Video large language models have not yet been widely deployed, largely due to their tendency to hallucinate. Typical benchmarks for Video-LLMs rely simply on multiple-choice questions. Unfortunately, VideoLLMs hallucinate far more aggressively on freeform text generation tasks like video captioning than they do on multiple choice verification tasks. To address this weakness, we propose ARGUS, a VideoLLM benchmark that measures freeform video captioning performance. By comparing VideoLLM outputs to human ground truth captions, ARGUS quantifies dual metrics. First, we measure the rate of hallucinations in the form of incorrect statements about video content or temporal relationships. Second, we measure the rate at which the model omits important descriptive details. Together, these dual metrics form a comprehensive view of video captioning performance.
Integrating Artificial Intelligence (AI) in educational settings has brought new learning approaches, transforming the practices of both students and educators. Among the various technologies driving this transformation, Large Language Models (LLMs) have emerged as powerful tools for creating educational materials and question answering, but there are still space for new applications. Educators commonly use Multiple-Choice Questions (MCQs) to assess student knowledge, but manually generating these questions is resource-intensive and requires significant time and cognitive effort. In our opinion, LLMs offer a promising solution to these challenges. This paper presents a novel comparative analysis of three widely known LLMs - Llama 2, Mistral, and GPT-3.5 - to explore their potential for creating informative and challenging MCQs. In our approach, we do not rely on the knowledge of the LLM, but we inject the knowledge into the prompt to contrast the hallucinations, giving the educators control over the test's source text, too. Our experiment involving 21 educators shows that GPT-3.5 generates the most effective MCQs across several known metrics. Additionally, it shows that there is still some reluctance to adopt AI in the educational field. This study sheds light on the potential of LLMs to generate MCQs and improve the educational experience, providing valuable insights for the future.
Although vision-language and large language models (VLM and LLM) offer promising opportunities for AI-driven educational assessment, their effectiveness in real-world classroom settings, particularly in underrepresented educational contexts, remains underexplored. In this study, we evaluated the performance of a state-of-the-art VLM and several LLMs on 646 handwritten exam responses from grade 4 students in six Indonesian schools, covering two subjects: Mathematics and English. These sheets contain more than 14K student answers that span multiple choice, short answer, and essay questions. Assessment tasks include grading these responses and generating personalized feedback. Our findings show that the VLM often struggles to accurately recognize student handwriting, leading to error propagation in downstream LLM grading. Nevertheless, LLM-generated feedback retains some utility, even when derived from imperfect input, although limitations in personalization and contextual relevance persist.
Background: Large language models (LLMs) have demonstrated substantial potential to support clinical practice. Other than Chat GPT4 and its predecessors, few LLMs, especially those of the leading and more powerful reasoning model class, have been subjected to medical specialty examination questions, including in the domain of primary care. This paper aimed to test the capabilities of leading LLMs as of May 2025 (o3, Claude Opus 4, Grok3, and Gemini 2.5 Pro) in primary care education, specifically in answering Member of the Royal College of General Practitioners (MRCGP) style examination questions. Methods: o3, Claude Opus 4, Grok3, and Gemini 2.5 Pro were tasked to answer 100 randomly chosen multiple choice questions from the Royal College of General Practitioners GP SelfTest on 25 May 2025. Questions included textual information, laboratory results, and clinical images. Each model was prompted to answer as a GP in the UK and was provided with full question information. Each question was attempted once by each model. Responses were scored against correct answers provided by GP SelfTest. Results: The total score of o3, Claude Opus 4, Grok3, and Gemini 2.5 Pro was 99.0%, 95.0%, 95.0%, and 95.0%, respectively. The average peer score for the same questions was 73.0%. Discussion: All models performed remarkably well, and all substantially exceeded the average performance of GPs and GP registrars who had answered the same questions. o3 demonstrated the best performance, while the performances of the other leading models were comparable with each other and were not substantially lower than that of o3. These findings strengthen the case for LLMs, particularly reasoning models, to support the delivery of primary care, especially those that have been specifically trained on primary care clinical data.
Large Language Models (LLMs) have achieved remarkable performance on a wide range of NLP benchmarks, often surpassing human-level accuracy. However, their reliability in high-stakes domains such as medicine, particularly in low-resource languages, remains underexplored. In this work, we introduce PersianMedQA, a large-scale, expert-validated dataset of multiple-choice Persian medical questions, designed to evaluate LLMs across both Persian and English. We benchmark over 40 state-of-the-art models, including general-purpose, Persian fine-tuned, and medical LLMs, in zero-shot and chain-of-thought (CoT) settings. Our results show that closed-source general models (e.g., GPT-4.1) consistently outperform all other categories, achieving 83.3% accuracy in Persian and 80.7% in English, while Persian fine-tuned models such as Dorna underperform significantly (e.g., 35.9% in Persian), often struggling with both instruction-following and domain reasoning. We also analyze the impact of translation, showing that while English performance is generally higher, Persian responses are sometimes more accurate due to cultural and clinical contextual cues. Finally, we demonstrate that model size alone is insufficient for robust performance without strong domain or language adaptation. PersianMedQA provides a foundation for evaluating multilingual and culturally grounded medical reasoning in LLMs. The PersianMedQA dataset can be accessed at: https://huggingface.co/datasets/MohammadJRanjbar/PersianMedQA
Hepato-pancreato-biliary (HPB) disorders represent a global public health challenge due to their high morbidity and mortality. Although large language models (LLMs) have shown promising performance in general medical question-answering tasks, the current evaluation benchmarks are mostly derived from standardized examinations or manually designed questions, lacking HPB coverage and clinical cases. To address these issues, we systematically eatablish an HPB disease evaluation benchmark comprising 3,535 closed-ended multiple-choice questions and 337 open-ended real diagnosis cases, which encompasses all the 33 main categories and 465 subcategories of HPB diseases defined in the International Statistical Classification of Diseases, 10th Revision (ICD-10). The multiple-choice questions are curated from public datasets and synthesized data, and the clinical cases are collected from prestigious medical journals, case-sharing platforms, and collaborating hospitals. By evalauting commercial and open-source general and medical LLMs on our established benchmark, namely ClinBench-HBP, we find that while commercial LLMs perform competently on medical exam questions, they exhibit substantial performance degradation on HPB diagnosis tasks, especially on complex, inpatient clinical cases. Those medical LLMs also show limited generalizability to HPB diseases. Our results reveal the critical limitations of current LLMs in the domain of HPB diseases, underscoring the imperative need for future medical LLMs to handle real, complex clinical diagnostics rather than simple medical exam questions. The benchmark will be released at https://clinbench-hpb.github.io.
Although Large Language Models (LLMs) have demonstrated strong language understanding and generation abilities across various languages, their cultural knowledge is often limited to English-speaking communities, which can marginalize the cultures of non-English communities. To address the problem, evaluation of the cultural awareness of the LLMs and the methods to develop culturally aware LLMs have been investigated. In this study, we focus on evaluating knowledge of folktales, a key medium for conveying and circulating culture. In particular, we focus on Japanese folktales, specifically on knowledge of Yokai. Yokai are supernatural creatures originating from Japanese folktales that continue to be popular motifs in art and entertainment today. Yokai have long served as a medium for cultural expression, making them an ideal subject for assessing the cultural awareness of LLMs. We introduce YokaiEval, a benchmark dataset consisting of 809 multiple-choice questions (each with four options) designed to probe knowledge about yokai. We evaluate the performance of 31 Japanese and multilingual LLMs on this dataset. The results show that models trained with Japanese language resources achieve higher accuracy than English-centric models, with those that underwent continued pretraining in Japanese, particularly those based on Llama-3, performing especially well. The code and dataset are available at https://github.com/CyberAgentA ILab/YokaiEval.
We introduce the $\underline{Ko}rean \underline{G}rammar \underline{E}valuation Bench\underline{M}ark (KoGEM)$, designed to assess the linguistic competence of LLMs and humans in Korean. KoGEM consists of 1.5k multiple-choice QA pairs covering five main categories and 16 subcategories. The zero-shot evaluation of 27 LLMs of various sizes and types reveals that while LLMs perform remarkably well on straightforward tasks requiring primarily definitional knowledge, they struggle with tasks that demand the integration of real-world experiential knowledge, such as phonological rules and pronunciation. Furthermore, our in-depth analysis suggests that incorporating such experiential knowledge could enhance the linguistic competence of LLMs. With KoGEM, we not only highlight the limitations of current LLMs in linguistic competence but also uncover hidden facets of LLMs in linguistic competence, paving the way for enhancing comprehensive language understanding. Our code and dataset are available at: https://github.com/SungHo3268/KoGEM.
Video understanding has made remarkable progress in recent years, largely driven by advances in deep models and the availability of large-scale annotated datasets. However, existing works typically ignore the inherent domain shifts encountered in real-world video applications, leaving domain generalization (DG) in video understanding underexplored. Hence, we propose Video Understanding Domain Generalization (VUDG), a novel dataset designed specifically for evaluating the DG performance in video understanding. VUDG contains videos from 11 distinct domains that cover three types of domain shifts, and maintains semantic similarity across different domains to ensure fair and meaningful evaluation. We propose a multi-expert progressive annotation framework to annotate each video with both multiple-choice and open-ended question-answer pairs. Extensive experiments on 9 representative large video-language models (LVLMs) and several traditional video question answering methods show that most models (including state-of-the-art LVLMs) suffer performance degradation under domain shifts. These results highlight the challenges posed by VUDG and the difference in the robustness of current models to data distribution shifts. We believe VUDG provides a valuable resource for prompting future research in domain generalization video understanding.
The performance of large language models (LLMs) continues to improve, as reflected in rising scores on standard benchmarks. However, the lack of transparency around training data raises concerns about potential overlap with evaluation sets and the fairness of reported results. Although prior work has proposed methods for detecting data leakage, these approaches primarily focus on identifying outliers and have not been evaluated under controlled simulated leakage conditions. In this work, we compare existing leakage detection techniques, namely permutation and n-gram-based methods, under a continual pretraining setup that simulates real-world leakage scenarios, and additionally explore a lightweight method we call semi-half question. Although semi-half offers a low-cost alternative, our analysis shows that the n-gram method consistently achieves the highest F1-score. We also refine these techniques to support instance-level detection and reduce computational overhead. Leveraging the best-performing method, we create cleaned versions of MMLU and HellaSwag, and re-evaluate several LLMs. Our findings present a practical path toward more reliable and transparent evaluations, and we recommend contamination checks as a standard step before releasing benchmark results.
Recent works on large language models (LLMs) have successfully demonstrated the emergence of reasoning capabilities via reinforcement learning (RL). Although recent efforts leverage group relative policy optimization (GRPO) for MLLMs post-training, they constantly explore one specific aspect, such as grounding tasks, math problems, or chart analysis. There are no works that can leverage multi-source MLLM tasks for stable reinforcement learning. In this work, we present a unified perspective to solve this problem. We present Mixed-R1, a unified yet straightforward framework that contains a mixed reward function design (Mixed-Reward) and a mixed post-training dataset (Mixed-45K). We first design a data engine to select high-quality examples to build the Mixed-45K post-training dataset. Then, we present a Mixed-Reward design, which contains various reward functions for various MLLM tasks. In particular, it has four different reward functions: matching reward for binary answer or multiple-choice problems, chart reward for chart-aware datasets, IoU reward for grounding problems, and open-ended reward for long-form text responses such as caption datasets. To handle the various long-form text content, we propose a new open-ended reward named Bidirectional Max-Average Similarity (BMAS) by leveraging tokenizer embedding matching between the generated response and the ground truth. Extensive experiments show the effectiveness of our proposed method on various MLLMs, including Qwen2.5-VL and Intern-VL on various sizes. Our dataset and model are available at https://github.com/xushilin1/mixed-r1.
Steering vectors are a lightweight method for controlling text properties by adding a learned bias to language model activations at inference time. So far, steering vectors have predominantly been evaluated in multiple-choice settings, while their effectiveness in free-form generation tasks remains understudied. Moving "Beyond Multiple Choice," we thoroughly evaluate the effectiveness of steering vectors in adaptively controlling topical focus, sentiment, toxicity, and readability in abstractive summaries of the NEWTS dataset. We find that steering effectively controls the targeted summary properties, but high steering strengths consistently degrade both intrinsic and extrinsic text quality. Compared to steering, prompting offers weaker control, while preserving text quality. Combining steering and prompting yields the strongest control over text properties and offers the most favorable efficacy-quality trade-off at moderate steering strengths. Our results underscore the practical trade-off between control strength and text quality preservation when applying steering vectors to free-form generation tasks.
If AI models can detect when they are being evaluated, the effectiveness of evaluations might be compromised. For example, models could have systematically different behavior during evaluations, leading to less reliable benchmarks for deployment and governance decisions. We investigate whether frontier language models can accurately classify transcripts based on whether they originate from evaluations or real-world deployment, a capability we call evaluation awareness. To achieve this, we construct a diverse benchmark of 1,000 prompts and transcripts from 61 distinct datasets. These span public benchmarks (e.g., MMLU, SWEBench), real-world deployment interactions, and agent trajectories from scaffolding frameworks (e.g., web-browsing agents). Frontier models clearly demonstrate above-random evaluation awareness (Gemini-2.5-Pro reaches an AUC of $0.83$), but do not yet surpass our simple human baseline (AUC of $0.92$). Furthermore, both AI models and humans are better at identifying evaluations in agentic settings compared to chat settings. Additionally, we test whether models can identify the purpose of the evaluation. Under multiple-choice and open-ended questioning, AI models far outperform random chance in identifying what an evaluation is testing for. Our results indicate that frontier models already exhibit a substantial, though not yet superhuman, level of evaluation-awareness. We recommend tracking this capability in future models.
Social bias is a critical issue in large vision-language models (VLMs), where fairness- and ethics-related problems harm certain groups of people in society. It is unknown to what extent VLMs yield social bias in generative responses. In this study, we focus on evaluating and mitigating social bias on both the model's response and probability distribution. To do so, we first evaluate four state-of-the-art VLMs on PAIRS and SocialCounterfactuals datasets with the multiple-choice selection task. Surprisingly, we find that models suffer from generating gender-biased or race-biased responses. We also observe that models are prone to stating their responses are fair, but indeed having mis-calibrated confidence levels towards particular social groups. While investigating why VLMs are unfair in this study, we observe that VLMs' hidden layers exhibit substantial fluctuations in fairness levels. Meanwhile, residuals in each layer show mixed effects on fairness, with some contributing positively while some lead to increased bias. Based on these findings, we propose a post-hoc method for the inference stage to mitigate social bias, which is training-free and model-agnostic. We achieve this by ablating bias-associated residuals while amplifying fairness-associated residuals on model hidden layers during inference. We demonstrate that our post-hoc method outperforms the competing training strategies, helping VLMs have fairer responses and more reliable confidence levels.
Traditional Chinese Medicine (TCM), as an effective alternative medicine, has been receiving increasing attention. In recent years, the rapid development of large language models (LLMs) tailored for TCM has underscored the need for an objective and comprehensive evaluation framework to assess their performance on real-world tasks. However, existing evaluation datasets are limited in scope and primarily text-based, lacking a unified and standardized multimodal question-answering (QA) benchmark. To address this issue, we introduce TCM-Ladder, the first multimodal QA dataset specifically designed for evaluating large TCM language models. The dataset spans multiple core disciplines of TCM, including fundamental theory, diagnostics, herbal formulas, internal medicine, surgery, pharmacognosy, and pediatrics. In addition to textual content, TCM-Ladder incorporates various modalities such as images and videos. The datasets were constructed using a combination of automated and manual filtering processes and comprise 52,000+ questions in total. These questions include single-choice, multiple-choice, fill-in-the-blank, diagnostic dialogue, and visual comprehension tasks. We trained a reasoning model on TCM-Ladder and conducted comparative experiments against 9 state-of-the-art general domain and 5 leading TCM-specific LLMs to evaluate their performance on the datasets. Moreover, we propose Ladder-Score, an evaluation method specifically designed for TCM question answering that effectively assesses answer quality regarding terminology usage and semantic expression. To our knowledge, this is the first work to evaluate mainstream general domain and TCM-specific LLMs on a unified multimodal benchmark. The datasets and leaderboard are publicly available at https://tcmladder.com or https://54.211.107.106 and will be continuously updated.
With the increasing integration of visual and textual content in Social Networking Services (SNS), evaluating the multimodal capabilities of Large Language Models (LLMs) is crucial for enhancing user experience, content understanding, and platform intelligence. Existing benchmarks primarily focus on text-centric tasks, lacking coverage of the multimodal contexts prevalent in modern SNS ecosystems. In this paper, we introduce SNS-Bench-VL, a comprehensive multimodal benchmark designed to assess the performance of Vision-Language LLMs in real-world social media scenarios. SNS-Bench-VL incorporates images and text across 8 multimodal tasks, including note comprehension, user engagement analysis, information retrieval, and personalized recommendation. It comprises 4,001 carefully curated multimodal question-answer pairs, covering single-choice, multiple-choice, and open-ended tasks. We evaluate over 25 state-of-the-art multimodal LLMs, analyzing their performance across tasks. Our findings highlight persistent challenges in multimodal social context comprehension. We hope SNS-Bench-VL will inspire future research towards robust, context-aware, and human-aligned multimodal intelligence for next-generation social networking services.
Open benchmarks are essential for evaluating and advancing large language models, offering reproducibility and transparency. However, their accessibility makes them likely targets of test set contamination. In this work, we introduce DyePack, a framework that leverages backdoor attacks to identify models that used benchmark test sets during training, without requiring access to the loss, logits, or any internal details of the model. Like how banks mix dye packs with their money to mark robbers, DyePack mixes backdoor samples with the test data to flag models that trained on it. We propose a principled design incorporating multiple backdoors with stochastic targets, enabling exact false positive rate (FPR) computation when flagging every model. This provably prevents false accusations while providing strong evidence for every detected case of contamination. We evaluate DyePack on five models across three datasets, covering both multiple-choice and open-ended generation tasks. For multiple-choice questions, it successfully detects all contaminated models with guaranteed FPRs as low as 0.000073% on MMLU-Pro and 0.000017% on Big-Bench-Hard using eight backdoors. For open-ended generation tasks, it generalizes well and identifies all contaminated models on Alpaca with a guaranteed false positive rate of just 0.127% using six backdoors.
Spatial intelligence is essential for multimodal large language models (MLLMs) operating in the complex physical world. Existing benchmarks, however, probe only single-image relations and thus fail to assess the multi-image spatial reasoning that real-world deployments demand. We introduce MMSI-Bench, a VQA benchmark dedicated to multi-image spatial intelligence. Six 3D-vision researchers spent more than 300 hours meticulously crafting 1,000 challenging, unambiguous multiple-choice questions from over 120,000 images, each paired with carefully designed distractors and a step-by-step reasoning process. We conduct extensive experiments and thoroughly evaluate 34 open-source and proprietary MLLMs, observing a wide gap: the strongest open-source model attains roughly 30% accuracy and OpenAI's o3 reasoning model reaches 40%, while humans score 97%. These results underscore the challenging nature of MMSI-Bench and the substantial headroom for future research. Leveraging the annotated reasoning processes, we also provide an automated error analysis pipeline that diagnoses four dominant failure modes, including (1) grounding errors, (2) overlap-matching and scene-reconstruction errors, (3) situation-transformation reasoning errors, and (4) spatial-logic errors, offering valuable insights for advancing multi-image spatial intelligence. Project page: https://runsenxu.com/projects/MMSI_Bench .
Video Anomaly Understanding (VAU) is essential for applications such as smart cities, security surveillance, and disaster alert systems, yet remains challenging due to its demand for fine-grained spatio-temporal perception and robust reasoning under ambiguity. Despite advances in anomaly detection, existing methods often lack interpretability and struggle to capture the causal and contextual aspects of abnormal events. This limitation is further compounded by the absence of comprehensive benchmarks for evaluating reasoning ability in anomaly scenarios. To address both challenges, we introduce VAU-R1, a data-efficient framework built upon Multimodal Large Language Models (MLLMs), which enhances anomaly reasoning through Reinforcement Fine-Tuning (RFT). Besides, we propose VAU-Bench, the first Chain-of-Thought benchmark tailored for video anomaly reasoning, featuring multiple-choice QA, detailed rationales, temporal annotations, and descriptive captions. Empirical results show that VAU-R1 significantly improves question answering accuracy, temporal grounding, and reasoning coherence across diverse contexts. Together, our method and benchmark establish a strong foundation for interpretable and reasoning-aware video anomaly understanding. Our code is available at https://github.com/GVCLab/VAU-R1.
Multimodal Large Language Models (MLLMs) are of great application across many domains, such as multimodal understanding and generation. With the development of diffusion models (DM) and unified MLLMs, the performance of image generation has been significantly improved, however, the study of image screening is rare and its performance with MLLMs is unsatisfactory due to the lack of data and the week image aesthetic reasoning ability in MLLMs. In this work, we propose a complete solution to address these problems in terms of data and methodology. For data, we collect a comprehensive medical image screening dataset with 1500+ samples, each sample consists of a medical image, four generated images, and a multiple-choice answer. The dataset evaluates the aesthetic reasoning ability under four aspects: \textit{(1) Appearance Deformation, (2) Principles of Physical Lighting and Shadow, (3) Placement Layout, (4) Extension Rationality}. For methodology, we utilize long chains of thought (CoT) and Group Relative Policy Optimization with Dynamic Proportional Accuracy reward, called DPA-GRPO, to enhance the image aesthetic reasoning ability of MLLMs. Our experimental results reveal that even state-of-the-art closed-source MLLMs, such as GPT-4o and Qwen-VL-Max, exhibit performance akin to random guessing in image aesthetic reasoning. In contrast, by leveraging the reinforcement learning approach, we are able to surpass the score of both large-scale models and leading closed-source models using a much smaller model. We hope our attempt on medical image screening will serve as a regular configuration in image aesthetic reasoning in the future.