With the rapid development of large language models (LLMs), LLM-as-a-judge has emerged as a widely adopted approach for text quality evaluation, including hallucination evaluation. While previous studies have focused exclusively on single-context evaluation (e.g., discourse faithfulness or world factuality), real-world hallucinations typically involve mixed contexts, which remains inadequately evaluated. In this study, we use summarization as a representative task to comprehensively evaluate LLMs’ capability in detecting mixed-context hallucinations, specifically distinguishing between factual and non-factual hallucinations. Through extensive experiments across direct generation and retrieval-based models of varying scales, our main observations are: (1) LLMs’ intrinsic knowledge introduces inherent biases in hallucination evaluation; (2) These biases particularly impact the detection of factual hallucinations, yielding a significant performance bottleneck; and (3) the fundamental challenge lies in effective knowledge utilization, balancing between LLMs’ intrinsic knowledge and external context for accurate mixed-context hallucination evaluation.
论文检索
输入标题、作者或关键词，从 3,086 篇学术成果中精准定位
Knowledge-intensive multi-hop question answering (QA) tasks, which require integrating evidence from multiple sources to address complex queries, often necessitate multiple rounds of retrieval and iterative generation by large language models (LLMs). However, incorporating many documents and extended contexts poses challenges—such as hallucinations and semantic drift—for lightweight LLMs with fewer parameters. This work proposes a novel framework called DEC (Dynamic Enhancement Chain). DEC first decomposes complex questions into logically coherent subquestions to form a hallucination-free reasoning chain. It then iteratively refines these subquestions through context-aware rewriting to generate effective query formulations. For retrieval, we introduce a lightweight discriminative keyword extraction module that leverages extracted keywords to achieve targeted, precise document recall with relatively low computational overhead. Extensive experiments on three multi-hop QA datasets demonstrate that DEC performs on par with or surpasses state-of-the-art benchmarks while significantly reducing token consumption. Notably, our approach attains state-of-the-art results on models with 8B parameters, showcasing its effectiveness in various scenarios, particularly in resource-constrained environments.
Automatic radiology report generation holds significant potential to streamline the labor-intensive process of report writing by radiologists, particularly for 3D radiographs such as CT scans. While CT scans are critical for clinical diagnostics, they remain less explored compared to 2D radiographs. To date, there has been no comprehensive benchmark for 3D radiograph report generation (3DRRG), nor sufficient investigation into the optimal training strategies for Vision Language Models (VLMs) in this context, particularly with respect to vision encoder choices, visual token compression, and model scaling.In this work, we make two three contributions. We curate CT-3DRRG, the largest publicly available 3D CT-report dataset, establishing a robust and diverse benchmark for evaluating VLM performance on 3DRRG. Furthermore, we propose a comprehensive training recipe for building high-performing VLMs for 3DRRG, exploring key factors such as vision encoder pretraining strategies, visual token compression, and the impact of data & model scale. Guided by these findings, we introduce Argus, a state-of-the-art family of VLMs that achieve superior performance across different model sizes and input 3D medical image resolutions, efficiently processing high-resolution 3D images up to 512 × 512 × 256.
See the World, Discover Knowledge: A Chinese Factuality Evaluation for Large Vision Language Models
PDF ↗The evaluation of factual accuracy in large vision language models (LVLMs) has lagged behind their rapid development, making it challenging to fully reflect these models’ knowledge capacity and reliability. In this paper, we introduce the first factuality-based visual question-answering benchmark in Chinese, named ChineseSimpleVQA, aimed at assessing the visual factuality of LVLMs across 8 major topics and 56 subtopics. The key features of this benchmark include a focus on the Chinese language, diverse knowledge types, a multi-hop question construction, high-quality data, static consistency, and easy-to-evaluate through short answers. Moreover, we contribute a rigorous data construction pipeline and decouple the visual factuality into two parts: seeing the world (i.e., object recognition) and discovering knowledge. This decoupling allows us to analyze the capability boundaries and execution mechanisms of LVLMs. Subsequently, we evaluate 34 advanced open-source and closed-source models, revealing critical performance gaps within this field.
Medical Vision-Language Pre-training (MedVLP) has made significant progress in enabling zero-shot tasks for medical image understanding. However, training MedVLP models typically requires large-scale datasets with paired, high-quality image-text data, which are scarce in the medical domain. Recent advancements in Large Language Models (LLMs) and diffusion models have made it possible to generate large-scale synthetic image-text pairs. This raises the question: Can MedVLP succeed using purely synthetic data? To address this, we use off-the-shelf generative models to create synthetic radiology reports and paired Chest X-ray (CXR) images, and propose an automated pipeline to build a diverse, high-quality synthetic dataset, enabling a rigorous study that isolates model and training settings, focusing entirely from the data perspective.Our results show that MedVLP models trained exclusively on synthetic data outperform those trained on real data by 3.8% in averaged AUC on zero-shot classification. Moreover, using a combination of synthetic and real data leads to a further improvement of 9.07%. Additionally, MedVLP models trained on synthetic or mixed data consistently outperform those trained on real data in zero-shot grounding, as well as in fine-tuned classification and segmentation tasks.Our analysis suggests MedVLP trained on well-designed synthetic data can outperform models trained on real datasets, which may be limited by low-quality samples and long-tailed distributions[^1].[^1]: All data and code will be released upon acceptance.
This paper introduces the Balanced Arabic Readability Evaluation Corpus (BAREC), a large-scale, fine-grained dataset for Arabic readability assessment. BAREC consists of 69,441 sentences spanning 1+ million words, carefully curated to cover 19 readability levels, from kindergarten to postgraduate comprehension. The corpus balances genre diversity, topical coverage, and target audiences, offering a comprehensive resource for evaluating Arabic text complexity. The corpus was fully manually annotated by a large team of annotators. The average pairwise inter-annotator agreement, measured by Quadratic Weighted Kappa, is 81.8%, reflecting a high level of substantial agreement.Beyond presenting the corpus, we benchmark automatic readability assessment across different granularity levels, comparing a range of techniques. Our results highlight the challenges and opportunities in Arabic readability modeling, demonstrating competitive performance across various methods.To support research and education, we make BAREC openly available, along with detailed annotation guidelines and benchmark results: http://barec.camel-lab.com.
Large language models (LLMs) have significantly advanced natural language processing, particularly through the integration of external tools and APIs. However, their effectiveness is frequently hampered by parameter mis-filling during tool calling. In this paper, we propose the Hierarchical Tool Error Checklist (HiTEC) framework to systematically diagnose and mitigate tool-calling errors without relying on extensive real-world interactions. HiTEC introduces a two-tiered approach: a global error checklist that identifies common, cross-tool issues, and a local error checklist that targets tool-specific and contextual failures. Building on this structure, we propose two deployments: HiTEC-In Context Learning (HiTEC-ICL) and HiTEC-Kahneman-Tversky Optimization (HiTEC-KTO). HiTEC-ICL embeds the global checklist in the initial prompts and leverages a two-round conversational interaction to dynamically refine parameter handling, while HiTEC-KTO generates high-quality negative examples to drive fine-tuning via preference-based optimization. Extensive experiments across five public datasets demonstrate that our framework significantly improves parameter-filling accuracy and tool-calling success rates compared to baseline methods.
SynGraph: A Dynamic Graph-LLM Synthesis Framework for Sparse Streaming User Sentiment Modeling
PDF ↗User reviews on e-commerce platforms exhibit dynamic sentiment patterns driven by temporal and contextual factors. Traditional sentiment analysis methods focus on static reviews, failing to capture the evolving temporal relationship between user sentiment rating and textual content. Sentiment analysis on streaming reviews addresses this limitation by modeling and predicting the temporal evolution of user sentiments. However, it suffers from data sparsity, manifesting in temporal, spatial, and combined forms. In this paper, we introduce SynGraph, a novel framework designed to address data sparsity in sentiment analysis on streaming reviews. SynGraph alleviates data sparsity by categorizing users into mid-tail, long-tail, and extreme scenarios and incorporating LLM-augmented enhancements within a dynamic graph-based structure. Experiments on real-world datasets demonstrate its effectiveness in addressing sparsity and improving sentiment modeling in streaming reviews.
Large Language Models (LLMs) have made remarkable advances in role-playing dialogue agents, demonstrating their utility in character simulations. However, it remains challenging for these agents to balance character portrayal utility with content safety because this essential character simulation often comes with the risk of generating unsafe content. To address this issue, we first conduct a systematic exploration of the safety-utility trade-off across multiple LLMs. Our analysis reveals that risk scenarios created by villain characters and user queries (referred to as risk coupling) contribute to this trade-off. Building on this, we propose a novel Adaptive Dynamic Multi-Preference (ADMP) method, which dynamically adjusts safety-utility preferences based on the degree of risk coupling and guides the model to generate responses biased toward utility or safety. We further introduce Coupling Margin Sampling (CMS) into coupling detection to enhance the model’s ability to handle high-risk scenarios. Experimental results demonstrate that our approach improves safety metrics while maintaining utility.
Existing multilingual vision-language (VL) benchmarks often only cover a handful of languages. Consequently, evaluations of large vision-language models (LVLMs) predominantly target high-resource languages, underscoring the need for evaluation data for low-resource languages. To address this limitation, we introduce MVL-SIB, a massively multilingual vision-language benchmark that evaluates both cross-modal and text-only topical matching across 205 languages – over 100 more than the most multilingual existing VL benchmarks encompass. We then benchmark a range of of open-weight LVLMs together with GPT-4o(-mini) on MVL-SIB. Our results reveal that LVLMs struggle in cross-modal topic matching in lower-resource languages, performing no better than chance on languages like N’Koo. Our analysis further reveals that VL support in LVLMs declines disproportionately relative to textual support for lower-resource languages, as evidenced by comparison of cross-modal and text-only topical matching performance. We further observe that open-weight LVLMs do not benefit from representing a topic with more than one image, suggesting that these models are not yet fully effective at handling multi-image tasks. By correlating performance on MVL-SIB with other multilingual VL benchmarks, we highlight that MVL-SIB serves as a comprehensive probe of multilingual VL understanding in LVLMs.
Grounded natural language understanding in Human-Robot Interaction (HRI) requires integrating linguistic, visual, and world knowledge to ensure effective task execution. We propose an approach that enhances Multi-Modal Large Language Models (MLLMs) with a novel explicit dialogue planning phase, allowing robotic agents to systematically refine their understanding of ambiguous commands through structured clarification steps. This reduces hallucinations and improves task feasibility.To evaluate this approach, we introduce a novel dataset of over 1,100 annotated dialogues in English and Italian, designed for fine-tuning and assessing Multi-Modal models in HRI scenarios. Experimental results show that dialogue planning improves response accuracy and quality, and contributes to cross-lingual generalisation, enabling models trained in one language to transfer effectively to another. To the best of our knowledge, this is the first application of structured, goal-driven, and explicit dialogue planning in Multi-Modal LLMs for grounded interaction.
Large Language Models (LLMs) have shown promise in automating high-labor data tasks, but the adoption of LLMs in high-stake scenarios faces two key challenges: their tendency to answer despite uncertainty and their difficulty handling long input contexts robustly.We investigate commonly used off-the-shelf LLMs’ ability to identify low-confidence outputs for human review through “check set selection”–a process where LLMs prioritize information needing human judgment.Using a case study on social media monitoring for disaster risk management,we define the “check set” as a list of tweets escalated to the disaster manager when the LLM has the least confidence, enabling human oversight within budgeted effort.We test two strategies for LLM check set selection: *individual confidence elicitation* – LLMs assesses confidence for each tweet classification individually, requiring more prompts with shorter contexts, and *direct set confidence elicitation* – LLM evaluates confidence for a list of tweet classifications at once, using less prompts but longer contexts.Our results reveal that set selection via individual probabilities is more reliable but that direct set confidence merits further investigation.Direct set selection challenges include inconsistent outputs, incorrect check set size, and low inter-annotator agreement. Despite these challenges, our approach improves collaborative disaster tweet classification by outperforming random-sample check set selection, demonstrating the potential of human-LLM collaboration.
Large Language Models (LLMs) excel in zero-shot and few-shot tasks, but achieving similar performance with encoder-only models like BERT and RoBERTa has been challenging due to their architecture. However, encoders offer advantages such as lower computational and memory costs. Recent work adapts them for zero-shot generalization using Statement Tuning, which reformulates tasks into finite templates. We extend this approach to multilingual NLP, exploring whether encoders can achieve zero-shot cross-lingual generalization and serve as efficient alternatives to memory-intensive LLMs for low-resource languages. Our results show that state-of-the-art encoder models generalize well across languages, rivaling multilingual LLMs while being more efficient. We also analyze multilingual Statement Tuning dataset design, efficiency gains, and language-specific generalization, contributing to more inclusive and resource-efficient NLP models. We release our code and models.
Commonsense, humans’ implicit understanding of everyday situations, is crucial for large language models (LLMs). Existing commonsense evaluations for LLMs primarily focus on downstream knowledge tasks, failing to probe whether LLMs truly understand and utilize knowledge or merely memorize it. They also rely heavily on human annotation and lack automated large-scale data generation. To address this, we propose to automatically construct a large benchmark named CoCo (Consistency of Commonsense) comprising 39K samples derived from commonsense knowledge graphs (CSKGs), paired with symbolic questions and ground-truth answers, which systematically assesses LLMs’ knowledge memorization, comprehension, and application and examines the consistency between these tasks. To enhance our evaluation, we also propose novel metrics and prompting strategies. Experimental results on multiple LLMs reveal that CoCo presents significant challenges, and our detailed analysis provides deeper insights into the strengths and limitations of LLMs’ commonsense abilities.
To address these limitations, we propose BDC, a novel framework that Boosts reasoning exploration via multi-agent collaboration, Disentangles heterogeneous data into specialized experts, and Customizes solutions through dynamic model composition. BDC integrates a Monte Carlo Tree-of-Agents algorithm, where multiple LLMs mutually verify and refine reasoning paths through reflection-guided pruning, enabling efficient exploration of high-quality solutions. To handle data diversity, we cluster problems by latent semantics, train composable LoRA experts on each cluster, and deploy an input-aware hypernetwork to dynamically merge these experts into tailored solvers. Experiments on APPS and CodeContest benchmarks demonstrate BDC’s superiority: it achieves up to 73.8% accuracy on hard problems, outperforming state-of-the-art methods like LATS and RethinkMCTS by 9–15%. This work lays the groundwork for advancing LLM capabilities in complex reasoning tasks, offering a novel System2-to-System1 solution.
Semantic Aware Linear Transfer by Recycling Pre-trained Language Models for Cross-lingual Transfer
PDF ↗Large Language Models (LLMs) are increasingly incorporating multilingual capabilities, fueling the demand to transfer them into target language-specific models. However, most approaches, which blend the source model’s embedding by replacing the source vocabulary with the target language-specific vocabulary, may constrain expressive capacity in the target language since the source model is predominantly trained on English data. In this paper, we propose Semantic Aware Linear Transfer (SALT), a novel cross-lingual transfer technique that recycles embeddings from target language Pre-trained Language Models (PLMs) to transmit the deep representational strengths of PLM-derived embedding to LLMs. SALT derives unique regression lines based on the similarity in the overlap of the source and target vocabularies to handle each non-overlapping token’s embedding space. Our extensive experiments show that SALT significantly outperforms other transfer methods, achieving lower loss and faster convergence during language adaptation. Notably, SALT achieves remarkable performance in cross-lingual understanding setups compared to other methods. Furthermore, we highlight the scalable use of PLMs to enhance the functionality of contemporary LLMs by conducting experiments with varying architectures.
Using large language models (LLMs) has a potential risk of privacy leakage since the data with sensitive information may be used for fine-tuning the LLMs. Differential privacy (DP) provides theoretical guarantees of privacy protection, but its practical application in LLMs still has the problem of privacy-utility trade-off. Researchers synthesized data with strong generation capabilities closed-source LLMs (i.e., GPT-4) under DP to alleviate this problem, but this method is not so flexible in fitting the given privacy distributions without fine-tuning. Besides, such methods can hardly balance the diversity of synthetic data and its relevance to target privacy data without accessing so much private data. To this end, this paper proposes DPGA-TextSyn, combining general LLMs with genetic algorithm (GA) to produce relevant and diverse synthetic text under DP constraints. First, we integrate the privacy gene (i.e., metadata) to generate better initial samples. Then, to achieve survival of the fittest and avoid homogeneity, we use privacy nearest neighbor voting and similarity suppression to select elite samples. In addition, we expand elite samples via genetic strategies such as mutation, crossover, and generation to expand the search scope of GA. Experiments show that this method significantly improves the performance of the model in downstream tasks while ensuring privacy.
This paper poses two critical issues in evaluating base models (without post-training): (1) Unstable evaluation during training: in the early stages of pre-training, the models lack the capability to answer questions as required, leading to unstable evaluation results. This instability makes it difficult to provide solid conclusions to guide the training, especially for key experiments such as data ablation and scaling law. (2) Inconsistency between base and instruct models: base models generally exhibit poorer evaluation performance compared to corresponding instruct models. This gap poses a challenge for assessing whether a base model with better evaluation can truly lead to a better instruct model. To address these issues, we propose **B**ase model **O**riented **S**ystematic **E**valuation (**BOSE**), a method specifically designed to optimize the evaluation of base models. Specifically, BOSE introduces two key innovations: In-Context Light-instruction Prompt (**ICLiP**) for open-ended tasks and **Blank-ppl** for multi-choice tasks with candidate options, which transforms the standard perplexity (ppl) metric into a fill-in-the-blank format to mitigate early-stage evaluation fluctuations. Furthermore, we are the first to propose Kendall’s rank correlation to quantitatively measure the evaluation stability and consistency. Experimental results demonstrate that BOSE significantly enhances both the stability of evaluations during pre-training and the consistency between base and instruct models, thereby providing more reliable guidance for the LLMs’ training.
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, the 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.
We present Sequential Policy Optimization for Simultaneous Machine Translation (SeqPO-SiMT), a new policy optimization framework that defines the simultaneous machine translation (SiMT) task as a sequential decision making problem, incorporating a tailored reward to enhance translation quality while reducing latency. In contrast to popular Reinforcement Learning from Human Feedback (RLHF) methods, such as PPO and DPO, which are typically applied in single-step tasks, SeqPO-SiMT effectively tackles the multi-step SiMT task. This intuitive framework allows the SiMT LLMs to simulate and refine the SiMT process using a tailored reward. We conduct experiments on six datasets from diverse domains for En → Zh and Zh → En SiMT tasks, demonstrating that SeqPO-SiMT consistently achieves significantly higher translation quality with lower latency. In particular, SeqPO-SiMT outperforms the supervised fine-tuning (SFT) model by 1.13 points in COMET, while reducing the Average Lagging by 6.17 in the NEWSTEST2021 En → Zh dataset. While SiMT operates with far less context than offline translation, the SiMT results of SeqPO-SiMT on 7B LLM surprisingly rival the offline translation of high-performing LLMs, including Qwen-2.5-7B-Instruct and LLaMA-3-8B-Instruct.