-the-art performance across multiple agent benchmarks, including GAIA, BrowseComp-EN/ZH, HLE, xBench-DeepSearch, and notably FutureX. We hope it could serve as an easily accessible, reproducible, and comparable baseline for the deep research community. △ Less

Submitted 26 February, 2026; originally announced February 2026.

Beyond Pass-by-Pass Optimization: Intent-Driven IR Optimization with Large Language Models

Authors: Lei Qiu, Zi Yang, Fang Lyu, Ming Zhong, Huimin Cui, Xiaobing Feng

Abstract: Modern compilers optimize programs through a sequence of modular passes over intermediate representations (IR). While this pass-by-pass paradigm offers engineering benefits, it suffers from a pass coordination problem: locally beneficial transformations may block more profitable optimizations in later stages. This limitation stems from the lack of an explicit notion of optimization intent, defined… ▽ More Modern compilers optimize programs through a sequence of modular passes over intermediate representations (IR). While this pass-by-pass paradigm offers engineering benefits, it suffers from a pass coordination problem: locally beneficial transformations may block more profitable optimizations in later stages. This limitation stems from the lack of an explicit notion of optimization intent, defined as a holistic strategy for coordinating multiple transformations toward a global performance objective. Recent LLM-based approaches formulate IR optimization as an end-to-end generation task, thereby avoiding the traditional pass-by-pass structure. However, optimization intent remains implicit in these methods, forcing models to jointly infer optimization strategy and generate low-level transformations, which limits both correctness and performance. We propose IntOpt, the first intent-driven IR optimizer that explicitly separates high-level optimization intent from low-level analysis and transformation. IntOpt organizes IR optimization into three stages: intent formulation, intent refinement, and intent realization, enabling globally coordinated transformations. Experiments show that IntOpt achieves 90.5% verified correctness and 2.660x average speedup on 200-program test set, outperforming state-of-the-art LLM-based optimizers in both correctness and performance, and surpassing modern compiler with the -O3 option on 37 benchmarks with speedups of up to 272.60x. △ Less

Submitted 19 February, 2026; originally announced February 2026.

Beyond Static Snapshots: Dynamic Modeling and Forecasting of Group-Level Value Evolution with Large Language Models

Authors: Qiankun Pi, Guixin Su, Jinliang Li, Mayi Xu, Xin Miao, Jiawei Jiang, Ming Zhong, Tieyun Qian

Abstract: Social simulation is critical for mining complex social dynamics and supporting data-driven decision making. LLM-based methods have emerged as powerful tools for this task by leveraging human-like social questionnaire responses to model group behaviors. Existing LLM-based approaches predominantly focus on group-level values at discrete time points, treating them as static snapshots rather than dyn… ▽ More Social simulation is critical for mining complex social dynamics and supporting data-driven decision making. LLM-based methods have emerged as powerful tools for this task by leveraging human-like social questionnaire responses to model group behaviors. Existing LLM-based approaches predominantly focus on group-level values at discrete time points, treating them as static snapshots rather than dynamic processes. However, group-level values are not fixed but shaped by long-term social changes. Modeling their dynamics is thus crucial for accurate social evolution prediction--a key challenge in both data mining and social science. This problem remains underexplored due to limited longitudinal data, group heterogeneity, and intricate historical event impacts. To bridge this gap, we propose a novel framework for group-level dynamic social simulation by integrating historical value trajectories into LLM-based human response modeling. We select China and the U.S. as representative contexts, conducting stratified simulations across four core sociodemographic dimensions (gender, age, education, income). Using the World Values Survey, we construct a multi-wave, group-level longitudinal dataset to capture historical value evolution, and then propose the first event-based prediction method for this task, unifying social events, current value states, and group attributes into a single framework. Evaluations across five LLM families show substantial gains: a maximum 30.88\% improvement on seen questions and 33.97\% on unseen questions over the Vanilla baseline. We further find notable cross-group heterogeneity: U.S. groups are more volatile than Chinese groups, and younger groups in both countries are more sensitive to external changes. These findings advance LLM-based social simulation and provide new insights for social scientists to understand and predict social value changes. △ Less

Submitted 15 February, 2026; originally announced February 2026.

Steer2Adapt: Dynamically Composing Steering Vectors Elicits Efficient Adaptation of LLMs

Authors: Pengrui Han, Xueqiang Xu, Keyang Xuan, Peiyang Song, Siru Ouyang, Runchu Tian, Yuqing Jiang, Cheng Qian, Pengcheng Jiang, Jiashuo Sun, Junxia Cui, Ming Zhong, Ge Liu, Jiawei Han, Jiaxuan You

Abstract: Activation steering has emerged as a promising approach for efficiently adapting large language models (LLMs) to downstream behaviors. However, most existing steering methods rely on a single static direction per task or concept, making them inflexible under task variation and inadequate for complex tasks that require multiple coordinated capabilities. To address this limitation, we propose STEER2… ▽ More Activation steering has emerged as a promising approach for efficiently adapting large language models (LLMs) to downstream behaviors. However, most existing steering methods rely on a single static direction per task or concept, making them inflexible under task variation and inadequate for complex tasks that require multiple coordinated capabilities. To address this limitation, we propose STEER2ADAPT, a lightweight framework that adapts LLMs by composing steering vectors rather than learning new ones from scratch. In many domains (e.g., reasoning or safety), tasks share a small set of underlying concept dimensions. STEER2ADAPT captures these dimensions as a reusable, low-dimensional semantic prior subspace, and adapts to new tasks by dynamically discovering a linear combination of basis vectors from only a handful of examples. Experiments across 9 tasks and 3 models in both reasoning and safety domains demonstrate the effectiveness of STEER2ADAPT, achieving an average improvement of 8.2%. Extensive analyses further show that STEER2ADAPT is a data-efficient, stable, and transparent inference-time adaptation method for LLMs. △ Less

Submitted 6 February, 2026; originally announced February 2026.

Rethinking the Reranker: Boundary-Aware Evidence Selection for Robust Retrieval-Augmented Generation

Authors: Jiashuo Sun, Pengcheng Jiang, Saizhuo Wang, Jiajun Fan, Heng Wang, Siru Ouyang, Ming Zhong, Yizhu Jiao, Chengsong Huang, Xueqiang Xu, Pengrui Han, Peiran Li, Jiaxin Huang, Ge Liu, Heng Ji, Jiawei Han

Abstract: Retrieval-Augmented Generation (RAG) systems remain brittle under realistic retrieval noise, even when the required evidence appears in the top-K results. A key reason is that retrievers and rerankers optimize solely for relevance, often selecting either trivial, answer-revealing passages or evidence that lacks the critical information required to answer the question, without considering whether t… ▽ More Retrieval-Augmented Generation (RAG) systems remain brittle under realistic retrieval noise, even when the required evidence appears in the top-K results. A key reason is that retrievers and rerankers optimize solely for relevance, often selecting either trivial, answer-revealing passages or evidence that lacks the critical information required to answer the question, without considering whether the evidence is suitable for the generator. We propose BAR-RAG, which reframes the reranker as a boundary-aware evidence selector that targets the generator's Goldilocks Zone -- evidence that is neither trivially easy nor fundamentally unanswerable for the generator, but is challenging yet sufficient for inference and thus provides the strongest learning signal. BAR-RAG trains the selector with reinforcement learning using generator feedback, and adopts a two-stage pipeline that fine-tunes the generator under the induced evidence distribution to mitigate the distribution mismatch between training and inference. Experiments on knowledge-intensive question answering benchmarks show that BAR-RAG consistently improves end-to-end performance under noisy retrieval, achieving an average gain of 10.3 percent over strong RAG and reranking baselines while substantially improving robustness. Code is publicly avaliable at https://github.com/GasolSun36/BAR-RAG. △ Less

Submitted 3 February, 2026; originally announced February 2026.

Comments: 19 pages, 8 tables, 5 figures

Symphony-Coord: Emergent Coordination in Decentralized Agent Systems

Authors: Zhaoyang Guan, Huixi Cao, Ming Zhong, Eric Yang, Lynn Ai, Yongxin Ni, Bill Shi

Abstract: Multi-agent large language model systems can tackle complex multi-step tasks by decomposing work and coordinating specialized behaviors. However, current coordination mechanisms typically rely on statically assigned roles and centralized controllers. As agent pools and task distributions evolve, these design choices lead to inefficient routing, poor adaptability, and fragile fault recovery capabil… ▽ More Multi-agent large language model systems can tackle complex multi-step tasks by decomposing work and coordinating specialized behaviors. However, current coordination mechanisms typically rely on statically assigned roles and centralized controllers. As agent pools and task distributions evolve, these design choices lead to inefficient routing, poor adaptability, and fragile fault recovery capabilities. We introduce Symphony-Coord, a decentralized multi-agent framework that transforms agent selection into an online multi-armed bandit problem, enabling roles to emerge organically through interaction. The framework employs a two-stage dynamic beacon protocol: (i) a lightweight candidate screening mechanism to limit communication and computational overhead; (ii) an adaptive LinUCB selector that routes subtasks based on context features derived from task requirements and agent states, continuously optimized through delayed end-to-end feedback. Under standard linear realizability assumptions, we provide sublinear regret bounds, indicating the system converges toward near-optimal allocation schemes. Validation through simulation experiments and real-world large language model benchmarks demonstrates that Symphony-Coord not only enhances task routing efficiency but also exhibits robust self-healing capabilities in scenarios involving distribution shifts and agent failures, achieving a scalable coordination mechanism without predefined roles. △ Less

Submitted 31 January, 2026; originally announced February 2026.

Comments: 41 pages,15 figures

Self-Compression of Chain-of-Thought via Multi-Agent Reinforcement Learning

Authors: Yiqun Chen, Jinyuan Feng, Wei Yang, Meizhi Zhong, Zhengliang Shi, Rui Li, Xiaochi Wei, Yan Gao, Yi Wu, Yao Hu, Zhiqiang Pu, Jiaxin Mao

Abstract: The inference overhead induced by redundant reasoning undermines the interactive experience and severely bottlenecks the deployment of Large Reasoning Models. Existing reinforcement learning (RL)-based solutions tackle this problem by coupling a length penalty with outcome-based rewards. This simplistic reward weighting struggles to reconcile brevity with accuracy, as enforcing brevity may comprom… ▽ More The inference overhead induced by redundant reasoning undermines the interactive experience and severely bottlenecks the deployment of Large Reasoning Models. Existing reinforcement learning (RL)-based solutions tackle this problem by coupling a length penalty with outcome-based rewards. This simplistic reward weighting struggles to reconcile brevity with accuracy, as enforcing brevity may compromise critical reasoning logic. In this work, we address this limitation by proposing a multi-agent RL framework that selectively penalizes redundant chunks, while preserving essential reasoning logic. Our framework, Self-Compression via MARL (SCMA), instantiates redundancy detection and evaluation through two specialized agents: \textbf{a Segmentation Agent} for decomposing the reasoning process into logical chunks, and \textbf{a Scoring Agent} for quantifying the significance of each chunk. The Segmentation and Scoring agents collaboratively define an importance-weighted length penalty during training, incentivizing \textbf{a Reasoning Agent} to prioritize essential logic without introducing inference overhead during deployment. Empirical evaluations across model scales demonstrate that SCMA reduces response length by 11.1\% to 39.0\% while boosting accuracy by 4.33\% to 10.02\%. Furthermore, ablation studies and qualitative analysis validate that the synergistic optimization within the MARL framework fosters emergent behaviors, yielding more powerful LRMs compared to vanilla RL paradigms. △ Less

Submitted 29 January, 2026; originally announced January 2026.

JADE: Bridging the Strategic-Operational Gap in Dynamic Agentic RAG

Authors: Yiqun Chen, Erhan Zhang, Tianyi Hu, Shijie Wang, Zixuan Yang, Meizhi Zhong, Xiaochi Wei, Yan Gao, Yi Wu, Yao Hu, Jiaxin Mao

Abstract: The evolution of Retrieval-Augmented Generation (RAG) has shifted from static retrieval pipelines to dynamic, agentic workflows where a central planner orchestrates multi-turn reasoning. However, existing paradigms face a critical dichotomy: they either optimize modules jointly within rigid, fixed-graph architectures, or empower dynamic planning while treating executors as frozen, black-box tools.… ▽ More The evolution of Retrieval-Augmented Generation (RAG) has shifted from static retrieval pipelines to dynamic, agentic workflows where a central planner orchestrates multi-turn reasoning. However, existing paradigms face a critical dichotomy: they either optimize modules jointly within rigid, fixed-graph architectures, or empower dynamic planning while treating executors as frozen, black-box tools. We identify that this \textit{decoupled optimization} creates a ``strategic-operational mismatch,'' where sophisticated planning strategies fail to materialize due to unadapted local executors, often leading to negative performance gains despite increased system complexity. In this paper, we propose \textbf{JADE} (\textbf{J}oint \textbf{A}gentic \textbf{D}ynamic \textbf{E}xecution), a unified framework for the joint optimization of planning and execution within dynamic, multi-turn workflows. By modeling the system as a cooperative multi-agent team unified under a single shared backbone, JADE enables end-to-end learning driven by outcome-based rewards. This approach facilitates \textit{co-adaptation}: the planner learns to operate within the capability boundaries of the executors, while the executors evolve to align with high-level strategic intent. Empirical results demonstrate that JADE transforms disjoint modules into a synergistic system, yielding remarkable performance improvements via joint optimization and enabling a flexible balance between efficiency and effectiveness through dynamic workflow orchestration. △ Less

Submitted 29 January, 2026; originally announced January 2026.

TIDE: Tuning-Integrated Dynamic Evolution for LLM-Based Automated Heuristic Design

Authors: Chentong Chen, Mengyuan Zhong, Ye Fan, Jialong Shi, Jianyong Sun

Abstract: Although Large Language Models have advanced Automated Heuristic Design, treating algorithm evolution as a monolithic text generation task overlooks the coupling between discrete algorithmic structures and continuous numerical parameters. Consequently, existing methods often discard promising algorithms due to uncalibrated constants and suffer from premature convergence resulting from simple simil… ▽ More Although Large Language Models have advanced Automated Heuristic Design, treating algorithm evolution as a monolithic text generation task overlooks the coupling between discrete algorithmic structures and continuous numerical parameters. Consequently, existing methods often discard promising algorithms due to uncalibrated constants and suffer from premature convergence resulting from simple similarity metrics. To address these limitations, we propose TIDE, a Tuning-Integrated Dynamic Evolution framework designed to decouple structural reasoning from parameter optimization. TIDE features a nested architecture where an outer parallel island model utilizes Tree Similarity Edit Distance to drive structural diversity, while an inner loop integrates LLM-based logic generation with a differential mutation operator for parameter tuning. Additionally, a UCB-based scheduler dynamically prioritizes high-yield prompt strategies to optimize resource allocation. Extensive experiments across nine combinatorial optimization problems demonstrate that TIDE discovers heuristics that significantly outperform state-of-the-art baselines in solution quality while achieving improved search efficiency and reduced computational costs. △ Less

Submitted 7 February, 2026; v1 submitted 28 January, 2026; originally announced January 2026.

DNF: Dual-Layer Nested Fingerprinting for Large Language Model Intellectual Property Protection

Authors: Zhenhua Xu, Yiran Zhao, Mengting Zhong, Dezhang Kong, Changting Lin, Tong Qiao, Meng Han

Abstract: The rapid growth of large language models raises pressing concerns about intellectual property protection under black-box deployment. Existing backdoor-based fingerprints either rely on rare tokens -- leading to high-perplexity inputs susceptible to filtering -- or use fixed trigger-response mappings that are brittle to leakage and post-hoc adaptation. We propose \textsc{Dual-Layer Nested Fingerpr… ▽ More The rapid growth of large language models raises pressing concerns about intellectual property protection under black-box deployment. Existing backdoor-based fingerprints either rely on rare tokens -- leading to high-perplexity inputs susceptible to filtering -- or use fixed trigger-response mappings that are brittle to leakage and post-hoc adaptation. We propose \textsc{Dual-Layer Nested Fingerprinting} (DNF), a black-box method that embeds a hierarchical backdoor by coupling domain-specific stylistic cues with implicit semantic triggers. Across Mistral-7B, LLaMA-3-8B-Instruct, and Falcon3-7B-Instruct, DNF achieves perfect fingerprint activation while preserving downstream utility. Compared with existing methods, it uses lower-perplexity triggers, remains undetectable under fingerprint detection attacks, and is relatively robust to incremental fine-tuning and model merging. These results position DNF as a practical, stealthy, and resilient solution for LLM ownership verification and intellectual property protection. △ Less

Submitted 21 January, 2026; v1 submitted 13 January, 2026; originally announced January 2026.

Comments: Accepted by ICASSP2026

Calibrating Agent-Based Financial Markets Simulators with Pretrainable Automatic Posterior Transformation-Based Surrogates

Authors: Boquan Jiang, Zhenhua Yang, Chenkai Wang, Muyao Zhong, Heping Fang, Peng Yang

Abstract: Calibrating Agent-Based Models (ABMs) is an important optimization problem for simulating the complex social systems, where the goal is to identify the optimal parameter of a given ABM by minimizing the discrepancy between the simulated data and the real-world observations. Unfortunately, it suffers from the extensive computational costs of iterative evaluations, which involves the expensive simul… ▽ More Calibrating Agent-Based Models (ABMs) is an important optimization problem for simulating the complex social systems, where the goal is to identify the optimal parameter of a given ABM by minimizing the discrepancy between the simulated data and the real-world observations. Unfortunately, it suffers from the extensive computational costs of iterative evaluations, which involves the expensive simulation with the candidate parameter. While Surrogate-Assisted Evolutionary Algorithms (SAEAs) have been widely adopted to alleviate the computational burden, existing methods face two key limitations: 1) surrogating the original evaluation function is hard due the nonlinear yet multi-modal nature of the ABMs, and 2) the commonly used surrogates cannot share the optimization experience among multiple calibration tasks, making the batched calibration less effective. To address these issues, this work proposes Automatic posterior transformation with Negatively Correlated Search and Adaptive Trust-Region (ANTR). ANTR first replaces the traditional surrogates with a pretrainable neural density estimator that directly models the posterior distribution of the parameters given observed data, thereby aligning the optimization objective with parameter-space accuracy. Furthermore, we incorporate a diversity-preserving search strategy to prevent premature convergence and an adaptive trust-region method to efficiently allocate computational resources. We take two representative ABM-based financial market simulators as the test bench as due to the high non-linearity. Experiments demonstrate that the proposed ANTR significantly outperforms conventional metaheuristics and state-of-the-art SAEAs in both calibration accuracy and computational efficiency, particularly in batch calibration scenarios across multiple market conditions. △ Less

Submitted 11 January, 2026; originally announced January 2026.

Comments: 32 pages, 4 figures

Adaptation of Agentic AI: A Survey of Post-Training, Memory, and Skills

Authors: Pengcheng Jiang, Jiacheng Lin, Zhiyi Shi, Zifeng Wang, Luxi He, Yichen Wu, Ming Zhong, Peiyang Song, Qizheng Zhang, Heng Wang, Xueqiang Xu, Hanwen Xu, Pengrui Han, Dylan Zhang, Jiashuo Sun, Chaoqi Yang, Kun Qian, Tian Wang, Changran Hu, Manling Li, Quanzheng Li, Hao Peng, Sheng Wang, Jingbo Shang, Chao Zhang , et al. (9 additional authors not shown)

Abstract: Large language model (LLM) agents are moving beyond prompting alone. ChatGPT marked the rise of general-purpose LLM assistants, DeepSeek showed that on-policy reinforcement learning with verifiable rewards can improve reasoning and tool use, and OpenClaw highlights a newer direction in which agents accumulate persistent memory and reusable skills. Yet the research landscape remains fragmented acro… ▽ More Large language model (LLM) agents are moving beyond prompting alone. ChatGPT marked the rise of general-purpose LLM assistants, DeepSeek showed that on-policy reinforcement learning with verifiable rewards can improve reasoning and tool use, and OpenClaw highlights a newer direction in which agents accumulate persistent memory and reusable skills. Yet the research landscape remains fragmented across post-training, retrieval, memory, and skill systems. This survey studies these developments under a single notion of \emph{adaptation}: improving an agent, its tools, or their interaction after pretraining. We organize the field with a four-paradigm framework spanning agent adaptation and tool adaptation. On the agent side, A1 (tool-execution-signaled) and A2 (agent-output-signaled) improve the agent itself through supervised fine-tuning, preference optimization, and reinforcement learning with verifiable rewards. On the tool side, T1 (agent-agnostic) provides reusable pre-trained modules any agent can call, while T2 (agent-supervised) uses the agent's outputs to train memory systems, skill libraries, or lightweight subagents. Using this framework, we review post-training methods, adaptive memory architectures, and agent skills; compare their trade-offs in cost, flexibility, and generalization; and summarize evaluation practices across deep research, software development, computer use, and drug discovery. We conclude by outlining open problems in agent-tool co-adaptation, continual learning, safety, and efficient deployment. △ Less

Submitted 9 March, 2026; v1 submitted 18 December, 2025; originally announced December 2025.

EgoLCD: Egocentric Video Generation with Long Context Diffusion

Authors: Liuzhou Zhang, Jiarui Ye, Yuanlei Wang, Ming Zhong, Mingju Cao, Wanke Xia, Bowen Zeng, Zeyu Zhang, Hao Tang

Abstract: Generating long, coherent egocentric videos is difficult, as hand-object interactions and procedural tasks require reliable long-term memory. Existing autoregressive models suffer from content drift, where object identity and scene semantics degrade over time. To address this challenge, we introduce EgoLCD, an end-to-end framework for egocentric long-context video generation that treats long video… ▽ More Generating long, coherent egocentric videos is difficult, as hand-object interactions and procedural tasks require reliable long-term memory. Existing autoregressive models suffer from content drift, where object identity and scene semantics degrade over time. To address this challenge, we introduce EgoLCD, an end-to-end framework for egocentric long-context video generation that treats long video synthesis as a problem of efficient and stable memory management. EgoLCD combines a Long-Term Sparse KV Cache for stable global context with an attention-based short-term memory, extended by LoRA for local adaptation. A Memory Regulation Loss enforces consistent memory usage, and Structured Narrative Prompting provides explicit temporal guidance. Extensive experiments on the EgoVid-5M benchmark demonstrate that EgoLCD achieves state-of-the-art performance in both perceptual quality and temporal consistency, effectively mitigating generative forgetting and representing a significant step toward building scalable world models for embodied AI. Code: https://github.com/AIGeeksGroup/EgoLCD. Website: https://aigeeksgroup.github.io/EgoLCD. △ Less

Submitted 4 December, 2025; originally announced December 2025.

Prompt Fairness: Sub-group Disparities in LLMs

Authors: Meiyu Zhong, Noel Teku, Ravi Tandon

Abstract: Large Language Models (LLMs), though shown to be effective in many applications, can vary significantly in their response quality. In this paper, we investigate this problem of prompt fairness: specifically, the phrasing of a prompt by different users/styles, despite the same question being asked in principle, may elicit different responses from an LLM. To quantify this disparity, we propose to us… ▽ More Large Language Models (LLMs), though shown to be effective in many applications, can vary significantly in their response quality. In this paper, we investigate this problem of prompt fairness: specifically, the phrasing of a prompt by different users/styles, despite the same question being asked in principle, may elicit different responses from an LLM. To quantify this disparity, we propose to use information-theoretic metrics that can capture two dimensions of bias: subgroup sensitivity, the variability of responses within a subgroup and cross group consistency, the variability of responses across subgroups. Our analysis reveals that certain subgroups exhibit both higher internal variability and greater divergence from others. Our empirical analysis reveals that certain demographic sub groups experience both higher internal variability and greater divergence from others, indicating structural inequities in model behavior. To mitigate these disparities, we propose practical interventions, including majority voting across multiple generations and prompt neutralization, which together improve response stability and enhance fairness across user populations. In the experiments, we observe clear prompt sensitivity disparities across demographic subgroups: before mitigation, cross-group divergence values reach 0.28 and typically fall in the from 0.14 to 0.22 range. After applying our neutralization and multi generation strategy, these divergences consistently decrease, with the largest gap reduced to 0.22 and many distances falling to 0.17 or below, indicating more stable and consistent outputs across subgroups. △ Less

Submitted 25 November, 2025; originally announced November 2025.

VCU-Bridge: Hierarchical Visual Connotation Understanding via Semantic Bridging

Authors: Ming Zhong, Yuanlei Wang, Liuzhou Zhang, Arctanx An, Renrui Zhang, Hao Liang, Ming Lu, Ying Shen, Wentao Zhang

Abstract: While Multimodal Large Language Models (MLLMs) excel on benchmarks, their processing paradigm differs from the human ability to integrate visual information. Unlike humans who naturally bridge details and high-level concepts, models tend to treat these elements in isolation. Prevailing evaluation protocols often decouple low-level perception from high-level reasoning, overlooking their semantic an… ▽ More While Multimodal Large Language Models (MLLMs) excel on benchmarks, their processing paradigm differs from the human ability to integrate visual information. Unlike humans who naturally bridge details and high-level concepts, models tend to treat these elements in isolation. Prevailing evaluation protocols often decouple low-level perception from high-level reasoning, overlooking their semantic and causal dependencies, which yields non-diagnostic results and obscures performance bottlenecks. We present VCU-Bridge, a framework that operationalizes a human-like hierarchy of visual connotation understanding: multi-level reasoning that advances from foundational perception through semantic bridging to abstract connotation, with an explicit evidence-to-inference trace from concrete cues to abstract conclusions. Building on this framework, we construct HVCU-Bench, a benchmark for hierarchical visual connotation understanding with explicit, level-wise diagnostics. Comprehensive experiments demonstrate a consistent decline in performance as reasoning progresses to higher levels. We further develop a data generation pipeline for instruction tuning guided by Monte Carlo Tree Search (MCTS) and show that strengthening low-level capabilities yields measurable gains at higher levels. Interestingly, it not only improves on HVCU-Bench but also brings benefits on general benchmarks (average +2.53%), especially with substantial gains on MMStar (+7.26%), demonstrating the significance of the hierarchical thinking pattern and its effectiveness in enhancing MLLM capabilities. The project page is at https://vcu-bridge.github.io . △ Less

Submitted 22 November, 2025; originally announced November 2025.

CompEvent: Complex-valued Event-RGB Fusion for Low-light Video Enhancement and Deblurring

Authors: Mingchen Zhong, Xin Lu, Dong Li, Senyan Xu, Ruixuan Jiang, Xueyang Fu, Baocai Yin

Abstract: Low-light video deblurring poses significant challenges in applications like nighttime surveillance and autonomous driving due to dim lighting and long exposures. While event cameras offer potential solutions with superior low-light sensitivity and high temporal resolution, existing fusion methods typically employ staged strategies, limiting their effectiveness against combined low-light and motio… ▽ More Low-light video deblurring poses significant challenges in applications like nighttime surveillance and autonomous driving due to dim lighting and long exposures. While event cameras offer potential solutions with superior low-light sensitivity and high temporal resolution, existing fusion methods typically employ staged strategies, limiting their effectiveness against combined low-light and motion blur degradations. To overcome this, we propose CompEvent, a complex neural network framework enabling holistic full-process fusion of event data and RGB frames for enhanced joint restoration. CompEvent features two core components: 1) Complex Temporal Alignment GRU, which utilizes complex-valued convolutions and processes video and event streams iteratively via GRU to achieve temporal alignment and continuous fusion; and 2) Complex Space-Frequency Learning module, which performs unified complex-valued signal processing in both spatial and frequency domains, facilitating deep fusion through spatial structures and system-level characteristics. By leveraging the holistic representation capability of complex-valued neural networks, CompEvent achieves full-process spatiotemporal fusion, maximizes complementary learning between modalities, and significantly strengthens low-light video deblurring capability. Extensive experiments demonstrate that CompEvent outperforms SOTA methods in addressing this challenging task. △ Less

Submitted 6 February, 2026; v1 submitted 18 November, 2025; originally announced November 2025.

MiroThinker: Pushing the Performance Boundaries of Open-Source Research Agents via Model, Context, and Interactive Scaling

Authors: MiroMind Team, Song Bai, Lidong Bing, Carson Chen, Guanzheng Chen, Yuntao Chen, Zhe Chen, Ziyi Chen, Jifeng Dai, Xuan Dong, Wenhan Dou, Yue Deng, Yunjie Fu, Junqi Ge, Chenxia Han, Tammy Huang, Zhenhang Huang, Jerry Jiao, Shilei Jiang, Tianyu Jiao, Xiaoqi Jian, Lei Lei, Ruilin Li, Ryan Luo, Tiantong Li , et al. (30 additional authors not shown)

Abstract: We present MiroThinker v1.0, an open-source research agent designed to advance tool-augmented reasoning and information-seeking capabilities. Unlike previous agents that only scale up model size or context length, MiroThinker explores interaction scaling at the model level, systematically training the model to handle deeper and more frequent agent-environment interactions as a third dimension of p… ▽ More We present MiroThinker v1.0, an open-source research agent designed to advance tool-augmented reasoning and information-seeking capabilities. Unlike previous agents that only scale up model size or context length, MiroThinker explores interaction scaling at the model level, systematically training the model to handle deeper and more frequent agent-environment interactions as a third dimension of performance improvement. Unlike LLM test-time scaling, which operates in isolation and risks degradation with longer reasoning chains, interactive scaling leverages environment feedback and external information acquisition to correct errors and refine trajectories. Through reinforcement learning, the model achieves efficient interaction scaling: with a 256K context window, it can perform up to 600 tool calls per task, enabling sustained multi-turn reasoning and complex real-world research workflows. Across four representative benchmarks-GAIA, HLE, BrowseComp, and BrowseComp-ZH-the 72B variant achieves up to 81.9%, 37.7%, 47.1%, and 55.6% accuracy respectively, surpassing previous open-source agents and approaching commercial counterparts such as GPT-5-high. Our analysis reveals that MiroThinker benefits from interactive scaling consistently: research performance improves predictably as the model engages in deeper and more frequent agent-environment interactions, demonstrating that interaction depth exhibits scaling behaviors analogous to model size and context length. These findings establish interaction scaling as a third critical dimension for building next-generation open research agents, complementing model capacity and context windows. △ Less

Submitted 18 November, 2025; v1 submitted 14 November, 2025; originally announced November 2025.

Comments: Technical Report

A Compilation Framework for Quantum Circuits with Mid-Circuit Measurement Error Awareness

Authors: Ming Zhong, Zhemin Zhang, Xiangyu Ren, Chenghong Zhu, Siyuan Niu, Zhiding Liang

Abstract: Mid-circuit measurement (MCM) provides the capability for qubit reuse and dynamic control in quantum processors, enabling more resource-efficient algorithms and supporting error-correction procedures. However, MCM introduces several sources of error, including measurement-induced crosstalk, idling-qubit decoherence, and reset infidelity, and these errors exhibit pronounced qubit-dependent variabil… ▽ More Mid-circuit measurement (MCM) provides the capability for qubit reuse and dynamic control in quantum processors, enabling more resource-efficient algorithms and supporting error-correction procedures. However, MCM introduces several sources of error, including measurement-induced crosstalk, idling-qubit decoherence, and reset infidelity, and these errors exhibit pronounced qubit-dependent variability within a single device. Since existing compilers such as the Qiskit-compiler and QR-Map (the state-of-art qubit reuse compiler) do not account for this variability, circuits with frequent MCM operations often experience substantial fidelity loss. In thie paper, we propose MERA, a compilation framework that performs MCM-error-aware layout, routing, and scheduling. MERA leverages lightweight profiling to obtain a stable per-qubit MCM error distribution, which it uses to guide error-aware qubit mapping and SWAP insertions. To further mitigate MCM-related decoherence and crosstalk, MERA augments as-late-as-possible scheduling with context-aware dynamic decoupling. Evaluated on 27 benchmark circuits, MERA achieves 24.94% -- 52.00% fidelity improvement over the Qiskit compiler (optimization level 3) without introducing additional overhead. On QR-Map-generated circuits, it improves fidelity by 29.26% on average and up to 122.58% in the best case, demonstrating its effectiveness for dynamic circuits dominated by MCM operations. △ Less

Submitted 13 November, 2025; originally announced November 2025.

Comments: 8 pages, 7 figures

ACM Class: C.1.3; D.3.4

Can a Small Model Learn to Look Before It Leaps? Dynamic Learning and Proactive Correction for Hallucination Detection

Authors: Zepeng Bao, Shen Zhou, Qiankun Pi, Jianhao Chen, Mayi Xu, Ming Zhong, Yuanyuan Zhu, Tieyun Qian

Abstract: Hallucination in large language models (LLMs) remains a critical barrier to their safe deployment. For hallucination detection to be practical in real-world scenarios, the use of efficient small models is essential to ensure low latency and minimal resource consumption. However, existing methods rely on fixed verification strategies, where simply tuning small models to mimic fixed verification tra… ▽ More Hallucination in large language models (LLMs) remains a critical barrier to their safe deployment. For hallucination detection to be practical in real-world scenarios, the use of efficient small models is essential to ensure low latency and minimal resource consumption. However, existing methods rely on fixed verification strategies, where simply tuning small models to mimic fixed verification trajectories fails to capture the adaptability required for diverse hallucination patterns, thereby inducing planning instability. To address this limitation, we propose a ``Learning to Evaluate and Adaptively Plan'' (LEAP) framework, which shifts hallucination detection from fixed execution to dynamic strategy learning. Specifically, LEAP first employs a powerful teacher model to iteratively explore and refine verification strategies through a failure-driven loop. This dynamic planning capability is then distilled into an efficient student model, augmented by a novel proactive correction mechanism that enables the model to evaluate and optimize its verification strategy before execution. Experiments on three benchmarks demonstrate that LEAP outperforms state-of-the-art methods, offering an effective and scalable solution for reliable hallucination detection. △ Less

Submitted 4 March, 2026; v1 submitted 8 November, 2025; originally announced November 2025.

CoIDO: Efficient Data Selection for Visual Instruction Tuning via Coupled Importance-Diversity Optimization

Authors: Yichen Yan, Ming Zhong, Qi Zhu, Xiaoling Gu, Jinpeng Chen, Huan Li

Abstract: Multimodal large language models (MLLMs) rely heavily on instruction tuning to align vision and language capabilities, yet the computational cost of training on large-scale datasets remains a major bottleneck. Existing data selection methods aim to mitigate this by selecting important and diverse subsets, but they often suffer from two critical drawbacks: high computational overhead from processin… ▽ More Multimodal large language models (MLLMs) rely heavily on instruction tuning to align vision and language capabilities, yet the computational cost of training on large-scale datasets remains a major bottleneck. Existing data selection methods aim to mitigate this by selecting important and diverse subsets, but they often suffer from two critical drawbacks: high computational overhead from processing the entire dataset and suboptimal data selection due to separate treatment of importance and diversity. We introduce CoIDO, a novel dual-objective framework that jointly optimizes data importance and diversity to overcome these challenges. Unlike existing approaches that require costly evaluations across the whole dataset, CoIDO employs a lightweight plug-in scorer. This scorer is trained on just a small random sample of data to learn the distribution of the candidate set, drastically reducing computational demands. By leveraging a homoscedastic uncertainty-based formulation, CoIDO effectively balances importance and diversity during training, enabling efficient and scalable data selection. In our experiments, we trained the CoIDO scorer using only 20 percent of randomly sampled data. Once trained, CoIDO was applied to the entire dataset to select a 20 percent subset for instruction tuning. On the widely used LLaVA-1.5-7B model across ten downstream tasks, this selected subset achieved an impressive 98.2 percent of the performance of full-data fine-tuning, on average. △ Less

Submitted 11 October, 2025; originally announced October 2025.

Comments: 22 pages, 8 figures, 39th Conference on Neural Information Processing Systems (NeurIPS 2025)

Efficient High-Accuracy PDEs Solver with the Linear Attention Neural Operator

Authors: Ming Zhong, Zhenya Yan

Abstract: Neural operators offer a powerful data-driven framework for learning mappings between function spaces, in which the transformer-based neural operator architecture faces a fundamental scalability-accuracy trade-off: softmax attention provides excellent fidelity but incurs quadratic complexity $\mathcal{O}(N^2 d)$ in the number of mesh points $N$ and hidden dimension $d$, while linear attention vari… ▽ More Neural operators offer a powerful data-driven framework for learning mappings between function spaces, in which the transformer-based neural operator architecture faces a fundamental scalability-accuracy trade-off: softmax attention provides excellent fidelity but incurs quadratic complexity $\mathcal{O}(N^2 d)$ in the number of mesh points $N$ and hidden dimension $d$, while linear attention variants reduce cost to $\mathcal{O}(N d^2)$ but often suffer significant accuracy degradation. To address the aforementioned challenge, in this paper, we present a novel type of neural operators, Linear Attention Neural Operator (LANO), which achieves both scalability and high accuracy by reformulating attention through an agent-based mechanism. LANO resolves this dilemma by introducing a compact set of $M$ agent tokens $(M \ll N)$ that mediate global interactions among $N$ tokens. This agent attention mechanism yields an operator layer with linear complexity $\mathcal{O}(MN d)$ while preserving the expressive power of softmax attention. Theoretically, we demonstrate the universal approximation property, thereby demonstrating improved conditioning and stability properties. Empirically, LANO surpasses current state-of-the-art neural PDE solvers, including Transolver with slice-based softmax attention, achieving average $19.5\%$ accuracy improvement across standard benchmarks. By bridging the gap between linear complexity and softmax-level performance, LANO establishes a scalable, high-accuracy foundation for scientific machine learning applications. △ Less

Submitted 19 October, 2025; originally announced October 2025.

Comments: 31 pages, 8 figures

TaskAudit: Detecting Functiona11ity Errors in Mobile Apps via Agentic Task Execution

Authors: Mingyuan Zhong, Xia Chen, Davin Win Kyi, Chen Li, James Fogarty, Jacob O. Wobbrock

Abstract: Accessibility checkers are tools in support of accessible app development, and their use is encouraged by accessibility best practices. However, most current checkers evaluate static or mechanically-generated contexts, failing to capture common accessibility errors impacting mobile app functionality. In this work, we define functiona11ity errors as accessibility barriers that only manifest through… ▽ More Accessibility checkers are tools in support of accessible app development, and their use is encouraged by accessibility best practices. However, most current checkers evaluate static or mechanically-generated contexts, failing to capture common accessibility errors impacting mobile app functionality. In this work, we define functiona11ity errors as accessibility barriers that only manifest through interaction (i.e., named according to a blend of "functionality" and "accessibility"). We introduce TaskAudit, which comprises three components: a Task Generator that constructs interactive tasks from app screens, a Task Executor that uses agents with a screen reader proxy to perform these tasks, and an Accessibility Analyzer that detects and reports accessibility errors by examining interaction traces. Our evaluation on real-world apps shows that TaskAudit detects 48 functiona11ity errors from 54 app screens, compared to between 4 and 20 with existing checkers. Our analysis demonstrates common error patterns that TaskAudit can detect in addition to those from prior work, including label-functionality mismatch, cluttered navigation, and inappropriate feedback. △ Less

Submitted 4 February, 2026; v1 submitted 14 October, 2025; originally announced October 2025.

Comments: CHI 2026

ACM Class: H.5.2

SusBench: An Online Benchmark for Evaluating Dark Pattern Susceptibility of Computer-Use Agents

Authors: Longjie Guo, Chenjie Yuan, Mingyuan Zhong, Robert Wolfe, Ruican Zhong, Yue Xu, Bingbing Wen, Hua Shen, Lucy Lu Wang, Alexis Hiniker

Abstract: As LLM-based computer-use agents (CUAs) begin to autonomously interact with real-world interfaces, understanding their vulnerability to manipulative interface designs becomes increasingly critical. We introduce SusBench, an online benchmark for evaluating the susceptibility of CUAs to UI dark patterns, designs that aim to manipulate or deceive users into taking unintentional actions. Drawing nine… ▽ More As LLM-based computer-use agents (CUAs) begin to autonomously interact with real-world interfaces, understanding their vulnerability to manipulative interface designs becomes increasingly critical. We introduce SusBench, an online benchmark for evaluating the susceptibility of CUAs to UI dark patterns, designs that aim to manipulate or deceive users into taking unintentional actions. Drawing nine common dark pattern types from existing taxonomies, we developed a method for constructing believable dark patterns on real-world consumer websites through code injections, and designed 313 evaluation tasks across 55 websites. Our study with 29 participants showed that humans perceived our dark pattern injections to be highly realistic, with the vast majority of participants not noticing that these had been injected by the research team. We evaluated five state-of-the-art CUAs on the benchmark. We found that both human participants and agents are particularly susceptible to the dark patterns of Preselection, Trick Wording, and Hidden Information, while being resilient to other overt dark patterns. Our findings inform the development of more trustworthy CUAs, their use as potential human proxies in evaluating deceptive designs, and the regulation of an online environment increasingly navigated by autonomous agents. △ Less

Submitted 23 February, 2026; v1 submitted 13 October, 2025; originally announced October 2025.

Comments: Accepted as a full paper to IUI 2026

Conformal Sparsification for Bandwidth-Efficient Edge-Cloud Speculative Decoding

Authors: Payel Bhattacharjee, Fengwei Tian, Meiyu Zhong, Guangyi Zhang, Osvaldo Simeone, Ravi Tandon

Abstract: Edge-cloud speculative decoding (SD) accelerates inference by having a cloud-based large language model (LLM) that verifies draft tokens generated by a resource-constrained small language model (SLM) at the edge. A central bottleneck is the limited bandwidth of the edge-cloud link, which necessitates efficient compression of draft token distributions. We first derive an information-theoretic bound… ▽ More Edge-cloud speculative decoding (SD) accelerates inference by having a cloud-based large language model (LLM) that verifies draft tokens generated by a resource-constrained small language model (SLM) at the edge. A central bottleneck is the limited bandwidth of the edge-cloud link, which necessitates efficient compression of draft token distributions. We first derive an information-theoretic bound that decomposes the token rejection rate into contributions from SLM-LLM distribution mismatch and from quantization distortion. Guided by this analysis, we propose the Sparse Quantize-and-Sample SD (SQS-SD) framework, which exploits distributional sparsity through structured sparsification and lattice-based quantization. Within this framework, K-SQS applies fixed top-K truncation, while C-SQS adaptively adjusts the retained token set via online conformal prediction to ensure bounded deviation from the dense distribution. Empirical results confirm that both approaches improve end-to-end latency and rejection rates in complimentary operating regimes. △ Less

Submitted 10 October, 2025; originally announced October 2025.

Comments: 39th Conference on Neural Information Processing Systems (NeurIPS 2025) Workshop: AI and ML for Next-Generation Wireless Communications and Networking (AI4NextG)

Enhancing Self-Supervised Learning with Semantic Pairs A New Dataset and Empirical Study

Authors: Mohammad Alkhalefi, Georgios Leontidis, Mingjun Zhong

Abstract: Instance discrimination is a self-supervised representation learning paradigm wherein individual instances within a dataset are treated as distinct classes. This is typically achieved by generating two disparate views of each instance by applying stochastic transformations, encouraging the model to learn representations invariant to the common underlying object across these views. While this appro… ▽ More Instance discrimination is a self-supervised representation learning paradigm wherein individual instances within a dataset are treated as distinct classes. This is typically achieved by generating two disparate views of each instance by applying stochastic transformations, encouraging the model to learn representations invariant to the common underlying object across these views. While this approach facilitates the acquisition of invariant representations for dataset instances under various handcrafted transformations (e.g., random cropping, colour jittering), an exclusive reliance on such data transformations for achieving invariance may inherently limit the model's generalizability to unseen datasets and diverse downstream tasks. The inherent limitation stems from the fact that the finite set of transformations within the data processing pipeline is unable to encompass the full spectrum of potential data variations. In this study, we provide the technical foundation for leveraging semantic pairs to enhance the generalizability of the model's representation and empirically demonstrate that incorporating semantic pairs mitigates the issue of limited transformation coverage. Specifically, we propose that by exposing the model to semantic pairs (i.e., two instances belonging to the same semantic category), we introduce varied real-world scene contexts, thereby fostering the development of more generalizable object representations. To validate this hypothesis, we constructed and released a novel dataset comprising curated semantic pairs and conducted extensive experimentation to empirically establish that their inclusion enables the model to learn more general representations, ultimately leading to improved performance across diverse downstream tasks. △ Less

Submitted 13 October, 2025; v1 submitted 9 October, 2025; originally announced October 2025.

Comments: 16 pages, 7 figures, 5 tables

Vibe Checker: Aligning Code Evaluation with Human Preference

Authors: Ming Zhong, Xiang Zhou, Ting-Yun Chang, Qingze Wang, Nan Xu, Xiance Si, Dan Garrette, Shyam Upadhyay, Jeremiah Liu, Jiawei Han, Benoit Schillings, Jiao Sun

Abstract: Large Language Models (LLMs) have catalyzed vibe coding, where users leverage LLMs to generate and iteratively refine code through natural language interactions until it passes their vibe check. Vibe check is tied to real-world human preference and goes beyond functionality: the solution should feel right, read cleanly, preserve intent, and remain correct. However, current code evaluation remains… ▽ More Large Language Models (LLMs) have catalyzed vibe coding, where users leverage LLMs to generate and iteratively refine code through natural language interactions until it passes their vibe check. Vibe check is tied to real-world human preference and goes beyond functionality: the solution should feel right, read cleanly, preserve intent, and remain correct. However, current code evaluation remains anchored to pass@k and captures only functional correctness, overlooking the non-functional instructions that users routinely apply. In this paper, we hypothesize that instruction following is the missing piece underlying vibe check that represents human preference in coding besides functional correctness. To quantify models' code instruction following capabilities with measurable signals, we present VeriCode, a taxonomy of 30 verifiable code instructions together with corresponding deterministic verifiers. We use the taxonomy to augment established evaluation suites, resulting in Vibe Checker, a testbed to assess both code instruction following and functional correctness. Upon evaluating 31 leading LLMs, we show that even the strongest models struggle to comply with multiple instructions and exhibit clear functional regression. Most importantly, a composite score of functional correctness and instruction following correlates the best with human preference, with the latter emerging as the primary differentiator on real-world programming tasks. Our work identifies core factors of the vibe check, providing a concrete path for benchmarking and developing models that better align with user preferences in coding. △ Less

Submitted 8 October, 2025; originally announced October 2025.

Comments: Preprint

DecEx-RAG: Boosting Agentic Retrieval-Augmented Generation with Decision and Execution Optimization via Process Supervision

Authors: Yongqi Leng, Yikun Lei, Xikai Liu, Meizhi Zhong, Bojian Xiong, Yurong Zhang, Yan Gao, Yi Wu, Yao Hu, Deyi Xiong

Abstract: Agentic Retrieval-Augmented Generation (Agentic RAG) enhances the processing capability for complex tasks through dynamic retrieval and adaptive workflows. Recent advances (e.g., Search-R1) have shown that outcome-supervised reinforcement learning demonstrate strong performance. However, this approach still suffers from inefficient exploration, sparse reward signals, and ambiguous global reward fe… ▽ More Agentic Retrieval-Augmented Generation (Agentic RAG) enhances the processing capability for complex tasks through dynamic retrieval and adaptive workflows. Recent advances (e.g., Search-R1) have shown that outcome-supervised reinforcement learning demonstrate strong performance. However, this approach still suffers from inefficient exploration, sparse reward signals, and ambiguous global reward feedback. To address these challenges, we propose DecEx-RAG, which models RAG as a Markov Decision Process (MDP) incorporating decision-making and execution, while introducing an efficient pruning strategy to optimize data expansion. Through comprehensive process-level policy optimization, DecEx-RAG significantly enhances the autonomous task decomposition, dynamic retrieval, and high-quality answer generation capabilities of large language models (LLMs). Experiments show that DecEx-RAG achieves an average absolute performance improvement of $6.2\%$ across six datasets, significantly outperforming existing baselines. Moreover, the pruning strategy improves data construction efficiency by nearly $6 \times$, providing an efficient solution for process-supervised RAG training. The code is available at https://github.com/sdsxdxl/DecEx-RAG. △ Less

Submitted 7 October, 2025; originally announced October 2025.

Spiral of Silence in Large Language Model Agents

Authors: Mingze Zhong, Meng Fang, Zijing Shi, Yuxuan Huang, Shunfeng Zheng, Yali Du, Ling Chen, Jun Wang

Abstract: The Spiral of Silence (SoS) theory holds that individuals with minority views often refrain from speaking out for fear of social isolation, enabling majority positions to dominate public discourse. When the 'agents' are large language models (LLMs), however, the classical psychological explanation is not directly applicable, since SoS was developed for human societies. This raises a central questi… ▽ More The Spiral of Silence (SoS) theory holds that individuals with minority views often refrain from speaking out for fear of social isolation, enabling majority positions to dominate public discourse. When the 'agents' are large language models (LLMs), however, the classical psychological explanation is not directly applicable, since SoS was developed for human societies. This raises a central question: can SoS-like dynamics nevertheless emerge from purely statistical language generation in LLM collectives? We propose an evaluation framework for examining SoS in LLM agents. Specifically, we consider four controlled conditions that systematically vary the availability of 'History' and 'Persona' signals. Opinion dynamics are assessed using trend tests such as Mann-Kendall and Spearman's rank, along with concentration measures including kurtosis and interquartile range. Experiments across open-source and closed-source models show that history and persona together produce strong majority dominance and replicate SoS patterns; history signals alone induce strong anchoring; and persona signals alone foster diverse but uncorrelated opinions, indicating that without historical anchoring, SoS dynamics cannot emerge. The work bridges computational sociology and responsible AI design, highlighting the need to monitor and mitigate emergent conformity in LLM-agent systems. △ Less

Submitted 7 October, 2025; v1 submitted 28 September, 2025; originally announced October 2025.

Comments: Accepted to EMNLP 2025 (Findings)

Bayesian Distributional Models of Executive Functioning

Authors: Robert Kasumba, Zeyu Lu, Dom CP Marticorena, Mingyang Zhong, Paul Beggs, Anja Pahor, Geetha Ramani, Imani Goffney, Susanne M Jaeggi, Aaron R Seitz, Jacob R Gardner, Dennis L Barbour

Abstract: This study uses controlled simulations with known ground-truth parameters to evaluate how Distributional Latent Variable Models (DLVM) and Bayesian Distributional Active LEarning (DALE) perform in comparison to conventional Independent Maximum Likelihood Estimation (IMLE). DLVM integrates observations across multiple executive function tasks and individuals, allowing parameter estimation even unde… ▽ More This study uses controlled simulations with known ground-truth parameters to evaluate how Distributional Latent Variable Models (DLVM) and Bayesian Distributional Active LEarning (DALE) perform in comparison to conventional Independent Maximum Likelihood Estimation (IMLE). DLVM integrates observations across multiple executive function tasks and individuals, allowing parameter estimation even under sparse or incomplete data conditions. DLVM consistently outperformed IMLE, especially under with smaller amounts of data, and converges faster to highly accurate estimates of the true distributions. In a second set of analyses, DALE adaptively guided sampling to maximize information gain, outperforming random sampling and fixed test batteries, particularly within the first 80 trials. These findings establish the advantages of combining DLVM's cross-task inference with DALE's optimal adaptive sampling, providing a principled basis for more efficient cognitive assessments. △ Less

Submitted 7 October, 2025; v1 submitted 30 September, 2025; originally announced October 2025.

Comments: 42 pages, 8 figures, 1 table

Robustness of One-to-Many Interdependent Higher-order Networks Against Cascading Failures

Authors: Cheng Qian, Dandan Zhao, Bo Zhang, Ming Zhong, Jianmin Han, Shenghong Li, Hao Peng, Wei Wang

Abstract: In the real world, the stable operation of a network is usually inseparable from the mutual support of other networks. In such an interdependent network, a node in one layer may depend on multiple nodes in another layer, forming a complex one-to-many dependency relationship. Meanwhile, there may also be higher-order interactions between multiple nodes within a layer, which increases the connectivi… ▽ More In the real world, the stable operation of a network is usually inseparable from the mutual support of other networks. In such an interdependent network, a node in one layer may depend on multiple nodes in another layer, forming a complex one-to-many dependency relationship. Meanwhile, there may also be higher-order interactions between multiple nodes within a layer, which increases the connectivity within the layer. However, existing research on one-to-many interdependence often neglects intra-layer higher-order structures and lacks a unified theoretical framework for inter-layer dependencies. Moreover, current research on interdependent higher-order networks typically assumes idealized one-to-one inter-layer dependencies, which does not reflect the complexity of real-world systems. These limitations hinder a comprehensive understanding of how such networks withstand failures. Therefore, this paper investigates the robustness of one-to-many interdependent higher-order networks under random attacks. Depending on whether node survival requires at least one dependency edge or multiple dependency edges, we propose four inter-layer interdependency conditions and analyze the network's robustness after cascading failures induced by random attacks. Using percolation theory, we establish a unified theoretical framework that reveals how higher-order interaction structures within intra-layers and inter-layer coupling parameters affect network reliability and system resilience. Additionally, we extend our study to partially interdependent hypergraphs. We validate our theoretical analysis on both synthetic and real-data-based interdependent hypergraphs, offering insights into the optimization of network design for enhanced reliability. △ Less

Submitted 28 September, 2025; originally announced September 2025.

HICode: Hierarchical Inductive Coding with LLMs

Authors: Mian Zhong, Pristina Wang, Anjalie Field

Abstract: Despite numerous applications for fine-grained corpus analysis, researchers continue to rely on manual labeling, which does not scale, or statistical tools like topic modeling, which are difficult to control. We propose that LLMs have the potential to scale the nuanced analyses that researchers typically conduct manually to large text corpora. To this effect, inspired by qualitative research metho… ▽ More Despite numerous applications for fine-grained corpus analysis, researchers continue to rely on manual labeling, which does not scale, or statistical tools like topic modeling, which are difficult to control. We propose that LLMs have the potential to scale the nuanced analyses that researchers typically conduct manually to large text corpora. To this effect, inspired by qualitative research methods, we develop HICode, a two-part pipeline that first inductively generates labels directly from analysis data and then hierarchically clusters them to surface emergent themes. We validate this approach across three diverse datasets by measuring alignment with human-constructed themes and demonstrating its robustness through automated and human evaluations. Finally, we conduct a case study of litigation documents related to the ongoing opioid crisis in the U.S., revealing aggressive marketing strategies employed by pharmaceutical companies and demonstrating HICode's potential for facilitating nuanced analyses in large-scale data. △ Less

Submitted 22 September, 2025; originally announced September 2025.

Comments: Long paper accepted at EMNLP 2025 main conference, 19 pages, 8 figures

A Survey on Retrieval And Structuring Augmented Generation with Large Language Models

Authors: Pengcheng Jiang, Siru Ouyang, Yizhu Jiao, Ming Zhong, Runchu Tian, Jiawei Han

Abstract: Large Language Models (LLMs) have revolutionized natural language processing with their remarkable capabilities in text generation and reasoning. However, these models face critical challenges when deployed in real-world applications, including hallucination generation, outdated knowledge, and limited domain expertise. Retrieval And Structuring (RAS) Augmented Generation addresses these limitation… ▽ More Large Language Models (LLMs) have revolutionized natural language processing with their remarkable capabilities in text generation and reasoning. However, these models face critical challenges when deployed in real-world applications, including hallucination generation, outdated knowledge, and limited domain expertise. Retrieval And Structuring (RAS) Augmented Generation addresses these limitations by integrating dynamic information retrieval with structured knowledge representations. This survey (1) examines retrieval mechanisms including sparse, dense, and hybrid approaches for accessing external knowledge; (2) explore text structuring techniques such as taxonomy construction, hierarchical classification, and information extraction that transform unstructured text into organized representations; and (3) investigate how these structured representations integrate with LLMs through prompt-based methods, reasoning frameworks, and knowledge embedding techniques. It also identifies technical challenges in retrieval efficiency, structure quality, and knowledge integration, while highlighting research opportunities in multimodal retrieval, cross-lingual structures, and interactive systems. This comprehensive overview provides researchers and practitioners with insights into RAS methods, applications, and future directions. △ Less

Submitted 12 September, 2025; originally announced September 2025.

Comments: KDD'25 survey track

The Gold Medals in an Empty Room: Diagnosing Metalinguistic Reasoning in LLMs with Camlang

Authors: Fenghua Liu, Yulong Chen, Yixuan Liu, Zhujun Jin, Solomon Tsai, Ming Zhong

Abstract: Large Language Models (LLMs) achieve gold-medal performance across many benchmarks, yet it remains unclear whether such success reflects genuine reasoning or pattern matching. From a cognitive science perspective, an informative test is whether models can master an unfamiliar language through explicit metalinguistic deductive learning, a paradigm where human learners can reliably internalise gramm… ▽ More Large Language Models (LLMs) achieve gold-medal performance across many benchmarks, yet it remains unclear whether such success reflects genuine reasoning or pattern matching. From a cognitive science perspective, an informative test is whether models can master an unfamiliar language through explicit metalinguistic deductive learning, a paradigm where human learners can reliably internalise grammatical systems through metalinguistic reasoning. We address this question with Camlang, a novel constructed language that exhibits naturalistic yet unattested feature combinations. Camlang consists of two explicit resources, a grammar book and a bilingual dictionary, which mirror adult second-language learning via explicit grammar rules and lexical lookup, and enable us to disentangle errors in morpho-syntax, lexical semantics, and sentence-level reasoning. Human experiments show that these resources are sufficient for participants to acquire Camlang and successfully solve Camlang tasks. To operationalise evaluation, we adapt CommonsenseQA into Camlang, creating Camlang-CSQA-v0, the first task in a broader suite where solving questions requires applying grammar rules and lexical mappings. Experimental results show that GPT-5 achieves 98\% EM accuracy in English but only 47\% in Camlang, far below human performance at 87\%, while other state-of-the-art reasoning LLMs perform even worse. Human verification further reveals that most model successes stem from shallow lexical alignment while GPT-5 shows emerging metalinguistic awareness to a limited extent but not systematic grammatical mastery as humans. Camlang establishes a cognitively grounded evaluation paradigm that exposes fundamental gaps between current models and human metalinguistic competence. △ Less

Submitted 30 August, 2025; originally announced September 2025.

Comments: Working in progress

Uncertain data assimilation for urban wind flow simulations with OpenLB-UQ

Authors: Mingliang Zhong, Dennis Teutscher, Adrian Kummerländer, Mathias J. Krause, Martin Frank, Stephan Simonis

Abstract: Accurate prediction of urban wind flow is essential for urban planning, pedestrian safety, and environmental management. Yet, it remains challenging due to uncertain boundary conditions and the high cost of conventional CFD simulations. This paper presents the use of the modular and efficient uncertainty quantification (UQ) framework OpenLB-UQ for urban wind flow simulations. We specifically use t… ▽ More Accurate prediction of urban wind flow is essential for urban planning, pedestrian safety, and environmental management. Yet, it remains challenging due to uncertain boundary conditions and the high cost of conventional CFD simulations. This paper presents the use of the modular and efficient uncertainty quantification (UQ) framework OpenLB-UQ for urban wind flow simulations. We specifically use the lattice Boltzmann method (LBM) coupled with a stochastic collocation (SC) approach based on generalized polynomial chaos (gPC). The framework introduces a relative-error noise model for inflow wind speeds based on real measurements. The model is propagated through a non-intrusive SC LBM pipeline using sparse-grid quadrature. Key quantities of interest, including mean flow fields, standard deviations, and vertical profiles with confidence intervals, are efficiently computed without altering the underlying deterministic solver. We demonstrate this on a real urban scenario, highlighting how uncertainty localizes in complex flow regions such as wakes and shear layers. The results show that the SC LBM approach provides accurate, uncertainty-aware predictions with significant computational efficiency, making OpenLB-UQ a practical tool for real-time urban wind analysis. △ Less

Submitted 25 August, 2025; originally announced August 2025.

Format as a Prior: Quantifying and Analyzing Bias in LLMs for Heterogeneous Data

Authors: Jiacheng Liu, Mayi Xu, Qiankun Pi, Wenli Li, Ming Zhong, Yuanyuan Zhu, Mengchi Liu, Tieyun Qian

Abstract: Large Language Models (LLMs) are increasingly employed in applications that require processing information from heterogeneous formats, including texts, tables, infoboxes, and knowledge graphs. However, systematic biases toward particular formats may undermine LLMs' ability to integrate heterogeneous data impartially, potentially resulting in reasoning errors and increased risks in downstream tasks… ▽ More Large Language Models (LLMs) are increasingly employed in applications that require processing information from heterogeneous formats, including texts, tables, infoboxes, and knowledge graphs. However, systematic biases toward particular formats may undermine LLMs' ability to integrate heterogeneous data impartially, potentially resulting in reasoning errors and increased risks in downstream tasks. Yet it remains unclear whether such biases are systematic, which data-level factors drive them, and what internal mechanisms underlie their emergence. In this paper, we present the first comprehensive study of format bias in LLMs through a three-stage empirical analysis. The first stage explores the presence and direction of bias across a diverse range of LLMs. The second stage examines how key data-level factors influence these biases. The third stage analyzes how format bias emerges within LLMs' attention patterns and evaluates a lightweight intervention to test its effectiveness. Our results show that format bias is consistent across model families, driven by information richness, structure quality, and representation type, and is closely associated with attention imbalance within the LLMs. Based on these investigations, we identify three future research directions to reduce format bias: enhancing data pre-processing through format repair and normalization, introducing inference-time interventions such as attention re-weighting, and developing format-balanced training corpora. These directions will support the design of more robust and fair heterogeneous data processing systems. △ Less

Submitted 13 January, 2026; v1 submitted 12 August, 2025; originally announced August 2025.

Comments: Accepted by AAAI 2026, camera ready version

OpenLB-UQ: An Uncertainty Quantification Framework for Incompressible Fluid Flow Simulations

Authors: Mingliang Zhong, Adrian Kummerländer, Shota Ito, Mathias J. Krause, Martin Frank, Stephan Simonis

Abstract: Uncertainty quantification (UQ) is crucial in computational fluid dynamics to assess the reliability and robustness of simulations, given the uncertainties in input parameters. OpenLB is an open-source lattice Boltzmann method library designed for efficient and extensible simulations of complex fluid dynamics on high-performance computers. In this work, we leverage the efficiency of OpenLB for lar… ▽ More Uncertainty quantification (UQ) is crucial in computational fluid dynamics to assess the reliability and robustness of simulations, given the uncertainties in input parameters. OpenLB is an open-source lattice Boltzmann method library designed for efficient and extensible simulations of complex fluid dynamics on high-performance computers. In this work, we leverage the efficiency of OpenLB for large-scale flow sampling with a dedicated and integrated UQ module. To this end, we focus on non-intrusive stochastic collocation methods based on generalized polynomial chaos and Monte Carlo sampling. The OpenLB-UQ framework is extensively validated in convergence tests with respect to statistical metrics and sample efficiency using selected benchmark cases, including two-dimensional Taylor--Green vortex flows with up to four-dimensional uncertainty and a flow past a cylinder. Our results confirm the expected convergence rates and show promising scalability, demonstrating robust statistical accuracy as well as computational efficiency. OpenLB-UQ enhances the capability of the OpenLB library, offering researchers a scalable framework for UQ in incompressible fluid flow simulations and beyond. △ Less

Submitted 19 August, 2025; originally announced August 2025.

HiFo-Prompt: Prompting with Hindsight and Foresight for LLM-based Automatic Heuristic Design

Authors: Chentong Chen, Mengyuan Zhong, Ye Fan, Jialong Shi, Jianyong Sun

Abstract: LLM-based Automatic Heuristic Design (AHD) within Evolutionary Computation (EC) frameworks has shown promising results. However, its effectiveness is hindered by the use of static operators and the lack of knowledge accumulation mechanisms. We introduce HiFo-Prompt, a framework that guides LLMs with two synergistic prompting strategies: Foresight and Hindsight. Foresight-based prompts adaptively s… ▽ More LLM-based Automatic Heuristic Design (AHD) within Evolutionary Computation (EC) frameworks has shown promising results. However, its effectiveness is hindered by the use of static operators and the lack of knowledge accumulation mechanisms. We introduce HiFo-Prompt, a framework that guides LLMs with two synergistic prompting strategies: Foresight and Hindsight. Foresight-based prompts adaptively steer the search based on population dynamics, managing the exploration-exploitation trade-off. In addition, hindsight-based prompts mimic human expertise by distilling successful heuristics from past generations into fundamental, reusable design principles. This dual mechanism transforms transient discoveries into a persistent knowledge base, enabling the LLM to learn from its own experience. Empirical results demonstrate that HiFo-Prompt significantly outperforms state-of-the-art LLM-based AHD methods, generating higher-quality heuristics while achieving substantially faster convergence and superior query efficiency. Our code is available at https://github.com/Challenger-XJTU/HiFo-Prompt. △ Less

Submitted 8 February, 2026; v1 submitted 18 August, 2025; originally announced August 2025.

Comments: Accepted at ICLR 2026

NeuronTune: Fine-Grained Neuron Modulation for Balanced Safety-Utility Alignment in LLMs

Authors: Birong Pan, Mayi Xu, Qiankun Pi, Jianhao Chen, Yuanyuan Zhu, Ming Zhong, Tieyun Qian

Abstract: Ensuring robust safety alignment while preserving utility is critical for the reliable deployment of Large Language Models (LLMs). However, current techniques fundamentally suffer from intertwined deficiencies: insufficient robustness against malicious attacks, frequent refusal of benign queries, degradation in generated text quality and general task performance--the former two reflecting deficits… ▽ More Ensuring robust safety alignment while preserving utility is critical for the reliable deployment of Large Language Models (LLMs). However, current techniques fundamentally suffer from intertwined deficiencies: insufficient robustness against malicious attacks, frequent refusal of benign queries, degradation in generated text quality and general task performance--the former two reflecting deficits in robust safety and the latter constituting utility impairment. We trace these limitations to the coarse-grained layer-wise interventions in existing methods. To resolve this, we propose NeuronTune, a fine-grained framework that dynamically modulates sparse neurons to achieve simultaneous safety-utility optimization. Our approach first identifies safety-critical and utility-preserving neurons across all layers via attribution, then employs meta-learning to adaptively amplify safety-neuron activations and suppress utility-neuron activations. Crucially, NeuronTune enables tunable adjustment of intervention scope via neuron-count thresholds, supporting flexible adaptation to security-critical or utility-priority scenarios. Extensive experimental results demonstrate that our method significantly outperforms existing state-of-the-art technologies, achieving superior model safety while maintaining excellent utility. △ Less

Submitted 13 August, 2025; originally announced August 2025.

A Survey on Training-free Alignment of Large Language Models

Authors: Birong Pan, Yongqi Li, Weiyu Zhang, Wenpeng Lu, Mayi Xu, Shen Zhou, Yuanyuan Zhu, Ming Zhong, Tieyun Qian

Abstract: The alignment of large language models (LLMs) aims to ensure their outputs adhere to human values, ethical standards, and legal norms. Traditional alignment methods often rely on resource-intensive fine-tuning (FT), which may suffer from knowledge degradation and face challenges in scenarios where the model accessibility or computational resources are constrained. In contrast, training-free (TF) a… ▽ More The alignment of large language models (LLMs) aims to ensure their outputs adhere to human values, ethical standards, and legal norms. Traditional alignment methods often rely on resource-intensive fine-tuning (FT), which may suffer from knowledge degradation and face challenges in scenarios where the model accessibility or computational resources are constrained. In contrast, training-free (TF) alignment techniques--leveraging in-context learning, decoding-time adjustments, and post-generation corrections--offer a promising alternative by enabling alignment without heavily retraining LLMs, making them adaptable to both open-source and closed-source environments. This paper presents the first systematic review of TF alignment methods, categorizing them by stages of pre-decoding, in-decoding, and post-decoding. For each stage, we provide a detailed examination from the viewpoint of LLMs and multimodal LLMs (MLLMs), highlighting their mechanisms and limitations. Furthermore, we identify key challenges and future directions, paving the way for more inclusive and effective TF alignment techniques. By synthesizing and organizing the rapidly growing body of research, this survey offers a guidance for practitioners and advances the development of safer and more reliable LLMs. △ Less

Submitted 10 September, 2025; v1 submitted 12 August, 2025; originally announced August 2025.

Comments: Accepted to EMNLP 2025 (findings), camera-ready version

Privacy-protected Retrieval-Augmented Generation for Knowledge Graph Question Answering

Authors: Yunfeng Ning, Mayi Xu, Jintao Wen, Qiankun Pi, Yuanyuan Zhu, Ming Zhong, Jiawei Jiang, Tieyun Qian

Abstract: LLMs often suffer from hallucinations and outdated or incomplete knowledge. RAG is proposed to address these issues by integrating external knowledge like that in KGs into LLMs. However, leveraging private KGs in RAG systems poses significant privacy risks due to the black-box nature of LLMs and potential insecure data transmission, especially when using third-party LLM APIs lacking transparency a… ▽ More LLMs often suffer from hallucinations and outdated or incomplete knowledge. RAG is proposed to address these issues by integrating external knowledge like that in KGs into LLMs. However, leveraging private KGs in RAG systems poses significant privacy risks due to the black-box nature of LLMs and potential insecure data transmission, especially when using third-party LLM APIs lacking transparency and control. In this paper, we investigate the privacy-protected RAG scenario for the first time, where entities in KGs are anonymous for LLMs, thus preventing them from accessing entity semantics. Due to the loss of semantics of entities, previous RAG systems cannot retrieve question-relevant knowledge from KGs by matching questions with the meaningless identifiers of anonymous entities. To realize an effective RAG system in this scenario, two key challenges must be addressed: (1) How can anonymous entities be converted into retrievable information. (2) How to retrieve question-relevant anonymous entities. Hence, we propose a novel ARoG framework including relation-centric abstraction and structure-oriented abstraction strategies. For challenge (1), the first strategy abstracts entities into high-level concepts by dynamically capturing the semantics of their adjacent relations. It supplements meaningful semantics which can further support the retrieval process. For challenge (2), the second strategy transforms unstructured natural language questions into structured abstract concept paths. These paths can be more effectively aligned with the abstracted concepts in KGs, thereby improving retrieval performance. To guide LLMs to effectively retrieve knowledge from KGs, the two strategies strictly protect privacy from being exposed to LLMs. Experiments on three datasets demonstrate that ARoG achieves strong performance and privacy-robustness. △ Less

Submitted 3 December, 2025; v1 submitted 12 August, 2025; originally announced August 2025.

Comments: Accepted by AAAI 2026, camera ready version

Caption: Generating Informative Content Labels for Image Buttons Using Next-Screen Context

Authors: Mingyuan Zhong, Ajit Mallavarapu, Qing Nie

Abstract: We present Caption, an LLM-powered content label generation tool for visual interactive elements on mobile devices. Content labels are essential for screen readers to provide announcements for image-based elements, but are often missing or uninformative due to developer neglect. Automated captioning systems attempt to address this, but are limited to on-screen context, often resulting in inaccurat… ▽ More We present Caption, an LLM-powered content label generation tool for visual interactive elements on mobile devices. Content labels are essential for screen readers to provide announcements for image-based elements, but are often missing or uninformative due to developer neglect. Automated captioning systems attempt to address this, but are limited to on-screen context, often resulting in inaccurate or unspecific labels. To generate more accurate and descriptive labels, Caption collects next-screen context on interactive elements by navigating to the destination screen that appears after an interaction and incorporating information from both the origin and destination screens. Preliminary results show Caption generates more accurate labels than both human annotators and an LLM baseline. We expect Caption to empower developers by providing actionable accessibility suggestions and directly support on-demand repairs by screen reader users. △ Less

Submitted 12 August, 2025; originally announced August 2025.

SlideAudit: A Dataset and Taxonomy for Automated Evaluation of Presentation Slides

Authors: Zhuohao Jerry Zhang, Ruiqi Chen, Mingyuan Zhong, Jacob O. Wobbrock

Abstract: Automated evaluation of specific graphic designs like presentation slides is an open problem. We present SlideAudit, a dataset for automated slide evaluation. We collaborated with design experts to develop a thorough taxonomy of slide design flaws. Our dataset comprises 2400 slides collected and synthesized from multiple sources, including a subset intentionally modified with specific design probl… ▽ More Automated evaluation of specific graphic designs like presentation slides is an open problem. We present SlideAudit, a dataset for automated slide evaluation. We collaborated with design experts to develop a thorough taxonomy of slide design flaws. Our dataset comprises 2400 slides collected and synthesized from multiple sources, including a subset intentionally modified with specific design problems. We then fully annotated them using our taxonomy through strictly trained crowdsourcing from Prolific. To evaluate whether AI is capable of identifying design flaws, we compared multiple large language models under different prompting strategies, and with an existing design critique pipeline. We show that AI models struggle to accurately identify slide design flaws, with F1 scores ranging from 0.331 to 0.655. Notably, prompting techniques leveraging our taxonomy achieved the highest performance. We further conducted a remediation study to assess AI's potential for improving slides. Among 82.0% of slides that showed significant improvement, 87.8% of them were improved more with our taxonomy, further demonstrating its utility. △ Less

Submitted 5 August, 2025; originally announced August 2025.

Comments: UIST 2025

FlowETL: An Autonomous Example-Driven Pipeline for Data Engineering

Authors: Mattia Di Profio, Mingjun Zhong, Yaji Sripada, Marcel Jaspars

Abstract: The Extract, Transform, Load (ETL) workflow is fundamental for populating and maintaining data warehouses and other data stores accessed by analysts for downstream tasks. A major shortcoming of modern ETL solutions is the extensive need for a human-in-the-loop, required to design and implement context-specific, and often non-generalisable transformations. While related work in the field of ETL aut… ▽ More The Extract, Transform, Load (ETL) workflow is fundamental for populating and maintaining data warehouses and other data stores accessed by analysts for downstream tasks. A major shortcoming of modern ETL solutions is the extensive need for a human-in-the-loop, required to design and implement context-specific, and often non-generalisable transformations. While related work in the field of ETL automation shows promising progress, there is a lack of solutions capable of automatically designing and applying these transformations. We present FlowETL, a novel example-based autonomous ETL pipeline architecture designed to automatically standardise and prepare input datasets according to a concise, user-defined target dataset. FlowETL is an ecosystem of components which interact together to achieve the desired outcome. A Planning Engine uses a paired input-output datasets sample to construct a transformation plan, which is then applied by an ETL worker to the source dataset. Monitoring and logging provide observability throughout the entire pipeline. The results show promising generalisation capabilities across 14 datasets of various domains, file structures, and file sizes. △ Less

Submitted 30 July, 2025; originally announced July 2025.

From Queries to Criteria: Understanding How Astronomers Evaluate LLMs

Authors: Alina Hyk, Kiera McCormick, Mian Zhong, Ioana Ciucă, Sanjib Sharma, John F Wu, J. E. G. Peek, Kartheik G. Iyer, Ziang Xiao, Anjalie Field

Abstract: There is growing interest in leveraging LLMs to aid in astronomy and other scientific research, but benchmarks for LLM evaluation in general have not kept pace with the increasingly diverse ways that real people evaluate and use these models. In this study, we seek to improve evaluation procedures by building an understanding of how users evaluate LLMs. We focus on a particular use case: an LLM-po… ▽ More There is growing interest in leveraging LLMs to aid in astronomy and other scientific research, but benchmarks for LLM evaluation in general have not kept pace with the increasingly diverse ways that real people evaluate and use these models. In this study, we seek to improve evaluation procedures by building an understanding of how users evaluate LLMs. We focus on a particular use case: an LLM-powered retrieval-augmented generation bot for engaging with astronomical literature, which we deployed via Slack. Our inductive coding of 368 queries to the bot over four weeks and our follow-up interviews with 11 astronomers reveal how humans evaluated this system, including the types of questions asked and the criteria for judging responses. We synthesize our findings into concrete recommendations for building better benchmarks, which we then employ in constructing a sample benchmark for evaluating LLMs for astronomy. Overall, our work offers ways to improve LLM evaluation and ultimately usability, particularly for use in scientific research. △ Less

Submitted 5 August, 2025; v1 submitted 21 July, 2025; originally announced July 2025.

Comments: Accepted to the Conference on Language Modeling 2025 (COLM), 22 pages, 6 figures

AssistedDS: Benchmarking How External Domain Knowledge Assists LLMs in Automated Data Science

Authors: An Luo, Xun Xian, Jin Du, Fangqiao Tian, Ganghua Wang, Ming Zhong, Shengchun Zhao, Xuan Bi, Zirui Liu, Jiawei Zhou, Jayanth Srinivasa, Ashish Kundu, Charles Fleming, Mingyi Hong, Jie Ding

Abstract: Large language models (LLMs) have advanced the automation of data science workflows. Yet it remains unclear whether they can critically leverage external domain knowledge as human data scientists do in practice. To answer this question, we introduce AssistedDS (Assisted Data Science), a benchmark designed to systematically evaluate how LLMs handle domain knowledge in tabular prediction tasks. Assi… ▽ More Large language models (LLMs) have advanced the automation of data science workflows. Yet it remains unclear whether they can critically leverage external domain knowledge as human data scientists do in practice. To answer this question, we introduce AssistedDS (Assisted Data Science), a benchmark designed to systematically evaluate how LLMs handle domain knowledge in tabular prediction tasks. AssistedDS features both synthetic datasets with explicitly known generative mechanisms and real-world Kaggle competitions, each accompanied by curated bundles of helpful and adversarial documents. These documents provide domain-specific insights into data cleaning, feature engineering, and model selection. We assess state-of-the-art LLMs on their ability to discern and apply beneficial versus harmful domain knowledge, evaluating submission validity, information recall, and predictive performance. Our results demonstrate three key findings: (1) LLMs frequently exhibit an uncritical adoption of provided information, significantly impairing their predictive performance when adversarial content is introduced, (2) helpful guidance is often insufficient to counteract the negative influence of adversarial information, and (3) in Kaggle datasets, LLMs often make errors in handling time-series data, applying consistent feature engineering across different folds, and interpreting categorical variables correctly. These findings highlight a substantial gap in current models' ability to critically evaluate and leverage expert knowledge, underscoring an essential research direction for developing more robust, knowledge-aware automated data science systems. Our data and code are publicly available here: https://github.com/jeremyxianx/Assisted-DS △ Less

Submitted 22 October, 2025; v1 submitted 25 May, 2025; originally announced June 2025.

MSC Class: 62-07; 62-08; 68T05; 68T07; 68T01; 68T50 ACM Class: I.2.0; I.2.6; I.2.7; I.5.1; I.5.4; H.2.8; G.3

Aligning VLM Assistants with Personalized Situated Cognition

Authors: Yongqi Li, Shen Zhou, Xiaohu Li, Xin Miao, Jintao Wen, Mayi Xu, Jianhao Chen, Birong Pan, Hankun Kang, Yuanyuan Zhu, Ming Zhong, Tieyun Qian

Abstract: Vision-language models (VLMs) aligned with general human objectives, such as being harmless and hallucination-free, have become valuable assistants of humans in managing visual tasks. However, people with diversified backgrounds have different cognition even in the same situation. Consequently, they may have personalized expectations for VLM assistants. This highlights the urgent need to align VLM… ▽ More Vision-language models (VLMs) aligned with general human objectives, such as being harmless and hallucination-free, have become valuable assistants of humans in managing visual tasks. However, people with diversified backgrounds have different cognition even in the same situation. Consequently, they may have personalized expectations for VLM assistants. This highlights the urgent need to align VLM assistants with personalized situated cognition for real-world assistance. To study this problem, we first simplify it by characterizing individuals based on the sociological concept of Role-Set. Then, we propose to evaluate the individuals' actions to examine whether the personalized alignment is achieved. Further, we construct a benchmark named PCogAlignBench, which includes 18k instances and 20 individuals with different Role-Sets. Finally, we present a framework called PCogAlign, which constructs a cognition-aware and action-based reward model for personalized alignment. Experimental results and human evaluations demonstrate the reliability of the PCogAlignBench and the effectiveness of our proposed PCogAlign. We will open-source the constructed benchmark and code at https://github.com/NLPGM/PCogAlign. △ Less

Submitted 1 June, 2025; originally announced June 2025.

Comments: Accepted to ACL 2025 (main), camera-ready version

Bridge the Gap between Past and Future: Siamese Model Optimization for Context-Aware Document Ranking

Authors: Songhao Wu, Quan Tu, Mingjie Zhong, Hong Liu, Jia Xu, Jinjie Gu, Rui Yan

Abstract: In the realm of information retrieval, users often engage in multi-turn interactions with search engines to acquire information, leading to the formation of sequences of user feedback behaviors. Leveraging the session context has proven to be beneficial for inferring user search intent and document ranking. A multitude of approaches have been proposed to exploit in-session context for improved doc… ▽ More In the realm of information retrieval, users often engage in multi-turn interactions with search engines to acquire information, leading to the formation of sequences of user feedback behaviors. Leveraging the session context has proven to be beneficial for inferring user search intent and document ranking. A multitude of approaches have been proposed to exploit in-session context for improved document ranking. Despite these advances, the limitation of historical session data for capturing evolving user intent remains a challenge. In this work, we explore the integration of future contextual information into the session context to enhance document ranking. We present the siamese model optimization framework, comprising a history-conditioned model and a future-aware model. The former processes only the historical behavior sequence, while the latter integrates both historical and anticipated future behaviors. Both models are trained collaboratively using the supervised labels and pseudo labels predicted by the other. The history-conditioned model, referred to as ForeRanker, progressively learns future-relevant information to enhance ranking, while it singly uses historical session at inference time. To mitigate inconsistencies during training, we introduce the peer knowledge distillation method with a dynamic gating mechanism, allowing models to selectively incorporate contextual information. Experimental results on benchmark datasets demonstrate the effectiveness of our ForeRanker, showcasing its superior performance compared to existing methods. △ Less

Submitted 20 May, 2025; originally announced May 2025.

ACM Class: H.3.3