Hallucinations are a common issue that undermine the reliability of large language models (LLMs). Recent studies have identified a specific subset of hallucinations, known as confabulations, which arise due to predictive uncertainty of LLMs. To detect confabulations, various methods for estimating predictive uncertainty in natural language generation (NLG) have been developed. These methods are typically evaluated by correlating uncertainty estimates with the correctness of generated text, with question-answering (QA) datasets serving as the standard benchmark. However, commonly used approximate correctness functions have substantial disagreement between each other and, consequently, in the ranking of the uncertainty estimation methods. This allows one to inflate the apparent performance of uncertainty estimation methods. We propose using several alternative risk indicators for risk correlation experiments that improve robustness of empirical assessment of UE algorithms for NLG. For QA tasks, we show that marginalizing over multiple LLM-as-a-judge variants leads to reducing the evaluation biases. Furthermore, we explore structured tasks as well as out of distribution and perturbation detection tasks which provide robust and controllable risk indicators. Finally, we propose to use an Elo rating of uncertainty estimation methods to give an objective summarization over extensive evaluation settings.

Drag-based image editing has long suffered from distortions in the target region, largely because the priors of earlier base models, Stable Diffusion, are insufficient to project optimized latents back onto the natural image manifold. With the shift from UNet-based DDPMs to more scalable DiT with flow matching (e.g., SD3.5, FLUX), generative priors have become significantly stronger, enabling advances across diverse editing tasks. However, drag-based editing has yet to benefit from these stronger priors. This work proposes the first framework to effectively harness FLUX's rich prior for drag-based editing, dubbed DragFlow, achieving substantial gains over baselines. We first show that directly applying point-based drag editing to DiTs performs poorly: unlike the highly compressed features of UNets, DiT features are insufficiently structured to provide reliable guidance for point-wise motion supervision. To overcome this limitation, DragFlow introduces a region-based editing paradigm, where affine transformations enable richer and more consistent feature supervision. Additionally, we integrate pretrained open-domain personalization adapters (e.g., IP-Adapter) to enhance subject consistency, while preserving background fidelity through gradient mask-based hard constraints. Multimodal large language models (MLLMs) are further employed to resolve task ambiguities. For evaluation, we curate a novel Region-based Dragging benchmark (ReD Bench) featuring region-level dragging instructions. Extensive experiments on DragBench-DR and ReD Bench show that DragFlow surpasses both point-based and region-based baselines, setting a new state-of-the-art in drag-based image editing. Code and datasets will be publicly available upon publication.

Existing safety evaluation methods for large language models (LLMs) suffer from inherent limitations, including evaluator bias and detection failures arising from model homogeneity, which collectively undermine the robustness of risk evaluation processes. This paper seeks to re-examine the risk evaluation paradigm by introducing a theoretical framework that reconstructs the underlying risk concept space. Specifically, we decompose the latent risk concept space into three mutually exclusive subspaces: the explicit risk subspace (encompassing direct violations of safety guidelines), the implicit risk subspace (capturing potential malicious content that requires contextual reasoning for identification), and the non-risk subspace. Furthermore, we propose RADAR, a multi-agent collaborative evaluation framework that leverages multi-round debate mechanisms through four specialized complementary roles and employs dynamic update mechanisms to achieve self-evolution of risk concept distributions. This approach enables comprehensive coverage of both explicit and implicit risks while mitigating evaluator bias. To validate the effectiveness of our framework, we construct an evaluation dataset comprising 800 challenging cases. Extensive experiments on our challenging testset and public benchmarks demonstrate that RADAR significantly outperforms baseline evaluation methods across multiple dimensions, including accuracy, stability, and self-evaluation risk sensitivity. Notably, RADAR achieves a 28.87% improvement in risk identification accuracy compared to the strongest baseline evaluation method. Large language models (LLMs) demonstrate considerable safety robustness across common language tasks, enabled by key techniques including alignment, defenses against prompt-based attacks, and dynamic safety monitoring (Trivedi et al., 2025; Zhang et al., 2025a; Y u et al., 2024; Xiong et al., 2024). Nevertheless, within AI governance contexts, LLMs remain susceptible to carefully crafted harmful requests ( e.g., jailbreak prompts), highlighting the critical need for robust safety evaluation frameworks.

Few-shot learning (FSL) aims to recognize novel concepts from only a few labeled support samples. Recent studies enhance support features by incorporating additional semantic information or designing complex semantic fusion modules. However, they still suffer from hallucinating semantics that contradict the visual evidence due to the lack of grounding in actual instances, resulting in noisy guidance and costly corrections. To address these issues, we propose a novel framework, bridging Vision and Text with LLMs for Few-Shot Learning (VT-FSL), which constructs precise cross-modal prompts conditioned on Large Language Models (LLMs) and support images, seamlessly integrating them through a geometry-aware alignment. It mainly consists of Cross-modal Iterative Prompting (CIP) and Cross-modal Geometric Alignment (CGA). Specifically, the CIP conditions an LLM on both class names and support images to generate precise class descriptions iteratively in a single structured reasoning pass. These descriptions not only enrich the semantic understanding of novel classes but also enable the zero-shot synthesis of semantically consistent images. The descriptions and synthetic images act respectively as complementary textual and visual prompts, providing high-level class semantics and low-level intra-class diversity to compensate for limited support data. Furthermore, the CGA jointly aligns the fused textual, support, and synthetic visual representations by minimizing the kernelized volume of the 3-dimensional parallelotope they span. It captures global and nonlinear relationships among all representations, enabling structured and consistent multimodal integration. The proposed VT-FSL method establishes new state-of-the-art performance across ten diverse benchmarks, including standard, cross-domain, and fine-grained few-shot learning scenarios. Code is available at https://github.com/peacelwh/VT-FSL.

Background: Currently, domain - specific large language models (LLMs) in traditional Chinese medicine (TCM) are primarily designed for clinical practice and medical education, yet they demonstrate substantial limitations when applied to research contexts owing to inadeq uate adaptability to complex tasks, thereby constraining their scientific utility. Moreover, the absence of comprehensive evaluation datasets and computational resource constraints hinder rigorous performance assessments and prevent extensive comparative o r ablation experiments, ultimately resulting in suboptimal model performance and weakened persuasiveness. Objective: To address these challenges, this study proposed a method for constructing a specialized LLM for the TCM domain based on contextual data integration and domain knowledge fusion and successfully developed a privatized LLM for the TCM profession, TianHui. Methods: Firstly, we acquired a large amount of TCM data, including academic literature resources, published book materials, online public data, and other supplementary materials, and pre - processed them to finally generate the 0.97G unsupervised dataset and 611312 QAs. Then, we adopted a phased training strategy (Pre - Training (PT) and Supervised Fine - Tuning (SFT)) and integrated three key technologies, Quantized Low - Rank Adaptation (QLoRA) parameter efficient fine - tuning, DeepSpeed Stage 2 distributed traini ng optimization, and Flash Attention 2 accelerated computation, to achieve optimal allocation of computational resources while guaranteeing training stability. Finally, we evaluated TianHui using 12 different types of benchmark test datasets and conducted extensive comparison experiments and ablation experiments. Results: The benchmark test data showed that TianHui demonstrated excellent performance in 12 TCM - related application scenarios. It ranked in the top three in each evaluation index in six test datasets: APQ, TCMCD, HFR, HCCA, DHPE, and TLAW. Meanwhile, it achieved optimal performance in all indicators of the six test data sets: TCMEE, APR, GCPMI, TCMKQA, TCMRC, and ADTG.

Legal documents are often long, dense, and difficult to comprehend, not only for laypeople but also for legal experts. While automated document summarization has great potential to improve access to legal knowledge, prevailing task-based evaluators overlook divergent user and stakeholder needs. Tool development is needed to encompass the technicality of a case summary for a litigator yet be accessible for a self-help public researching for their lawsuit. We introduce PersonaMatrix, a persona-by-criterion evaluation framework that scores summaries through the lens of six personas, including legal and non-legal users. We also introduce a controlled dimension-shifted pilot dataset of U.S. civil rights case summaries that varies along depth, accessibility, and procedural detail as well as Diversity-Coverage Index (DCI) to expose divergent optima of legal summary between persona-aware and persona-agnostic judges. This work enables refinement of legal AI summarization systems for both expert and non-expert users, with the potential to increase access to legal knowledge. The code base and data are publicly available in GitHub.

Speculative decoding is a widely adopted technique for accelerating inference in large language models (LLMs), yet its application to vision-language models (VLMs) remains underexplored, with existing methods achieving only modest speedups (<1.5x). This gap is increasingly significant as multimodal capabilities become central to large-scale models. We hypothesize that large VLMs can effectively filter redundant image information layer by layer without compromising textual comprehension, whereas smaller draft models struggle to do so. To address this, we introduce Vision-Aware Speculative Decoding (ViSpec), a novel framework tailored for VLMs. ViSpec employs a lightweight vision adaptor module to compress image tokens into a compact representation, which is seamlessly integrated into the draft model's attention mechanism while preserving original image positional information. Additionally, we extract a global feature vector for each input image and augment all subsequent text tokens with this feature to enhance multimodal coherence. To overcome the scarcity of multimodal datasets with long assistant responses, we curate a specialized training dataset by repurposing existing datasets and generating extended outputs using the target VLM with modified prompts. Our training strategy mitigates the risk of the draft model exploiting direct access to the target model's hidden states, which could otherwise lead to shortcut learning when training solely on target model outputs. Extensive experiments validate ViSpec, achieving, to our knowledge, the first substantial speedup in VLM speculative decoding. Code is available at https://github.com/KangJialiang/ViSpec.

Large language model agents increasingly depend on memory to sustain long horizon interaction, but existing frameworks remain limited. Most expose only a few basic primitives such as encode, retrieve, and delete, while higher order operations like merge, promote, demote, split, lock, and expire are missing or inconsistently supported. Moreover, there is no formal and executable specification for memory commands, leaving scope and lifecycle rules implicit and causing unpredictable behavior across systems. We introduce Text2Mem, a unified memory operation language that provides a standardized pathway from natural language to reliable execution. Text2Mem defines a compact yet expressive operation set aligned with encoding, storage, and retrieval. Each instruction is represented as a JSON based schema instance with required fields and semantic invariants, which a parser transforms into typed operation objects with normalized parameters. A validator ensures correctness before execution, while adapters map typed objects either to a SQL prototype backend or to real memory frameworks. Model based services such as embeddings or summarization are integrated when required. All results are returned through a unified execution contract. This design ensures safety, determinism, and portability across heterogeneous backends. We also outline Text2Mem Bench, a planned benchmark that separates schema generation from backend execution to enable systematic evaluation. Together, these components establish the first standardized foundation for memory control in agents.

Minority languages in China, such as Tibetan, Uyghur, and Traditional Mongolian, face significant challenges due to their unique writing systems, which differ from international standards. This discrepancy has led to a severe lack of relevant corpora, particularly for supervised tasks like headline generation. To address this gap, we introduce a novel dataset, Chinese Minority Headline Generation (CMHG), which includes 100,000 entries for Tibetan, and 50,000 entries each for Uyghur and Mongolian, specifically curated for headline generation tasks. Additionally, we propose a high-quality test set annotated by native speakers, designed to serve as a benchmark for future research in this domain. We hope this dataset will become a valuable resource for advancing headline generation in Chinese minority languages and contribute to the development of related benchmarks.

This study presents a framework for automated evaluation of dynamically evolving topic models using Large Language Models (LLMs). Topic modeling is essential for organizing and retrieving scholarly content in digital library systems, helping users navigate complex and evolving knowledge domains. However, widely used automated metrics, such as coherence and diversity, often capture only narrow statistical patterns and fail to explain semantic failures in practice. We introduce a purpose-oriented evaluation framework that employs nine LLM-based metrics spanning four key dimensions of topic quality: lexical validity, intra-topic semantic soundness, inter-topic structural soundness, and document-topic alignment soundness. The framework is validated through adversarial and sampling-based protocols, and is applied across datasets spanning news articles, scholarly publications, and social media posts, as well as multiple topic modeling methods and open-source LLMs. Our analysis shows that LLM-based metrics provide interpretable, robust, and task-relevant assessments, uncovering critical weaknesses in topic models such as redundancy and semantic drift, which are often missed by traditional metrics. These results support the development of scalable, fine-grained evaluation tools for maintaining topic relevance in dynamic datasets. All code and data supporting this work are accessible at https://github.com/zhiyintan/topic-model-LLMjudgment.