论文检索

Fukun Liu, Adam T. Greer, Gengchen Mai, Jin Sun 0011

Plankton are small drifting organisms found throughout the world's oceans and can be indicators of ocean health. One component of this plankton community is the zooplankton, which includes gelatinous animals and crustaceans (e.g. shrimp), as well as the early life stages (i.e., eggs and larvae) of many commercially important fishes. Being able to monitor zooplankton abundances accurately and understand how populations change in relation to ocean conditions is invaluable to marine science research, with important implications for future marine seafood productivity. While new imaging technologies generate massive amounts of video data of zooplankton, analyzing them using general-purpose computer vision tools turns out to be highly challenging due to the high similarity in appearance between the zooplankton and its background (e.g., marine snow). In this work, we present the ZooplanktonBench, a benchmark dataset containing images and videos of zooplankton associated with rich geospatial metadata (e.g., geographic coordinates, depth, etc.) in various water ecosystems. ZooplanktonBench defines a collection of tasks to detect, classify, and track zooplankton in challenging settings, including highly cluttered environments, living vs non-living classification, objects with similar shapes, and relatively small objects. Our dataset presents unique challenges and opportunities for state-of-the-art computer vision systems to evolve and improve visual understanding in dynamic environments characterized by significant variation and the need for geo-awareness. The code and settings described in this paper can be found on our website: https://lfk118.github.io/ZooplanktonBench_Webpage.

Junyong Lin, Lu Dai 0001, Ruiqian Han, Yijie Sui, Ruilin Wang, Xingliang Sun, Qinglin Wu, Min Feng, Hao Liu 0026, Hui Xiong 0001

Scientific researchers need intensive information about datasets to effectively evaluate and develop theories and methodologies. The information needs regarding datasets are implicitly embedded in particular research tasks, rather than explicitly expressed in search queries. However, existing scientific retrieval and question-answering (QA) datasets typically address straightforward questions, which do not align with the distribution of real-world research inquiries. To bridge this gap, we developed ScIRGen, a dataset generation framework for scientific QA & retrieval that more accurately reflects the information needs of professional science researchers, and uses it to create a large-scale scientific retrieval-augmented generation (RAG) dataset with realistic queries, datasets and papers. Technically, we designed a dataset-oriented information extraction method that leverages academic papers to augment the dataset representation. We then proposed a question generation framework by employing cognitive taxonomy to ensure the quality of synthesized questions. We also design a method to automatically filter synthetic answers based on the perplexity shift of LLMs, which is highly aligned with human judgment of answers' validity. Collectively, these methodologies culminated in the creation of the 61k QA dataset, ScIRGen-Geo. We benchmarked representative methods on the ScIRGen-Geo dataset for their question-answering and retrieval capabilities, finding out that current methods still suffer from reasoning from complex questions. This work advances the development of more sophisticated tools to support the intricate information needs of the scientific community.

Junhong Lin 0004, Xiaojie Guo 0002, Shuaicheng Zhang, Yada Zhu, Julian Shun

Graph mining has become crucial in fields such as social science, finance, and cybersecurity. Many large-scale real-world networks exhibit both heterogeneity, where multiple node and edge types exist in the graph, and heterophily, where connected nodes may have dissimilar labels and attributes. However, existing benchmarks primarily focus on either heterophilic homogeneous graphs or homophilic heterogeneous graphs, leaving a significant gap in understanding how models perform on graphs with both heterogeneity and heterophily. To bridge this gap, we introduce H2GB, a large-scale node-classification graph benchmark that brings together the complexities of both the heterophily and heterogeneity properties of real-world graphs. H2GB encompasses 9 real-world datasets spanning 5 diverse domains, 28 baseline models, and a unified benchmarking library with a standardized data loader, evaluator, unified modeling framework, and an extensible framework for reproducibility. We establish a standardized workflow supporting both model selection and development, enabling researchers to easily benchmark graph learning methods. Extensive experiments across 28 baselines reveal that current methods struggle with heterophilic and heterogeneous graphs, underscoring the need for improved approaches. Finally, we present a new variant of the model, H2G-former, developed following our standardized workflow, that excels at this challenging benchmark. Both the benchmark and the framework are publicly available at Github and PyPI, with documentation hosted at https://junhongmit.github.io/H2GB.

Zhe Li 0011, Xiangfei Qiu, Peng Chen 0038, Yihang Wang 0004, Hanyin Cheng, Yang Shu 0001, Jilin Hu, Chenjuan Guo, Aoying Zhou, Christian S. Jensen 等

Time Series Forecasting (TSF) is key functionality in numerous fields, such as financial investment, weather services, and energy management. Although increasingly capable TSF methods occur, many of them require domain-specific data collection and model training and do not generalize well when applied in other domains. Time Series Foundation Models (TSFMs) that are pre-trained on massive heterogeneous time series data aim to overcome these limitations. The prospects for generalizability have spurred the development of a new generation of TSFMs. This study proposes a benchmark, TSFM-Bench, to facilitate comprehensive and unified evaluation of TSFMs. TSFM-Bench covers a wide range of TSFMs, including those based on large language models and those pre-trained on time series data. TSFM-Bench supports multiple forecasting scenarios, including zero-shot, few-shot, and full-shot, enabling assessment across the full range of adaptation strategies. TSFM-Bench also provides a standardized experimental protocols for critical evaluation processes such as dataset splitting, loading, normalization, and few-shot sampling, facilitating consistency and fairness. We report on an extensive evaluation of TSFMs across a diverse range of datasets spanning multiple domains and exhibiting varied statistical characteristics. Specifically, we identify pros and cons and inherent limitations of existing TSFMs, and we propose potential directions for new model designs.

Yuchen Li 0006, Hao Zhang 0156, Haojie Zhang, Hengyi Cai, Xinyu Ma 0001, Shuaiqiang Wang, Haoyi Xiong, Zhaochun Ren, Maarten de Rijke, Dawei Yin 0001

The exponential growth of online content and increasingly diverse user needs have underscored the necessity for ranking models that go beyond traditional relevance assessments. Although several open-source benchmarks have significantly advanced academic research in Learning-to-Rank (LTR), these datasets predominantly focus on either text-based relevance or user behavior (click-through or dwell time) signals separately. This separation has inadvertently burdened academic progress by limiting the exploration of multifaceted, satisfaction-oriented ranking models. In contrast, industry research has begun to delve into integrated approaches that fuse prior (relevance, authority, recency, and quality) with posterior (user interaction such as clicks and dwell time) signals, thereby better capturing true user satisfaction. In this paper, we introduce FULTR-a large-scale, prior-posterior FUsion LTR dataset. FULTR comprises over 224M queries and 683M documents from Baidu Search, combining both: (1) a rich prior-attribute set with detailed textual relevance, authority, recency, and quality features, and (2) a comprehensive posterior-attribute set enriched by user click data, dwell time, and positional information. By unifying these dual perspectives, FULTR establishes a robust, reproducible benchmark for satisfaction-oriented ranking, enabling researchers to develop models that better capture real-world search behaviors and user satisfaction. In addition, we propose a strong LTR baseline that merges a satisfaction ranker that leverages pre-trained language models to integrate diverse satisfaction signals, with a behavior ranker that captures user interactions using a dual-tower approach. Their outputs are combined via a fusion layer, yielding significant performance gains in multiple evaluation metrics, as confirmed by extensive experiments and ablation studies. We are confident that our contribution not only democratizes access to industrial-grade fusion data for the research community but also paves the way for more effective and holistic LTR model design. FULTR is available to the research community at https://github.com/zhanghao731/FULTR.

Ekaterina Kuzmina, Dmitrii Kriukov, Mikhail Lebedev 0001, Dmitry V. Dylov

Adaptive deep brain stimulation (aDBS) has emerged as a promising treatment for Parkinson's disease (PD). In aDBS, a surgically placed electrode sends dynamically altered stimuli to the brain based on neurophysiological feedback: an invasive gadget that limits the amount of data one could collect for optimizing the control offline. As a consequence, a plethora of synthetic models of PD and those of the control algorithms have been proposed. Herein, we introduce the first neurophysiologically realistic benchmark for comparing said models. Specifically, our methodology covers not only conventional basal ganglia circuit dynamics and pathological oscillations, but also captures 15 previously dismissed physiological attributes, such as signal instabilities and noise, neural drift, electrode conductance changes and individual variability - all modeled as spatially distributed and temporally registered features via beta-band activity in the brain and a feedback. Furthermore, we purposely built our framework as a structured environment for training and evaluating deep reinforcement learning (RL) algorithms, opening new possibilities for optimizing aDBS control strategies and inviting the machine learning community to contribute to the emerging field of intelligent neurostimulation interfaces. Code repository: https://github.com/NevVerVer/DBS-Gym

Dmitrii Kriukov, Evgeniy Efimov, Ekaterina Kuzmina, Anastasiia Dudkovskaia, Ekaterina E. Khrameeva, Dmitry V. Dylov

The success of clinical trials of longevity drugs relies heavily on identifying integrative health and aging biomarkers, such as biological age. Epigenetic aging clocks predict the biological age of an individual using their DNA methylation profiles, commonly retrieved from blood samples. However, there is no standardized methodology to validate and compare epigenetic clock models as yet. We propose ComputAgeBench, a unifying framework that comprises such a methodology and a dataset for comprehensive benchmarking of different clinically relevant aging clocks. Our methodology exploits the core idea that reliable aging clocks must be able to distinguish between healthy individuals and those with aging-accelerating conditions. Specifically, we collected and harmonized 66 public datasets of blood DNA methylation, covering 19 such conditions across different ages, and tested 13 published clock models. Additionally, we compiled 46 separate datasets to facilitate the training of new aging clocks. We believe our work will bring the fields of aging biology and machine learning closer together for the research on reliable biomarkers of health and aging. Code https://github.com/ComputationalAgingLab/ComputAge Dataset https://huggingface.co/datasets/computage/computage_bench

Ruochen Kong 0001, Taylor Anderson 0001, Matthew Scotch, David J. Heslop, Yonchanok Khaokaew, Hao Xue 0001, Li Xiong 0001, Chandini Raina MacIntyre, Flora D. Salim, Andreas Züfle

Massive datasets related to infectious diseases became available after the COVID-19 pandemic, supporting data-driven approaches in modeling and forecasting infectious diseases. However, these approaches are known to exacerbate data biases present in the training data such as having certain demographic groups being over or underrepresented in the data. Such data collection biases may propagate through the modeling and prediction pipelines to decision-making, and the consequences are relatively unknown. Therefore, efforts are needed to understand how data collection bias affects data-driven infectious disease models. This datasets and benchmarks paper provides a suite of datasets, each corresponding to a simulated disease spread among a population of 5000 simulated agents over 90 days in Atlanta and San Francisco. For each dataset, we provide not only the full (simulated ground truth) of the disease spread in terms of when, where, and by whom the disease spreads, but also information on which cases are observed when different types and degrees of data collection bias are applied. The agents' characteristics, check-ins, and social network data are also available to support downstream tasks. Additionally, we also describe how to use the simulation to re-generate the data and to generate new datasets in different regions and with different parameters. With the provided datasets and the simulation tools, researchers studying the spread of infectious diseases may better understand, account for, and correct the systematic bias caused by the inherent real-world data bias, and hence improve the prediction of infectious diseases.

Suyeon Kim, SeongKu Kang, Dongwoo Kim 0002, Jungseul Ok, Hwanjo Yu

该来源暂未提供摘要。

Chaewon Kang, Seoyoon Jeong, Jonghyun Lee, Daejin Choi, Simon S. Woo, Jinyoung Han

The rapid development and prevalence of generative AI have made it easy for people to create high-quality deepfake images and videos, but their abuses have also increased exponentially. To mitigate potential social disruption, it is crucial to quickly detect the authenticity of each deepfake content hidden in a sea of information. While researchers have worked on developing deep learning-based methods, the deepfake datasets utilized in these studies are far from the real world in terms of their qualities; most popular deepfake datasets are human-distinguishable. To address this problem, we present a novel deepfake dataset, HiDF, a high-quality and human-indistinguishable deepfake dataset consisting of 62K images and 8K videos. HiDF is a meticulously curated dataset that includes diverse subjects that have undergone rigorous quality checks. A comparison of the quality between HiDF and existing deepfake datasets demonstrates that HiDF is human-indistinguishable. Hence, it can be a valuable benchmark dataset for deepfake detection tasks. Data and code (https://github.com/DSAIL-SKKU/HiDF) are publicly available for future deepfake detection research.

Robert J. Joyce, Gideon Miller, Phil Roth 0002, Richard Zak, Elliott Zaresky-Williams, Hyrum S. Anderson, Edward Raff, James Holt

A lack of accessible data has historically restricted malware analysis research, and practitioners have relied heavily on datasets provided by industry sources to advance. Existing public datasets are limited by narrow scope - most include files targeting a single platform, have labels supporting just one type of malware classification task, and make no effort to capture the evasive files that make malware detection difficult in practice. We present EMBER2024, a new dataset that enables holistic evaluation of malware classifiers. Created in collaboration with the authors of EMBER2017 and EMBER2018, the EMBER2024 dataset includes hashes, metadata, feature vectors, and labels for more than 3.2 million files from six file formats. Our dataset supports the training and evaluation of machine learning models on seven malware classification tasks, including malware detection, malware family classification, and malware behavior identification. EMBER2024 is the first to include a collection of malicious files that initially went undetected by a set of antivirus products, creating a ''challenge'' set to assess classifier performance against evasive malware. This work also introduces EMBER feature version 3, with added support for several new feature types. We are releasing the EMBER2024 dataset to promote reproducibility and empower researchers in the pursuit of new malware research topics.

Zhuohang Jiang, Pangjing Wu, Ziran Liang, Peter Q. Chen, Xu Yuan 0007, Ye Jia, Jiancheng Tu, Chen Li 0023, Peter H. F. Ng, Qing Li 0001

Structure reasoning is a fundamental capability of large language models (LLMs), enabling them to reason about structured commonsense and answer multi-hop questions. However, existing benchmarks for structure reasoning mainly focus on horizontal and coordinate structures (e.g. graphs), overlooking the hierarchical relationships within them. Hierarchical structure reasoning is crucial for human cognition, particularly in memory organization and problem-solving. It also plays a key role in various real-world tasks, such as information extraction and decision-making. To address this gap, we propose HiBench, the first framework designed to systematically benchmark the hierarchical reasoning capabilities of LLMs from initial structure generation to final proficiency assessment. It encompasses six representative scenarios, covering both fundamental and practical aspects, and consists of 30 tasks with varying hierarchical complexity, totaling 39,519 queries. To evaluate LLMs comprehensively, we develop five capability dimensions that depict different facets of hierarchical structure understanding. Through extensive evaluation of 20 LLMs from 10 model families, we reveal key insights into their capabilities and limitations: 1) existing LLMs show proficiency in basic hierarchical reasoning tasks; 2) they still struggle with more complex structures and implicit hierarchical representations, especially in structural modification and textual reasoning. Based on these findings, we create a small yet well-designed instruction dataset, which enhances LLMs' performance on HiBench by an average of 88.84% (Llama-3.1-8B) and 31.38% (Qwen2.5-7B) across all tasks. The HiBench dataset and toolkit are available at https://github.com/jzzzzh/HiBench to encourage evaluation.

Tenghao Huang, Dong Hee Lee, John Sweeney, Jiatong Shi, Emily Steliotes, Matthew Lange, Jonathan May, Muhao Chen 0001

Flavor development in the food industry is increasingly challenged by the need for rapid innovation and precise flavor profile creation. Traditional flavor research methods typically rely on iterative, subjective testing, which lacks the efficiency and scalability required for modern demands. This paper presents three contributions to address these challenges. Firstly, we define a new problem domain for scientific agents in flavor science, conceptualized as the generation of hypotheses for flavor profile sourcing and understanding. By leveraging their capacity to identify relevant evidence and reason within large context spaces, language model-backed agents can perform the labor-intensive tasks of flavor sourcing and understanding with enhanced efficiency and precision. To facilitate research in this area, we introduce the FoodPuzzle dataset, a challenging benchmark consisting of 978 food items and 1,766 flavor molecule profiles. We propose a novel Scientific Agent approach, integrating in-context learning and retrieval augmented techniques to generate grounded hypotheses in the domain of food science. Experimental results indicate that our model significantly surpasses traditional methods in flavor profile prediction tasks, demonstrating its potential to transform flavor development practices.

Benhao Huang, Yingzhuo Yu, Jin Huang, Xingjian Zhang 0002, Jiaqi W. Ma

The quality of datasets plays an increasingly crucial role in the research and development of modern artificial intelligence (AI). Despite the proliferation of open dataset platforms nowadays, data quality issues, such as incomplete documentation, inaccurate labels, ethical concerns, and outdated information, remain common in widely used datasets. Furthermore, these issues are often subtle and difficult to be detected by rule-based scripts, therefore requiring identification and verification by dataset users or maintainers-a process that is both time-consuming and prone to human mistakes. With the surging ability of large language models (LLM), it's promising to streamline the discovery of hidden dataset issues with LLM agents. To achieve this, one significant challenge is enabling LLM agents to detect issues in the wild rather than simply fixing known ones. In this work, we establish a benchmark to measure LLM agent's ability to tackle this challenge. We carefully curate 221 real-world test cases from eight popular dataset platforms and propose an automatic evaluation framework using GPT-4o. Our proposed framework shows strong empirical alignment with expert evaluations, validated through extensive comparisons with human annotations. Without any hints, most competitive Curator agent can only reveal ~30% of the data quality issues in the proposed dataset, highlighting the complexity of this task and indicating that applying LLM agents to real-world dataset curation still requires further in-depth exploration and innovation. The data and code is available at https://github.com/TRAIS-Lab/dca-bench.

Zixuan Gu, Qiufeng Fan, Long Sun, Yang Liu 0165, Xiaojun Ye 0001

With the advancement of Large Language Models (LLMs), LLM applications have expanded into a growing number of fields. However, users with data privacy concerns face limitations in directly utilizing LLM APIs, while private deployments incur significant computational demands. This creates a substantial challenge in achieving secure LLM adaptation under constrained local resources. To address this issue, collaborative learning methods, such as Split Learning (SL), offer a resource-efficient and privacy-preserving solution for adapting LLMs to private domains. In this study, we introduce VFLAIR-LLM (available at https://github.com/FLAIR-THU/VFLAIR-LLM), an extensible and lightweight split learning framework for LLMs, enabling privacy-preserving LLM inference and fine-tuning in resource-constrained environments. Our library provides two LLM partition settings, supporting three task types and 18 datasets. In addition, we provide standard modules for implementing and evaluating attacks and defenses. We benchmark 5 attacks and 9 defenses under various Split Learning for LLM(SL-LLM) settings, offering concrete insights and recommendations on the choice of model partition configurations, defense strategies, and relevant hyperparameters for real-world applications.

Philippe Gervais, Anastasiia Fadeeva, Andrii Maksai

Recognition of handwritten mathematical expressions allows to transfer scientific notes into their digital form. It facilitates the sharing, searching, and preservation of scientific information. We introduce MathWriting, the largest online handwritten mathematical expression dataset to date. It consists of 230k human-written samples and an additional 400k synthetic ones. This dataset can also be used in its rendered form for offline HME recognition. One MathWriting sample consists of a formula written on a touch screen and a corresponding LaTeX expression. We also provide a normalized version of LaTeX expression to simplify the recognition task and enhance the result quality. We provide baseline performance of standard models like OCR and CTC Transformer as well as Vision-Language Models like PaLI on the dataset. The dataset together with an example colab is accessible on Github.

Michael Galarnyk, Veer Kejriwal, Agam Shah, Yash Bhardwaj, Nicholas Watney Meyer, Anand Krishnan, Sudheer Chava

Social media has amplified the reach of financial influencers known as ''finfluencers,'' who share stock recommendations on platforms like YouTube. Understanding their influence requires analyzing multimodal signals like tone, delivery style, and facial expressions, which extend beyond text-based financial analysis. We introduce VideoConviction, a multimodal dataset with 6,000+ expert annotations, produced through 457 hours of human effort, to benchmark multimodal large language models (MLLMs) and text-based large language models (LLMs) in financial discourse. Our results show that while multimodal inputs improve stock ticker extraction (e.g., extracting Apple's ticker AAPL), both MLLMs and LLMs struggle to distinguish investment actions and conviction-the strength of belief conveyed through confident delivery and detailed reasoning-often misclassifying general commentary as definitive recommendations. While high-conviction recommendations perform better than low-conviction ones, they still underperform the popular S&P 500 index fund. An inverse strategy-betting against finfluencer recommendations-outperforms the S&P 500 by 6.8% in annual returns but carries greater risk (Sharpe ratio of 0.41 vs. 0.65). Our benchmark enables a diverse evaluation of multimodal tasks, comparing model performance on both full video and segmented video inputs. This enables deeper advancements in multimodal financial research. Our code, dataset, and evaluation leaderboard are available under the CC BY-NC 4.0 license.

Kei Sen Fong, Mehul Motani

Symbolic Regression (SR) algorithms are powerful machine learning tools that discover mathematical expressions from data, but their evaluation is often hindered by the computational cost of optimizing numerical parameters (also called constants) within candidate expressions. To address this limitation, we introduce POVE (PreOptimized Vault of Expressions), a comprehensive repository of preoptimized symbolic expressions where numerical parameters have been efficiently computed using an established optimization technique in SR (i.e., BFGS) on widely used SR regression problems (i.e., SRBench). In addition to optimized constants, POVE stores the corresponding fitness values, quantifying the accuracy of each expression on its respective regression problem, allowing researchers to directly retrieve both optimized expressions and their precomputed performance metrics from memory (e.g., hash tables or dictionaries). By eliminating the need for computationally expensive numerical optimization steps, POVE enables significantly faster evaluation, large-scale benchmarking, and extensive hyperparameter analysis. It provides a structured and standardized foundation for comparing SR algorithms, assessing their consistency across large numbers of random seeds and investigating the impact of hyperparameters without the overhead of numerical parameter optimization. By offering a collection of precomputed expressions, parameters, and fitness values across diverse regression problems, POVE enhances reproducibility, accelerates experimental workflows, and enables scalable SR research. It serves as a valuable resource for both developing novel SR methodologies and improving existing algorithms by providing a reliable reference for optimal expression discovery. With POVE, SR research can shift away from costly optimization, enabling deeper algorithmic exploration and more comprehensive systematic evaluation. POVE also lowers the barrier to entry for SR research and makes large-scale evaluation significantly more accessible. POVE is publicly available at: https://github.com/kentridgeai/POVE

Muhammad Hasan Ferdous, Emam Hossain, Md. Osman Gani

Robust causal discovery in time series datasets depends on reliable benchmark datasets with known ground-truth causal relationships. However, such datasets remain scarce, and existing synthetic alternatives often overlook critical temporal properties inherent in real-world data, including nonstationarity driven by trends and seasonality, irregular sampling intervals, and the presence of unobserved confounders. To address these challenges, we introduce TimeGraph, a comprehensive suite of synthetic time-series benchmark datasets that systematically incorporates both linear and nonlinear dependencies while modeling key temporal characteristics such as trends, seasonal effects, and heterogeneous noise patterns. Each dataset is accompanied by a fully specified causal graph featuring varying densities and diverse noise distributions and is provided in two versions: one including unobserved confounders and one without, thereby offering extensive coverage of real-world complexity while preserving methodological neutrality. We further demonstrate the utility of TimeGraph through systematic evaluations of state-of-the-art causal discovery algorithms including PCMCI+, LPCMCI, and FGES across a diverse array of configurations and metrics. Our experiments reveal significant variations in algorithmic performance under realistic temporal conditions, underscoring the need for robust synthetic benchmarks in the fair and transparent assessment of causal discovery methods. The complete TimeGraph suite, including dataset generation scripts, evaluation metrics, and recommended experimental protocols, is freely available to facilitate reproducible research and foster community-driven advancements in time-series causal discovery.

Jie Feng 0002, Jun Zhang 0087, Tianhui Liu, Xin Zhang 0106, Tianjian Ouyang, Junbo Yan, Yuwei Du, Siqi Guo, Yong Li 0008

As large language models (LLMs) continue to advance and gain widespread use, establishing systematic and reliable evaluation methodologies for LLMs and vision-language models (VLMs) has become essential to ensure their real-world effectiveness and reliability. There have been some early explorations about the usability of LLMs for limited urban tasks, but a systematic and scalable evaluation benchmark is still lacking. The challenge in constructing a systematic evaluation benchmark for urban research lies in the diversity of urban data, the complexity of application scenarios and the highly dynamic nature of the urban environment. In this paper, we design CityBench, an interactive simulator based evaluation platform, as the first systematic benchmark for evaluating the capabilities of LLMs for diverse tasks in urban research. First, we build CityData to integrate the diverse urban data and CitySimu to simulate fine-grained urban dynamics. Based on CityData and CitySimu, we design 8 representative urban tasks in 2 categories of perception-understanding and decision-making as the CityBench. With extensive results from 30 well-known LLMs and VLMs in 13 cities around the world, we find that advanced LLMs and VLMs can achieve competitive performance in diverse urban tasks requiring commonsense and semantic understanding abilities, e.g., understanding the human dynamics and semantic inference of urban images. Meanwhile, they fail to solve the challenging urban tasks requiring professional knowledge and high-level numerical abilities, e.g., geospatial prediction and traffic control task. These findings provide critical insights for the effective utilization and further development of LLMs to advance urban-related tasks and research in the future.