Retrieval-augmented generation improves the factual accuracy of Large Language Models (LLMs) by incorporating external context, but often suffers from irrelevant retrieved content that hinders effectiveness. Context compression addresses this issue by filtering out irrelevant information from context before LLM generation. However, existing methods struggle to adaptively adjust compression rates for different context, maintain low latency and integrate information across multiple documents. To overcome these limitations, We introduce AttnComp, an adaptive, efficient and context-aware compression framework. By leveraging the attention mechanism of LLMs to identify relevant information, AttnComp employs a Top-P compression algorithm to retain the minimal set of documents whose cumulative attention weights exceeds a predefined threshold. In addition to compression, AttnComp estimates response confidence by assessing the overall relevance of the retrieved content, enabling users to gauge response reliability. Experiments demonstrate that AttnComp outperforms existing compression methods and uncompressed baselines, achieving higher accuracy with substantial compression rates and lower latency.

Model editing, the process of efficiently modifying factual knowledge in pre-trained language models, is critical for maintaining their accuracy and relevance. However, existing editing methods often introduce unintended side effects, degrading model performance in unpredictable ways. While much research has focused on improving editing algorithms, the role of the target knowledge's intrinsic properties remains a significant, underexplored factor. This paper addresses this gap by first proposing the ``Knowledge Spectrum,'' a systematic framework for categorizing knowledge based on its real-world popularity, the model's pre-edit familiarity, and the linguistic structure of the eliciting question. Our empirical analysis reveals that these characteristics are strong predictors of editing success and stability. Informed by these findings, we introduce the ``Knowledge-Diagnostic Framework,'' an adaptive strategy that tailors editing intensity to the diagnosed difficulty of a knowledge item. We demonstrate that this framework significantly improves success rates for challenging edits while optimizing computational resources. Our work provides a more comprehensive understanding of the factors governing model editing.

Large Vision-Language Models (LVLMs) have recently demonstrated remarkable progress, yet hallucination remains a critical barrier, particularly in chart understanding, which requires sophisticated perceptual and cognitive abilities as well as rigorous factual accuracy. While prior work has investigated hallucinations and chart comprehension independently, their intersection remains largely unexplored. To address this gap, we present ChartHal, a benchmark that features a fine-grained taxonomy of hallucination scenarios in chart understanding, along with a human-validated dataset of 1,062 samples. Our evaluation shows that state-of-the-art LVLMs suffer from severe hallucinations on ChartHal, including proprietary models such as GPT-5 and o4-mini, which achieve only 34.46% and 22.79% accuracy, respectively. Further analysis reveals that questions involving information absent from or contradictory to charts are especially likely to trigger hallucinations, underscoring the urgent need for more robust mitigation strategies. Code and data are available at https://github.com/ymcui/ChartHal .

Dialogue agents based on large language models (LLMs) have shown promising performance in proactive dialogue, which requires effective strategy planning. However, existing approaches to strategy planning for proactive dialogue face several limitations: limited strategy coverage, preference bias in planning, and reliance on costly additional training. To address these, we propose PRINCIPLES: a synthetic strategy memory for proactive dialogue agents. PRINCIPLES is derived through offline self-play simulations and serves as reusable knowledge that guides strategy planning during inference, eliminating the need for additional training and data annotation. We evaluate PRINCIPLES in both emotional support and persuasion domains, demonstrating consistent improvements over strong baselines. Furthermore, PRINCIPLES maintains its robustness across extended and more diverse evaluation settings. See our project page at https://huggingface.co/spaces/kimnamssya/Principles.

We explore the applicability of text-to-code to solve real-world problems that are typically solved in natural language, such as legal judgment and medical QA. Unlike previous works, our approach leverages the explicit reasoning provided by program generation. We present ICRAG, a framework that transforms natural language into executable programs through iterative refinement using external knowledge from domain resources and GitHub. Across 13 benchmarks, ICRAG achieves up to 161.1\% relative improvement. We provide a detailed analysis of the generated code and the impact of external knowledge, and we discuss the limitations of applying text-to-code approaches to real-world natural language tasks.

Knowledge of slang is a desirable feature of LLMs in the context of user interaction, as slang often reflects an individual's social identity. Several works on informal language processing have defined and curated benchmarks for tasks such as detection and identification of slang. In this paper, we focus on queer slang. Queer slang can be mistakenly flagged as hate speech or can evoke negative responses from LLMs during user interaction. Research efforts so far have not focused explicitly on queer slang. In particular, detection and processing of queer slang have not been thoroughly evaluated due to the lack of a high-quality annotated benchmark. To address this gap, we curate SLAyiNG, the first dataset containing annotated queer slang derived from subtitles, social media posts, and podcasts, reflecting real-world usage. We describe our data curation process, including the collection of slang terms and definitions, scraping sources for examples that reflect usage of these terms, and our ongoing annotation process. As preliminary results, we calculate inter-annotator agreement for human annotators and OpenAI's model o3-mini, evaluating performance on the task of sense disambiguation. Reaching an average Krippendorff's alpha of 0.746, we argue that state-of-the-art reasoning models can serve as tools for pre-filtering, but the complex and often sensitive nature of queer language data requires expert and community-driven annotation efforts.

Recent advancements in reinforcement learning (RL) have enhanced the reasoning abilities of large language models (LLMs), yet the impact on multimodal LLMs (MLLMs) is limited. Particularly in vision-intensive tasks like geometric reasoning, MLLMs hallucinate frequently, leading to inaccurate reasoning. We attribute this to the perceptual bottleneck in MLLMs, which caps the benefits of reasoning training. To quantify this, we design a Geo-Perception Question-Answering (GeoPQA) benchmark, targeting basic geometric concepts and spatial relationships. Experiments on GeoPQA reveal significant shortcomings of MLLMs in visual perception, which constrain RL reward signals for effective training. To address this bottleneck, we propose a two-stage RL training framework by first enhancing the visual perception of geometric structures, then fostering reasoning capabilities. Applied to Qwen2.5-VL-3B-Instruct, our two-stage training improves geometric reasoning by 9.7% and geometric problem solving by 9.1%, compared to the direct reasoning training approach. Our method also generalizes to other vision-intensive domains like figure understanding, highlighting the importance of perceptual grounding in effective MLLM reasoning.

Medical fact-checking has become increasingly critical as more individuals seek medical information online. However, existing datasets predominantly focus on human-generated content, leaving the verification of content generated by large language models (LLMs) relatively unexplored. To address this gap, we introduce MedFact, the first evidence-based Chinese medical fact-checking dataset of LLM-generated medical content. It consists of 1,321 questions and 7,409 claims, mirroring the complexities of real-world medical scenarios. We conduct comprehensive experiments in both in-context learning (ICL) and fine-tuning settings, showcasing the capability and challenges of current LLMs on this task, accompanied by an in-depth error analysis to point out key directions for future research. Our dataset is publicly available at https://github.com/AshleyChenNLP/MedFact.

While various multimodal multi-image evaluation datasets have been emerged, but these datasets are primarily based on English, and there has yet to be a Chinese multi-image dataset. To fill this gap, we introduce RealBench, the first Chinese multimodal multi-image dataset, which contains 9393 samples and 69910 images. RealBench distinguishes itself by incorporating real user-generated content, ensuring high relevance to real-world applications. Additionally, the dataset covers a wide variety of scenes, image resolutions, and image structures, further increasing the difficulty of multi-image understanding. Ultimately, we conduct a comprehensive evaluation of RealBench using 21 multimodal LLMs of different sizes, including closed-source models that support multi-image inputs as well as open-source visual and video models. The experimental results indicate that even the most powerful closed-source models still face challenges when handling multi-image Chinese scenarios. Moreover, there remains a noticeable performance gap of around 71.8\% on average between open-source visual/video models and closed-source models. These results show that RealBench provides an important research foundation for further exploring multi-image understanding capabilities in the Chinese context.

Spelling correction from visual input poses unique challenges for vision language models (VLMs), as it requires not only detecting but also correcting textual errors directly within images. We present ReViCo (Real Visual Correction), the first benchmark that systematically evaluates VLMs on real-world visual spelling correction across Chinese and English. ReViCo contains naturally occurring errors collected from real-world image data and supports fine-grained evaluation at both image and token levels. Through comprehensive experiments on representative cascaded (Qwen) and native (InternVL) open-source models, as well as closed-source systems (GPT-4o, Claude), we show that current VLMs fall significantly short of human performance, particularly in correction. To address these limitations, we explore two solution paradigms: a Joint OCR-Correction pipeline and a Background Information enhanced approach, both of which yield consistent performance gains. Our analysis highlights fundamental limitations of existing architectures and provides actionable insights for advancing multimodal spelling correction.