Role-playing enables large language models (LLMs) to engage users in immersive and personalized interactions, but it also introduces significant safety risks. Existing role-play fine-tuning techniques improve role adaptability but may degrade safety performance, particularly for villainous characters. In this work, we conduct the first comprehensive assessment of role-play fine-tuning risks by training 95 role-specific LLMs using RoleBench. Our experiments reveal that role-play fine-tuning leads to a noticeable decline in safety performance, with safety risks varying based on character traits. To tackle this challenge, we propose Safety-Aware Role-Play Fine-Tuning (SaRFT), a novel method designed to balance role-playing capabilities and safety. Extensive experiments on LLaMA-3-8B-Instruct, Gemma-2-9B-it, and Qwen2.5-7B-Instruct demonstrate that SaRFT consistently outperforms state-of-the-art baselines under both LoRA and full-parameter fine-tuning settings. Our findings highlight the necessity of role-adaptive safety measures and provide insights into mitigating role-specific safety risks in role-playing LLMs.

Visual Question Answering (VQA) focuses on providing answers to natural language questions by utilizing information from images. Although cutting-edge multimodal large language models (MLLMs) such as GPT-4o achieve strong performance on VQA tasks, they frequently fall short in accessing domain-specific or the latest knowledge. To mitigate this issue, retrieval-augmented generation (RAG) leveraging external knowledge bases (KBs), referred to as KB-VQA, emerges as a promising approach. Nevertheless, conventional unimodal retrieval techniques, which translate images into textual descriptions, often result in the loss of critical visual details. This study presents fine-grained knowledge units, which merge textual snippets with entity images stored in vector databases. Furthermore, we introduce a knowledge unit retrieval-augmented generation framework (KU-RAG) that integrates fine-grained retrieval with MLLMs. The proposed KU-RAG framework ensures precise retrieval of relevant knowledge and enhances reasoning capabilities through a knowledge correction chain. Experimental findings demonstrate that our approach significantly boosts the performance of leading KB-VQA methods, achieving improvements of up to 10%.

With the widespread application of Large Language Models across various domains, their security issues have increasingly garnered significant attention from both academic and industrial communities. This study conducts sampling and normalization of the parameters of the LLM to generate visual representations and heatmaps of parameter distributions, revealing notable discrepancies in parameter distributions among certain layers within the hidden layers. Further analysis involves calculating statistical metrics for each layer, followed by the computation of a Comprehensive Sensitivity Score based on these metrics, which identifies the lower layers as being particularly sensitive to the generation of harmful content. Based on this finding, we employ a Freeze training strategy, selectively performing Supervised Fine-Tuning only on the lower layers. Experimental results demonstrate that this method significantly reduces training duration and GPU memory consumption while maintaining a high jailbreak success rate and a high harm score, outperforming the results achieved by applying the LoRA method for SFT across all layers. Additionally, the method has been successfully extended to other open-source large models, validating its generality and effectiveness across different model architectures. Furthermore, we compare our method with ohter jailbreak method, demonstrating the superior performance of our approach. By innovatively proposing a method to statistically analyze and compare large model parameters layer by layer, this study provides new insights into the interpretability of large models. These discoveries emphasize the necessity of continuous research and the implementation of adaptive security measures in the rapidly evolving field of LLMs to prevent potential jailbreak attack risks, thereby promoting the development of more robust and secure LLMs.

The growing popularity and widespread adoption of large language models (LLMs) necessitates the development of tools that enhance the effectiveness of user interactions with these models. Understanding the structures and functions of these models poses a significant challenge for users. Visual analytics-driven tools enables users to explore and compare, facilitating better decision-making. This paper presents a visual analytics-driven tool equipped with interactive controls for key hyperparameters, including top-p, frequency and presence penalty, enabling users to explore, examine and compare the outputs of LLMs. In a user study, we assessed the tool's effectiveness, which received favorable feedback for its visual design, with particular commendation for the interface layout and ease of navigation. Additionally, the feedback provided valuable insights for enhancing the effectiveness of Human-LLM interaction tools.

We present WebFAQ, a large-scale collection of open-domain question answering datasets derived from FAQ-style schema.org annotations. In total, the data collection consists of 96 million natural question-answer (QA) pairs across 75 languages, including 47 million (49%) non-English samples. WebFAQ further serves as the foundation for 20 monolingual retrieval benchmarks with a total size of 11.2 million QA pairs (5.9 million non-English). These datasets are carefully curated through refined filtering and near-duplicate detection, yielding high-quality resources for training and evaluating multilingual dense retrieval models. To empirically confirm WebFAQ's efficacy, we use the collected QAs to fine-tune an in-domain pretrained XLM-RoBERTa model. Through this process of dataset-specific fine-tuning, the model achieves significant retrieval performance gains, which generalize - beyond WebFAQ - to other multilingual retrieval benchmarks evaluated in zero-shot setting. Last but not least, we utilize WebFAQ to construct a set of QA-aligned bilingual corpora spanning over 1000 language pairs using state-of-the-art bitext mining and automated LLM-assessed translation evaluation. Due to our advanced, automated method of bitext dataset generation, the resulting bilingual corpora demonstrate higher translation quality compared to similar datasets. WebFAQ and all associated resources are publicly available on GitHub and HuggingFace.

Real-time video segmentation is a promising feature for AI-assisted surgery, providing intraoperative guidance by identifying surgical tools and anatomical structures. However, deploying state-of-the-art segmentation models, such as SAM2, in real-time settings is computationally demanding, which makes it essential to balance frame rate and segmentation performance. In this study, we investigate the impact of frame rate on zero-shot surgical video segmentation, evaluating SAM2's effectiveness across multiple frame sampling rates for cholecystectomy procedures. Surprisingly, our findings indicate that in conventional evaluation settings, frame rates as low as a single frame per second can outperform 25 FPS, as fewer frames smooth out segmentation inconsistencies. However, when assessed in a real-time streaming scenario, higher frame rates yield superior temporal coherence and stability, particularly for dynamic objects such as surgical graspers. Finally, we investigate human perception of real-time surgical video segmentation among professionals who work closely with such data and find that respondents consistently prefer high FPS segmentation mask overlays, reinforcing the importance of real-time evaluation in AI-assisted surgery.

With reasoning language models such as OpenAI-o3 and DeepSeek-R1 emerging, large language models (LLMs) have entered a new phase of development. However, existing benchmarks for coding evaluation are gradually inadequate to assess the capability of advanced LLMs in code reasoning. To bridge the gap for high-level code reasoning assessment, we propose ProBench to benchmark LLMs in competitive programming, drawing inspiration from the International Collegiate Programming Contest. ProBench collects a comprehensive set of competitive programming problems from Codeforces, Luogu, and Nowcoder platforms during the period from July to December 2024, obtaining real test results through online submissions to ensure the fairness and accuracy of the evaluation. We establish a unified problem attribute system, including difficulty grading and algorithm tagging. With carefully collected and annotated data in ProBench, we systematically assess 9 latest LLMs in competitive programming across multiple dimensions, including thought chain analysis, error type diagnosis, and reasoning depth evaluation. Experimental results show that QwQ-32B-Preview achieves the best score of 20.93 followed by DeepSeek-V3 with a score of 16.38, suggesting that models trained with specialized reasoning tasks significantly outperform general-purpose models (even larger than reasoning-oriented models) in programming. Further analysis also reveals key areas for programming capability enhancement, e.g., algorithm adaptability and reasoning sufficiency, providing important insights for the future development of reasoning models.

While LLMs excel in zero-shot tasks, their performance in linguistic challenges like syntactic parsing has been less scrutinized. This paper studies state-of-the-art open-weight LLMs on the task by comparing them to baselines that do not have access to the input sentence, including baselines that have not been used in this context such as random projective trees or optimal linear arrangements. The results show that most of the tested LLMs cannot outperform the best uninformed baselines, with only the newest and largest versions of LLaMA doing so for most languages, and still achieving rather low performance. Thus, accurate zero-shot syntactic parsing is not forthcoming with open LLMs.

The Javanese language features a complex system of honorifics that vary according to the social status of the speaker, listener, and referent. Despite its cultural and linguistic significance, there has been limited progress in developing a comprehensive corpus to capture these variations for natural language processing (NLP) tasks. In this paper, we present Unggah-Ungguh, a carefully curated dataset designed to encapsulate the nuances of Unggah-Ungguh Basa, the Javanese speech etiquette framework that dictates the choice of words and phrases based on social hierarchy and context. Using Unggah-Ungguh, we assess the ability of language models (LMs) to process various levels of Javanese honorifics through classification and machine translation tasks. To further evaluate cross-lingual LMs, we conduct machine translation experiments between Javanese (at specific honorific levels) and Indonesian. Additionally, we explore whether LMs can generate contextually appropriate Javanese honorifics in conversation tasks, where the honorific usage should align with the social role and contextual cues. Our findings indicate that current LMs struggle with most honorific levels, exhibitinga bias toward certain honorific tiers.

Large language models (LLMs) excel in both closed tasks (including problem-solving, and code generation) and open tasks (including creative writing), yet existing explanations for their capabilities lack connections to real-world human intelligence. To fill this gap, this paper systematically investigates LLM intelligence through the lens of ``human simulation'', addressing three core questions: (1) How do personality traits affect problem-solving in closed tasks? (2) How do traits shape creativity in open tasks? (3) How does single-agent performance influence multi-agent collaboration? By assigning Big Five personality traits to LLM agents and evaluating their performance in single- and multi-agent settings, we reveal that specific traits significantly influence reasoning accuracy (closed tasks) and creative output (open tasks). Furthermore, multi-agent systems exhibit collective intelligence distinct from individual capabilities, driven by distinguishing combinations of personalities. We demonstrate that LLMs inherently simulate human behavior through next-token prediction, mirroring human language, decision-making, and collaborative dynamics.