论文检索

Shiting Huang, Zhen Fang, Zehui Chen, Siyu Yuan, Junjie Ye, Yu Zeng, Lin Chen, Qi Mao, Feng Zhao

The ability of large language models (LLMs) to utilize external tools has enabled them to tackle an increasingly diverse range of tasks. However, as the tasks become more complex and long-horizon, the intricate tool utilization process may trigger various unexpected errors. Therefore, how to effectively handle such errors, including identifying, diagnosing, and recovering from them, has emerged as a key research direction for advancing tool learning. In this work, we first extensively analyze the types of errors encountered during the function-calling process on several competitive tool evaluation benchmarks. Based on it, we introduce CRITICTOOL, a comprehensive critique evaluation benchmark specialized for tool learning. Building upon a novel evolutionary strategy for dataset construction, CRITICTOOL holds diverse tool-use errors with varying complexities, which better reflects real-world scenarios. We conduct extensive experiments on CRITICTOOL, and validate the generalization and effectiveness of our constructed benchmark strategy. We also provide an in-depth analysis of the tool reflection ability on various LLMs, offering a new perspective on the field of tool learning in LLMs. The code is available at https://github.com/Shellorley0513/CriticTool.

Yu Sun, Xingyu Qian, Weiwen Xu, Hao Zhang, Chenghao Xiao, Long Li, Deli Zhao, Wenbing Huang, Tingyang Xu, Qifeng Bai 等

Reasoning-based large language models have excelled in mathematics and programming, yet their potential in knowledge-intensive medical question answering remains underexplored and insufficiently validated in clinical contexts. To bridge this gap, we introduce ReasonMed, the largest medical reasoning dataset to date, comprising 370k high-quality examples distilled from 1.75 million initial reasoning paths generated by complementary LLMs and curated through a cost-efficient easy-medium-difficult (EMD) pipeline. ReasonMed is built through a multi-agent generation, verification, and refinement process, in which an Error Refiner improves reasoning paths by correcting error-prone steps identified by a verifier. Using ReasonMed, we investigate effective strategies for training medical reasoning models and find that integrating detailed CoT reasoning with concise answer summaries yields the most robust fine-tuning results. Models trained on ReasonMed set a new benchmark: ReasonMed-7B surpasses the prior best sub-10B models by 4.17% and even exceeds LLaMA3.1-70B on PubMedQA by 4.60%. When scaled to ReasonMed-14B, it remains highly competitive, underscoring consistent scaling potential. The codes and datasets are available at https://github.com/YuSun-Work/ReasonMed.

WenHao Wang, Zijie Yu, Rui Ye, Jianqing Zhang, Guangyi Liu, Liang Liu, Siheng Chen, Yanfeng Wang

Mobile GUI agents have attracted tremendous research participation recently. Traditional approaches to mobile agent training rely on centralized data collection, leading to high cost and limited scalability. Distributed training utilizing federated learning offers an alternative by harnessing real-world user data, providing scalability and reducing costs. However, pivotal challenges, including the absence of standardized benchmarks, hinder progress in this field. To tackle the challenges, we introduce FedMABench, the first benchmark for federated training and evaluation of mobile GUI agents, specifically designed for heterogeneous scenarios. FedMABench features 6 datasets with 30+ subsets, 8 federated algorithms, 10+ base models, and over 800 apps across 5 categories, providing a comprehensive framework for evaluating mobile agents across diverse environments. Through extensive experiments, we uncover several key insights: federated algorithms consistently outperform local training; the distribution of specific apps plays a crucial role in heterogeneity; and, even apps from distinct categories can exhibit correlations during training. FedMABench is publicly available at: https://github.com/wwh0411/FedMABench.

Ramya Keerthy Thatikonda, Wray Buntine, Ehsan Shareghi

Logical reasoning is a critical benchmark for evaluating the capabilities of large language models (LLMs), as it reflects their ability to derive valid conclusions from given premises. While the combination of test-time scaling with dedicated outcome or process reward models has opened up new avenues to enhance LLMs performance in complex reasoning tasks, this space is under-explored in deductive logical reasoning. We present a set of Outcome Reward Models (ORMs) for deductive reasoning. To train the ORMs we mainly generate data using Chain-of-Thought (CoT) with single and multiple samples. Additionally, we propose a novel tactic to further expand the type of errors covered in the training dataset of the ORM. In particular, we propose an echo generation technique that leverages LLMs’ tendency to reflect incorrect assumptions made in prompts to extract additional training data, covering previously unexplored error types. While a standard CoT chain may contain errors likely to be made by the reasoner, the echo strategy deliberately steers the model toward incorrect reasoning. We show that ORMs trained on CoT and echo-augmented data demonstrate improved performance on the FOLIO, JustLogic, and ProverQA datasets across four different LLMs.

Suqing Wang, Zuchao Li, Shi Luohe, Bo Du, Hai Zhao, Yun Li, Qianren Wang

Large language models (LLMs) have achieved remarkable success across various domains, driving significant technological advancements and innovations. Despite the rapid growth in model scale and capability, systematic, data-driven research on how structural configurations affect performance remains scarce. To address this gap, we present a large-scale dataset encompassing diverse open-source LLM structures and their performance across multiple benchmarks. Leveraging this dataset, we conduct a systematic, data mining-driven analysis to validate and quantify the relationship between structural configurations and performance. Our study begins with a review of the historical development of LLMs and an exploration of potential future trends. We then analyze how various structural choices impact performance across benchmarks and further corroborate our findings using mechanistic interpretability techniques. By providing data-driven insights into LLM optimization, our work aims to guide the targeted development and application of future models.

Zesen Lyu, Dandan Zhang, Wei Ye, Fangdi Li, Zhihang Jiang, Yao Yang

Spatial reasoning is a core component of human cognition, enabling individuals to perceive, comprehend, and interact with the physical world. It relies on a nuanced understanding of spatial structures and inter-object relationships, serving as the foundation for complex reasoning and decision-making. To investigate whether current vision-language models (VLMs) exhibit similar capability, we introduce Jigsaw-Puzzles, a novel benchmark consisting of 1,100 carefully curated real-world images with high spatial complexity. Based on this dataset, we design five tasks to rigorously evaluate VLMs’ spatial perception, structural understanding, and reasoning capabilities, while deliberately minimizing reliance on domain-specific knowledge to better isolate and assess the general spatial reasoning capability. We conduct a comprehensive evaluation across 24 state-of-the-art VLMs. The results show that even the strongest model, Gemini-2.5-Pro, achieves only 77.14% overall accuracy and performs particularly poorly on the Order Generation task, with only 30.00% accuracy, far below the 90%+ performance achieved by human participants. This persistent gap underscores the need for continued progress, positioning Jigsaw-Puzzles as a challenging and diagnostic benchmark for advancing spatial reasoning research in VLMs. Our project page is at https://zesen01.github.io/jigsaw-puzzles.

Guanyu Hou, Jiaming He, Yinhang Zhou, Ji Guo, Yitong Qiao, Rui Zhang, Wenbo Jiang

Large Audio-Language Models (LALMs) are increasingly deployed in real-world applications, yet their robustness against malicious audio injection remains underexplored. To address this gap, this study systematically evaluates five leading LALMs across four attack scenarios: Audio Interference Attack, Instruction Following Attack, Context Injection Attack, and Judgment Hijacking Attack. We quantitatively assess their vulnerabilities and resilience using metrics: the Defense Success Rate, Context Robustness Score, and Judgment Robustness Index. The experiments reveal significant performance disparities, with no single model demonstrating consistent robustness across all attack types. Attack effectiveness is significantly influenced by the position of the malicious content, particularly when injected at the beginning of a sequence. Furthermore, our analysis uncovers a negative correlation between a model’s instruction-following capability and its robustness: models that strictly adhere to instructions tend to be more susceptible, whereas safety-aligned models exhibit greater resistance. To facilitate future research, this work introduces a comprehensive benchmark framework. Our findings underscore the critical need for integrating robustness into training pipelines and developing multi-modal defenses, ultimately facilitating the secure deployment of LALMs. The dataset used in this work is available on Hugging Face.

Sahana Ramnath, Anurag Mudgil, Brihi Joshi, Skyler Hallinan, Xiang Ren

Today, large language models are widely used as judges to evaluate responses from other language models. Hence, it is imperative to benchmark and improve these LLM-judges on real-world language model usage: a typical human-assistant conversation is lengthy, and shows significant diversity in topics, intents, and requirements across turns, e.g. social interactions, task requests, feedback. We present Amulet, a framework that leverages pertinent linguistic concepts of dialog-acts and maxims to improve the accuracy of LLM-judges on preference data with complex, multi-turn conversational context. Amulet presents valuable insights about (a) the communicative structures and intents present in the conversation (dialog acts), and (b) the satisfaction of conversational principles (maxims) by the preference pair responses, and uses them to make judgments. On 4 challenging datasets, Amulet shows that (a) humans frequently (60-70% of the time) change their intents from one turn of the conversation to the next, and (b) in ∼75% of instances, the preference responses can be differentiated via dialog acts and/or maxims, reiterating the latter’s significance in judging such data. Amulet can be used either as a judge by applying the framework to a single LLM, or integrated into a jury with different LLM judges; our judges and juries show strong improvements on relevant baselines for all 4 datasets.

Mintong Kang, Vinayshekhar Bannihatti Kumar, Shamik Roy, Abhishek Kumar, Sopan Khosla, Balakrishnan Murali Narayanaswamy, Rashmi Gangadharaiah

Text-to-image diffusion models often exhibit biases toward specific demographic groups, such as generating more males than females when prompted to generate images of engineers, raising ethical concerns and limiting their adoption. In this paper, we tackle the challenge of mitigating generation bias towards any target attribute value (e.g., “male” for “gender”) in diffusion models while preserving generation quality. We propose FairGen, an adaptive latent guidance mechanism which controls the generation distribution during inference. In FairGen, a latent guidance module dynamically adjusts the diffusion process to enforce specific attributes, while a memory module tracks the generation statistics and steers latent guidance to align with the targeted fair distribution of the attribute values. Further, given the limitations of existing datasets in comprehensively assessing bias in diffusion models, we introduce a holistic bias evaluation benchmark HBE, covering diverse domains and incorporating complex prompts across various applications. Extensive evaluations on HBE and Stable Bias datasets demonstrate that FairGen outperforms existing bias mitigation approaches, achieving substantial bias reduction (e.g., 68.5% gender bias reduction on Stable Diffusion 2). Ablation studies highlight FairGen’s ability to flexibly and precisely control generation distribution at any user-specified granularity, ensuring adaptive and targeted bias mitigation.

Shuya Feng, Yuan Hong

Training text embedding models under differential privacy constraints is challenging due to the high dimensionality of language data and the presence of rare, identifying linguistic features. We propose (Differentially Private Embedding Distillation), a framework that leverages teacher-student distillation with multi-layer noise injection to learn high-quality embeddings while providing differential privacy guarantees. DPED trains an ensemble of teacher models on disjoint subsets of sensitive text data, then transfers their knowledge to a student model through noisy aggregation at multiple layers. A rare-word-aware strategy adaptively handles infrequent words, improving privacy-utility trade-offs. Experiments on benchmark datasets demonstrate that DPED outperforms standard differentially private training methods, achieving substantially higher utility at the same privacy budget. Our approach protects individual word usage patterns in training documents, preventing models from memorizing unique linguistic fingerprints while maintaining practical utility for downstream NLP tasks. Source code is available at https://github.com/datasec-lab/DPED.

Lihui Liu

Knowledge graphs (KGs) enable reasoning tasks such as link prediction, question answering, and knowledge discovery. However, real-world KGs are often incomplete, making link prediction both essential and challenging. Existing methods, including embedding-based and path-based approaches, rely on Euclidean embeddings, which struggle to capture hierarchical structures. GNN-based methods aggregate information through message passing in Euclidean space, but they struggle to effectively encode the recursive tree-like structures that emerge in multi-hop reasoning. To address these challenges, we propose a hyperbolic GNN framework that embeds recursive learning trees in hyperbolic space and generates query-specific embeddings. By incorporating hierarchical message passing, our method naturally aligns with reasoning paths and dynamically adapts to queries, improving prediction accuracy. Unlike static embedding-based approaches, our model computes context-aware embeddings tailored to each query. Experiments on multiple benchmark datasets show that our approach consistently outperforms state-of-the-art methods, demonstrating its effectiveness in KG reasoning.

Guido Ivetta, Marcos J Gomez, Sofía Martinelli, Pietro Palombini, M Emilia Echeveste, Nair Carolina Mazzeo, Beatriz Busaniche, Luciana Benotti

Most resources for evaluating social biases in Large Language Models are developed without co-design from the communities affected by these biases, and rarely involve participatory approaches. We introduce HESEIA, a dataset of 46,499 sentences created in a professional development course. The course involved 370 high-school teachers and 5,370 students from 189 Latin-American schools. Unlike existing benchmarks, HESEIA captures intersectional biases across multiple demographic axes and school subjects. It reflects local contexts through the lived experience and pedagogical expertise of educators. Teachers used minimal pairs to create sentences that express stereotypes relevant to their school subjects and communities. We show the dataset diversity in term of demographic axes represented and also in terms of the knowledge areas included. We demonstrate that the dataset contains more stereotypes unrecognized by current LLMs than previous datasets. HESEIA is available to support bias assessments grounded in educational communities.

Sujoy Sarkar, Gourav Sarkar, Manoj Balaji Jagadeeshan, Jivnesh Sandhan, Amrith Krishna, Pawan Goyal

High lexical variation, ambiguous references, and long-range dependencies make entity resolution in literary texts particularly challenging. We present Mahānāma, the first large-scale dataset for end-to-end Entity Discovery and Linking (EDL) in Sanskrit, a morphologically rich and under-resourced language. Derived from the Mahābhārata , the world’s longest epic, the dataset comprises over 109K named entity mentions mapped to 5.5K unique entities, and is aligned with an English knowledge base to support cross-lingual linking. The complex narrative structure of Mahānāma, coupled with extensive name variation and ambiguity, poses significant challenges to resolution systems. Our evaluation reveals that current coreference and entity linking models struggle when evaluated on the global context of the test set. These results highlight the limitations of current approaches in resolving entities within such complex discourse. Mahānāma thus provides a unique benchmark for advancing entity resolution, especially in literary domains.

Dongwon Jung, Wenxuan Zhou, Muhao Chen

Training large language models (LLMs) with chain-of-thought (CoT) supervision has proven effective for enhancing their reasoning abilities. However, obtaining reliable and accurate reasoning supervision remains a significant challenge. We propose a scalable method for generating a high-quality CoT supervision dataset by leveraging the determinism of program execution. Unlike existing reasoning dataset generation methods that rely on costly human annotations or error-prone LLM-generated CoT, our approach extracts verifiable, step-by-step reasoning traces from code execution and transforms them into a natural language CoT reasoning. Experiments on reasoning benchmarks across various domains show that our method effectively equips LLMs with transferable reasoning abilities across diverse tasks. Furthermore, the ablation studies validate that our method produces highly accurate reasoning data and reduces overall token length during inference by reducing meaningless repetition and overthinking.

Li Zhou, Lutong Yu, Dongchu Xie, Shaohuan Cheng, Wenyan Li, Haizhou Li

Culture is a rich and dynamic domain that evolves across both geography and time. However, existing studies on cultural understanding with vision-language models (VLMs) primarily emphasize geographic diversity, often overlooking the critical temporal dimensions. To bridge this gap, we introduce Hanfu-Bench, a novel, expert-curated multimodal dataset. Hanfu, a traditional garment spanning ancient Chinese dynasties, serves as a representative cultural heritage that reflects the profound temporal aspects of Chinese culture while remaining highly popular in Chinese contemporary society. Hanfu-Bench comprises two core tasks: cultural visual understanding and cultural image transcreation. The former task examines temporal-cultural feature recognition based on single- or multi-image inputs through multiple-choice visual question answering, while the latter focuses on transforming traditional attire into modern designs through cultural element inheritance and modern context adaptation. Our evaluation shows that closed VLMs perform comparably to non-experts on visual cutural understanding but fall short by 10% to human experts, while open VLMs lags further behind non-experts. For the transcreation task, multi-faceted human evaluation indicates that the best-performing model achieves a success rate of only 42%. Our benchmark provides an essential testbed, revealing significant challenges in this new direction of temporal cultural understanding and creative adaptation.

Chuangtao Ma, Yongrui Chen, Tianxing Wu, Arijit Khan, Haofen Wang

Large language models (LLMs) have demonstrated remarkable performance on question-answering (QA) tasks because of their superior capabilities in natural language understanding and generation. However, LLM-based QA struggles with complex QA tasks due to poor reasoning capacity, outdated knowledge, and hallucinations. Several recent works synthesize LLMs and knowledge graphs (KGs) for QA to address the above challenges. In this survey, we propose a new structured taxonomy that categorizes the methodology of synthesizing LLMs and KGs for QA according to the categories of QA and the KG’s role when integrating with LLMs. We systematically survey state-of-the-art methods in synthesizing LLMs and KGs for QA and compare and analyze these approaches in terms of strength, limitations, and KG requirements. We then align the approaches with QA and discuss how these approaches address the main challenges of different complex QA. Finally, we summarize the advancements, evaluation metrics, and benchmark datasets and highlight open challenges and opportunities.

Jean De Dieu Nyandwi, Yueqi Song, Simran Khanuja, Graham Neubig

Multimodal Large Language Models excel in high-resource settings, but often misinterpret long-tail cultural entities and underperform in low-resource languages. To address this gap, we propose a data-centric approach that directly grounds MLLMs in cultural knowledge. Leveraging a large scale knowledge graph from Wikidata, we collect images that represent culturally significant entities, and generate synthetic multilingual visual question answering data. The resulting dataset, CulturalGround, comprises 22 million high-quality, culturally-rich VQA pairs spanning 42 countries and 39 languages. We train an open-source MLLM CulturalPangea on CulturalGround, interleaving standard multilingual instruction-tuning data to preserve general abilities. Cultural-Pangea achieves state-of-the-art performance among open models on various culture-focused multilingual multimodal benchmarks, outperforming prior models by an average of +5.0%without degrading results on mainstream vision–language tasks. Our findings show that our targeted, culturally grounded approach could substantially narrow the cultural gap in MLLMs and offer a practical path towards globally inclusive multimodal systems.

LiQing Xu, Qiwei Li, Tianshuo Peng, Zuchao Li, Hai Zhao, Ping Wang

Large language models (LLMs) have achieved remarkable success across diverse domains. However, their potential as effective language teachers—particularly in complex pedagogical scenarios like teaching Chinese as a second language—remains inadequately assessed. To address this gap, we propose the first pedagogical competence benchmark for LLMs, rigorously evaluating their performance against international standards for Chinese language teachers. Our framework spans three core dimensions: (1) basic knowledge evaluation, covering 32 subtopics across five major categories; (2) international teacher examination, based on data collected from international Chinese teacher certification exams; and (3) teaching practice evaluation, where target LLMs summarize knowledge points and design instructional content for student models, followed by testing the student models to assess the LLM’s ability to distill and teach key concepts.We conduct a comprehensive evaluation of 13 latest multilingual and Chinese LLMs. While most models demonstrate promising pedagogical potential, there remains substantial room for improvement in their teaching capabilities. This study contributes to the development of AI-assisted language education tools capable of rivaling human teaching excellence. The benchmark dataset and evaluation scripts used in this study are publicly available at https://github.com/Line-Kite/CLTE.

Weicheng Ma, John J. Guerrerio, Soroush Vosoughi

Research on stereotypes in large language models (LLMs) has largely focused on English-speaking contexts, due to the lack of datasets in other languages and the high cost of manual annotation in underrepresented cultures. To address this gap, we introduce a cost-efficient human-LLM collaborative annotation framework and apply it to construct EspanStereo, a Spanish-language stereotype dataset spanning multiple Spanish-speaking countries across Europe and Latin America. EspanStereo captures both well-documented stereotypes from prior literature and culturally specific biases absent from English-centric resources. Using LLMs to generate candidate stereotypes and in-culture annotators to validate them, we demonstrate the framework’s effectiveness in identifying nuanced, region-specific biases. Our evaluation of Spanish-supporting LLMs using EspanStereo reveals significant variation in stereotypical behavior across countries, highlighting the need for more culturally grounded assessments. Beyond Spanish, our framework is adaptable to other languages and regions, offering a scalable path toward multilingual stereotype benchmarks. This work broadens the scope of stereotype analysis in LLMs and lays the groundwork for comprehensive cross-cultural bias evaluation.

Xie Zhifei, Mingbao Lin, Zihang Liu, Pengcheng Wu, Shuicheng Yan, Chunyan Miao

Recent advancements in multimodal reasoning overlook the audio modality. We introduce Audio-Reasoner, a large-scale audio language model for deep reasoning. We meticulously curated a large-scale and diverse multi-task audio dataset with simple annotations. Then, we leverage closed-source models to conduct secondary labeling, QA generation, along with structured COT process. These datasets together form a high-quality reasoning dataset with 1.2 million reasoning-rich samples, which we name CoTA. Following inference scaling principles, we train Audio-Reasoner on CoTA, enabling it to achieve great logical capabilities in audio reasoning. Experiments show state-of-the-art performance across key benchmarks, including MMAU-mini (+25.42%), AIR-Bench chat/foundation (+14.57%/+10.13%), and MELD (+8.01%). Our findings stress the core of structured CoT training in advancing audio reasoning. The model, dataset, and code are open-sourced at [https://github.com/xzf-thu/Audio-Reasoner](https://github.com/xzf-thu/Audio-Reasoner) or [https://huggingface.co/datasets/zhifeixie/Audio-Reasoner-CoTA](https://huggingface.co/datasets/zhifeixie/Audio-Reasoner-CoTA).