Junjie Chen

Junjie Chen, Yuxi Dong, Haitao Li et al.

As large language models (LLMs) are increasingly used for long-form generation, reliably evaluating long-form outputs has become a critical challenge. LLM-as-a-judge offers a scalable alternative to human evaluation, yet its reliability in long-form output evaluation remains underexamined: existing meta-evaluation benchmarks focus mainly on short-form outputs. Compared with short-form evaluation, long-form evaluation is not merely a matter of output length; it often requires judges to handle more complex document-level demands. In this work, we introduce LongJudgeBench, a comprehensive benchmark for evaluating LLM judges on long-form outputs across diverse real-world scenarios and judging protocols. We systematically evaluate a broad range of LLM judges, covering multiple base models and judging settings. Our results reveal a substantial reliability gap: current LLM judges remain unstable across scenarios, and rubrics or references are helpful but not always sufficient. We hope LongJudgeBench will support future research on more robust, context-aware, and human-aligned LLM-as-a-judge methods. Our code is available at https://anonymous.4open.science/r/LongJudgeBench-F782.

Jiahao Li, Zhourun Wu, Wenhao Lin et al.

Motivation: Enhancers are important cis-regulatory elements that regulate a wide range of biological functions and enhance the transcription of target genes. Although many feature extraction methods have been proposed to improve the performance of enhancer identification, they cannot learn position-related multiscale contextual information from raw DNA sequences. Results: In this article, we propose a novel enhancer identification method (iEnhancer-ELM) based on BERT-like enhancer language models. iEnhancer-ELM tokenizes DNA sequences with multi-scale k-mers and extracts contextual information of different scale k-mers related with their positions via an multi-head attention mechanism. We first evaluate the performance of different scale k-mers, then ensemble them to improve the performance of enhancer identification. The experimental results on two popular benchmark datasets show that our model outperforms stateof-the-art methods. We further illustrate the interpretability of iEnhancer-ELM. For a case study, we discover 30 enhancer motifs via a 3-mer-based model, where 12 of motifs are verified by STREME and JASPAR, demonstrating our model has a potential ability to unveil the biological mechanism of enhancer. Availability and implementation: The models and associated code are available at https://github.com/chen-bioinfo/iEnhancer-ELM Contact: junjiechen@hit.edu.cn Supplementary information: Supplementary data are available at Bioinformatics Advances online.

Junjie Chen, Li Niu, Siyuan Zhou et al.

Semantic segmentation is an important and prevalent task, but severely suffers from the high cost of pixel-level annotations when extending to more classes in wider applications. To this end, we focus on the problem named weak-shot semantic segmentation, where the novel classes are learnt from cheaper image-level labels with the support of base classes having off-the-shelf pixel-level labels. To tackle this problem, we propose SimFormer, which performs dual similarity transfer upon MaskFormer. Specifically, MaskFormer disentangles the semantic segmentation task into two sub-tasks: proposal classification and proposal segmentation for each proposal. Proposal segmentation allows proposal-pixel similarity transfer from base classes to novel classes, which enables the mask learning of novel classes. We also learn pixel-pixel similarity from base classes and distill such class-agnostic semantic similarity to the semantic masks of novel classes, which regularizes the segmentation model with pixel-level semantic relationship across images. In addition, we propose a complementary loss to facilitate the learning of novel classes. Comprehensive experiments on the challenging COCO-Stuff-10K and ADE20K datasets demonstrate the effectiveness of our method. Codes are available at https://github.com/bcmi/SimFormer-Weak-Shot-Semantic-Segmentation.

Taohua Zhou, Yining Shi, Junjie Chen et al. · tsinghua

Environmental perception with the multi-modal fusion of radar and camera is crucial in autonomous driving to increase accuracy, completeness, and robustness. This paper focuses on utilizing millimeter-wave (MMW) radar and camera sensor fusion for 3D object detection. A novel method that realizes the feature-level fusion under the bird's-eye view (BEV) for a better feature representation is proposed. Firstly, radar points are augmented with temporal accumulation and sent to a spatial-temporal encoder for radar feature extraction. Meanwhile, multi-scale image 2D features which adapt to various spatial scales are obtained by image backbone and neck model. Then, image features are transformed to BEV with the designed view transformer. In addition, this work fuses the multi-modal features with a two-stage fusion model called point-fusion and ROI-fusion, respectively. Finally, a detection head regresses objects category and 3D locations. Experimental results demonstrate that the proposed method realizes the state-of-the-art (SOTA) performance under the most crucial detection metrics-mean average precision (mAP) and nuScenes detection score (NDS) on the challenging nuScenes dataset.

Kaituo Xu, Yan Jia, Kai Huang et al.

We present FireRedASR2S, a state-of-the-art industrial-grade all-in-one automatic speech recognition (ASR) system. It integrates four modules in a unified pipeline: ASR, Voice Activity Detection (VAD), Spoken Language Identification (LID), and Punctuation Prediction (Punc). All modules achieve SOTA performance on the evaluated benchmarks: FireRedASR2: An ASR module with two variants, FireRedASR2-LLM (8B+ parameters) and FireRedASR2-AED (1B+ parameters), supporting speech and singing transcription for Mandarin, Chinese dialects and accents, English, and code-switching. Compared to FireRedASR, FireRedASR2 delivers improved recognition accuracy and broader dialect and accent coverage. FireRedASR2-LLM achieves 2.89% average CER on 4 public Mandarin benchmarks and 11.55% on 19 public Chinese dialects and accents benchmarks, outperforming competitive baselines including Doubao-ASR, Qwen3-ASR, and Fun-ASR. FireRedVAD: An ultra-lightweight module (0.6M parameters) based on the Deep Feedforward Sequential Memory Network (DFSMN), supporting streaming VAD, non-streaming VAD, and multi-label VAD (mVAD). On the FLEURS-VAD-102 benchmark, it achieves 97.57% frame-level F1 and 99.60% AUC-ROC, outperforming Silero-VAD, TEN-VAD, FunASR-VAD, and WebRTC-VAD. FireRedLID: An Encoder-Decoder LID module supporting 100+ languages and 20+ Chinese dialects and accents. On FLEURS (82 languages), it achieves 97.18% utterance-level accuracy, outperforming Whisper and SpeechBrain. FireRedPunc: A BERT-style punctuation prediction module for Chinese and English. On multi-domain benchmarks, it achieves 78.90% average F1, outperforming FunASR-Punc (62.77%). To advance research in speech processing, we release model weights and code at https://github.com/FireRedTeam/FireRedASR2S.

Zhao Tian, Zifan Zhang, Tao Xiao et al.

Recent advances in coding agents have shown remarkable progress in software issue resolution. In practice, real-world issues are typically bug fixes or feature requests in which human developers naturally incorporate refactoring as part of the resolution process, resulting in tangled refactoring. Since LLMs are trained on large-scale open-source repositories, coding agents may inherit such behaviors. In this paper, we conduct an empirical study on Multi-SWE-bench, analyzing 3,691 valid patches generated by three agent frameworks with 12 LLMs. We find that coding agents introduce tangled refactorings less frequently (21.43% vs. 36.72%) and with lower intensity (0.66 vs. 1.75) than human developers, although they exhibit a broader diversity of refactoring types. Logistic regression analysis further shows that tangled refactorings are strongly associated with reduced compilability, while exhibiting no significant association with functional correctness. Based on these findings, we propose a refactoring-aware refinement approach that assesses the necessity and safety of tangled refactorings and selectively removes or repairs problematic operations. Our approach improves compilability from 19.34% to 38.33%, and additionally resolves 2.79% previously unresolved issues. Overall, this work presents the first step towards understanding tangled refactoring practices in agentic issue resolution and opens up avenues for future work.

Yiqi Nie, Fei Wang, Junjie Chen et al.

Memes represent a tightly coupled, multimodal form of social expression, in which visual context and overlaid text jointly convey nuanced affect and commentary. Inspired by cognitive reappraisal in psychology, we introduce Meme Reappraisal, a novel multimodal generation task that aims to transform negatively framed memes into constructive ones while preserving their underlying scenario, entities, and structural layout. Unlike prior works on meme understanding or generation, Meme Reappraisal requires emotion-controllable, structure-preserving multimodal transformation under multiple semantic and stylistic constraints. To support this task, we construct MER-Bench, a benchmark of real-world memes with fine-grained multimodal annotations, including source and target emotions, positively rewritten meme text, visual editing specifications, and taxonomy labels covering visual type, sentiment polarity, and layout structure. We further propose a structured evaluation framework based on a multimodal large language model (MLLM)-as-a-Judge paradigm, decomposing performance into modality-level generation quality, affect controllability, structural fidelity, and global affective alignment. Extensive experiments across representative image-editing and multimodal-generation systems reveal substantial gaps in satisfying the constraints of structural preservation, semantic consistency, and affective transformation. We believe MER-Bench establishes a foundation for research on controllable meme editing and emotion-aware multimodal generation. Our code is available at: https://github.com/one-seven17/MER-Bench.

Jieteng Yao, Junjie Chen, Li Niu et al.

Affordance learning considers the interaction opportunities for an actor in the scene and thus has wide application in scene understanding and intelligent robotics. In this paper, we focus on contextual affordance learning, i.e., using affordance as context to generate a reasonable human pose in a scene. Existing scene-aware human pose generation methods could be divided into two categories depending on whether using pose templates. Our proposed method belongs to the template-based category, which benefits from the representative pose templates. Moreover, inspired by recent transformer-based methods, we associate each query embedding with a pose template, and use the interaction between query embeddings and scene feature map to effectively predict the scale and offsets for each pose template. In addition, we employ knowledge distillation to facilitate the offset learning given the predicted scale. Comprehensive experiments on Sitcom dataset demonstrate the effectiveness of our method.

Lin Yang, Junjie Chen, Shutao Gao et al.

The rapid growth of deep learning (DL) has spurred interest in enhancing log-based anomaly detection. This approach aims to extract meaning from log events (log message templates) and develop advanced DL models for anomaly detection. However, these DL methods face challenges like heavy reliance on training data, labels, and computational resources due to model complexity. In contrast, traditional machine learning and data mining techniques are less data-dependent and more efficient but less effective than DL. To make log-based anomaly detection more practical, the goal is to enhance traditional techniques to match DL's effectiveness. Previous research in a different domain (linking questions on Stack Overflow) suggests that optimized traditional techniques can rival state-of-the-art DL methods. Drawing inspiration from this concept, we conducted an empirical study. We optimized the unsupervised PCA (Principal Component Analysis), a traditional technique, by incorporating lightweight semantic-based log representation. This addresses the issue of unseen log events in training data, enhancing log representation. Our study compared seven log-based anomaly detection methods, including four DL-based, two traditional, and the optimized PCA technique, using public and industrial datasets. Results indicate that the optimized unsupervised PCA technique achieves similar effectiveness to advanced supervised/semi-supervised DL methods while being more stable with limited training data and resource-efficient. This demonstrates the adaptability and strength of traditional techniques through small yet impactful adaptations.

Qingchao Shen, Zibo Xiao, Lili Huang et al.

Large Language Models (LLMs) are increasingly deployed across diverse domains, yet their vulnerability to jailbreak attacks, where adversarial inputs bypass safety mechanisms to elicit harmful outputs, poses significant security risks. While prior work has primarily focused on prompt injection attacks, these approaches often require resource-intensive prompt engineering and overlook other critical components, such as chat templates. This paper introduces TEMPLATEFUZZ, a fine-grained fuzzing framework that systematically exposes vulnerabilities in chat templates, a critical yet underexplored attack surface in LLMs. Specifically, TEMPLATEFUZZ (1) designs a series of element-level mutation rules to generate diverse chat template variants, (2) proposes a heuristic search strategy to guide the chat template generation toward the direction of amplifying the attack success rate (ASR) while preserving model accuracy, and (3) integrates an active learning-based strategy to derive a lightweight rule-based oracle for accurate and efficient jailbreak evaluation. Evaluated on twelve open-source LLMs across multiple attack scenarios, TEMPLATEFUZZ achieves an average ASR of 98.2% with only 1.1% accuracy degradation, outperforming state-of-the-art methods by 9.1%-47.9% in ASR and 8.4% in accuracy degradation. Moreover, even on five industry-leading commercial LLMs where chat templates cannot be specified, TEMPLATEFUZZ attains a 90% average ASR via chat template-based prompt injection attacks.

Ping Li, Junjie Chen, Binbin Lin et al.

Semantic segmentation plays an important role in widespread applications such as autonomous driving and robotic sensing. Traditional methods mostly use RGB images which are heavily affected by lighting conditions, \eg, darkness. Recent studies show thermal images are robust to the night scenario as a compensating modality for segmentation. However, existing works either simply fuse RGB-Thermal (RGB-T) images or adopt the encoder with the same structure for both the RGB stream and the thermal stream, which neglects the modality difference in segmentation under varying lighting conditions. Therefore, this work proposes a Residual Spatial Fusion Network (RSFNet) for RGB-T semantic segmentation. Specifically, we employ an asymmetric encoder to learn the compensating features of the RGB and the thermal images. To effectively fuse the dual-modality features, we generate the pseudo-labels by saliency detection to supervise the feature learning, and develop the Residual Spatial Fusion (RSF) module with structural re-parameterization to learn more promising features by spatially fusing the cross-modality features. RSF employs a hierarchical feature fusion to aggregate multi-level features, and applies the spatial weights with the residual connection to adaptively control the multi-spectral feature fusion by the confidence gate. Extensive experiments were carried out on two benchmarks, \ie, MFNet database and PST900 database. The results have shown the state-of-the-art segmentation performance of our method, which achieves a good balance between accuracy and speed.

Ping Li, Junjie Chen, Li Yuan et al.

To alleviate the expensive human labeling, semi-supervised semantic segmentation employs a few labeled images and an abundant of unlabeled images to predict the pixel-level label map with the same size. Previous methods often adopt co-training using two convolutional networks with the same architecture but different initialization, which fails to capture the sufficiently diverse features. This motivates us to use tri-training and develop the triple-view encoder to utilize the encoders with different architectures to derive diverse features, and exploit the knowledge distillation skill to learn the complementary semantics among these encoders. Moreover, existing methods simply concatenate the features from both encoder and decoder, resulting in redundant features that require large memory cost. This inspires us to devise a dual-frequency decoder that selects those important features by projecting the features from the spatial domain to the frequency domain, where the dual-frequency channel attention mechanism is introduced to model the feature importance. Therefore, we propose a Triple-view Knowledge Distillation framework, termed TriKD, for semi-supervised semantic segmentation, including the triple-view encoder and the dual-frequency decoder. Extensive experiments were conducted on two benchmarks, \ie, Pascal VOC 2012 and Cityscapes, whose results verify the superiority of the proposed method with a good tradeoff between precision and inference speed.

Lingfeng Tang, Daoping Zhang, Junjie Chen et al.

Host-GPU data movement has become a latency-critical bottleneck in LLM serving, surfacing in common paths such as model-weight movement and KV cache offload/fetch. Today, each host-GPU copy is effectively confined to the PCIe path of the target GPU, even though modern multi-GPU servers contain additional PCIe links on peer GPUs and high bandwidth GPU interconnects. This leaves substantial intra-server I/O capacity unused. To address this issue, we present Multipath Memory Access (MMA), a software-defined multipath memory access system for host--GPU data transfer. To the best of our knowledge, MMA is the first software-defined system to enable efficient multipath host--GPU data transfer within a single multi-GPU server. MMA expands a single host--GPU copy across available direct and relay paths without hardware, driver, or application changes. It preserves CUDA stream semantics with a dependency-preserving Dummy Task, coordinates distributed micro-transfer completion through a lightweight synchronization mechanism, and uses queue backpressure to route traffic without explicit link-state feedback. On an 8-GPU NVIDIA H20 server, MMA achieves 245 GB/s peak host-to-GPU bandwidth, a 4.62x improvement over native CUDA copies, and reduces TTFT for KV cache fetching by 1.14-2.38x and model wake-up/switching latency by 1.12-2.48x.

Haitao Li, Qian Dong, Junjie Chen et al.

The rapid advancement of Large Language Models (LLMs) has driven their expanding application across various fields. One of the most promising applications is their role as evaluators based on natural language responses, referred to as ''LLMs-as-judges''. This framework has attracted growing attention from both academia and industry due to their excellent effectiveness, ability to generalize across tasks, and interpretability in the form of natural language. This paper presents a comprehensive survey of the LLMs-as-judges paradigm from five key perspectives: Functionality, Methodology, Applications, Meta-evaluation, and Limitations. We begin by providing a systematic definition of LLMs-as-Judges and introduce their functionality (Why use LLM judges?). Then we address methodology to construct an evaluation system with LLMs (How to use LLM judges?). Additionally, we investigate the potential domains for their application (Where to use LLM judges?) and discuss methods for evaluating them in various contexts (How to evaluate LLM judges?). Finally, we provide a detailed analysis of the limitations of LLM judges and discuss potential future directions. Through a structured and comprehensive analysis, we aim aims to provide insights on the development and application of LLMs-as-judges in both research and practice. We will continue to maintain the relevant resource list at https://github.com/CSHaitao/Awesome-LLMs-as-Judges.

Zheyuan Yang, Liqiang Shang, Junjie Chen et al.

We introduce TableVista, a comprehensive benchmark for evaluating foundation models in multimodal table reasoning under visual and structural complexity. TableVista consists of 3,000 high-quality table reasoning problems, where each instance is expanded into 10 distinct visual variants through our multi-style rendering and transformation pipeline. This process encompasses diverse scenario styles, robustness perturbations, and vision-only configurations, culminating in 30,000 multimodal samples for a multi-dimensional evaluation. We conduct an extensive evaluation of 29 state-of-the-art open-source and proprietary foundation models on TableVista. Through comprehensive quantitative and qualitative analysis, we find that while evaluated models remain largely stable across diverse rendering styles, they exhibit pronounced performance degradation on complex structural layouts and vision-only settings, revealing that current models struggle to maintain reasoning consistency when structural complexity combines with visually integrated presentations. These findings highlight critical gaps in current multimodal capabilities, providing insights for advancing more robust and reliable table understanding models.

GLM-V Team, Wenyi Hong, Wenmeng Yu et al.

We present GLM-4.1V-Thinking and GLM-4.5V, a family of vision-language models (VLMs) designed to advance general-purpose multimodal understanding and reasoning. In this report, we share our key findings in the development of the reasoning-centric training framework. We first develop a capable vision foundation model with significant potential through large-scale pre-training, which arguably sets the upper bound for the final performance. We then propose Reinforcement Learning with Curriculum Sampling (RLCS) to unlock the full potential of the model, leading to comprehensive capability enhancement across a diverse range of tasks, including STEM problem solving, video understanding, content recognition, coding, grounding, GUI-based agents, and long document interpretation. In a comprehensive evaluation across 42 public benchmarks, GLM-4.5V achieves state-of-the-art performance on nearly all tasks among open-source models of similar size, and demonstrates competitive or even superior results compared to closed-source models such as Gemini-2.5-Flash on challenging tasks including Coding and GUI Agents. Meanwhile, the smaller GLM-4.1V-9B-Thinking remains highly competitive-achieving superior results to the much larger Qwen2.5-VL-72B on 29 benchmarks. We open-source both GLM-4.1V-9B-Thinking and GLM-4.5V. Code, models and more information are released at https://github.com/zai-org/GLM-V.

Xianda Chen, Mingxing Peng, PakHin Tiu et al.

Accurate modeling of car-following behaviors is essential for various applications in traffic management and autonomous driving systems. However, current approaches often suffer from limitations like high sensitivity to data quality and lack of interpretability. In this study, we propose GenFollower, a novel zero-shot prompting approach that leverages large language models (LLMs) to address these challenges. We reframe car-following behavior as a language modeling problem and integrate heterogeneous inputs into structured prompts for LLMs. This approach achieves improved prediction performance and interpretability compared to traditional baseline models. Experiments on the Waymo Open datasets demonstrate GenFollower's superior performance and ability to provide interpretable insights into factors influencing car-following behavior. This work contributes to advancing the understanding and prediction of car-following behaviors, paving the way for enhanced traffic management and autonomous driving systems.

Tingting Zhu, Bo Peng, Jifan Liang et al.

Semantic communications (SC) have been expected to be a new paradigm shifting to catalyze the next generation communication, whose main concerns shift from accurate bit transmission to effective semantic information exchange in communications. However, the previous and widely-used metrics for images are not applicable to evaluate the image semantic similarity in SC. Classical metrics to measure the similarity between two images usually rely on the pixel level or the structural level, such as the PSNR and the MS-SSIM. Straightforwardly using some tailored metrics based on deep-learning methods in CV community, such as the LPIPS, is infeasible for SC. To tackle this, inspired by BERTScore in NLP community, we propose a novel metric for evaluating image semantic similarity, named Vision Transformer Score (ViTScore). We prove theoretically that ViTScore has 3 important properties, including symmetry, boundedness, and normalization, which make ViTScore convenient and intuitive for image measurement. To evaluate the performance of ViTScore, we compare ViTScore with 3 typical metrics (PSNR, MS-SSIM, and LPIPS) through 4 classes of experiments: (i) correlation with BERTScore through evaluation of image caption downstream CV task, (ii) evaluation in classical image communications, (iii) evaluation in image semantic communication systems, and (iv) evaluation in image semantic communication systems with semantic attack. Experimental results demonstrate that ViTScore is robust and efficient in evaluating the semantic similarity of images. Particularly, ViTScore outperforms the other 3 typical metrics in evaluating the image semantic changes by semantic attack, such as image inverse with Generative Adversarial Networks (GANs). This indicates that ViTScore is an effective performance metric when deployed in SC scenarios.

Haitao Li, Junjie Chen, Jingli Yang et al.

With the increasing intelligence and autonomy of LLM agents, their potential applications in the legal domain are becoming increasingly apparent. However, existing general-domain benchmarks cannot fully capture the complexity and subtle nuances of real-world judicial cognition and decision-making. Therefore, we propose LegalAgentBench, a comprehensive benchmark specifically designed to evaluate LLM Agents in the Chinese legal domain. LegalAgentBench includes 17 corpora from real-world legal scenarios and provides 37 tools for interacting with external knowledge. We designed a scalable task construction framework and carefully annotated 300 tasks. These tasks span various types, including multi-hop reasoning and writing, and range across different difficulty levels, effectively reflecting the complexity of real-world legal scenarios. Moreover, beyond evaluating final success, LegalAgentBench incorporates keyword analysis during intermediate processes to calculate progress rates, enabling more fine-grained evaluation. We evaluated eight popular LLMs, highlighting the strengths, limitations, and potential areas for improvement of existing models and methods. LegalAgentBench sets a new benchmark for the practical application of LLMs in the legal domain, with its code and data available at \url{https://github.com/CSHaitao/LegalAgentBench}.

Trae Research Team, Pengfei Gao, Zhao Tian et al. · pku

Software issue resolution is a critical challenge in software engineering and has garnered increasing attention in recent years. With the rapid advancement of large language models (LLMs), substantial progress has been made in addressing real-world software engineering tasks. Recent studies have introduced ensemble reasoning techniques to enhance the performance of LLM-based issue resolution. However, existing prompting-based methods still face limitations in effectively exploring large ensemble spaces and lack the capacity for repository-level understanding, both of which constrain their overall effectiveness. In this paper, we propose Trae Agent, the first agent-based ensemble reasoning approach for repository-level issue resolution. Trae Agent formulates our goal as an optimal solution search problem and addresses two key challenges, i.e., large ensemble spaces and repository-level understanding, through modular agents for generation, pruning, and selection. We conduct extensive experiments using three leading LLMs on the widely-adopted SWE-bench benchmark, comparing Trae Agent against four state-of-the-art ensemble reasoning techniques. Experimental results demonstrate that Trae Agent consistently achieves superior performance, with an average improvement of 10.22% over all baselines in terms of Pass@1. Trae Agent has achieved first place on the SWE-bench Verified leaderboard, with a notable Pass@1 score of 75.20%. We are pleased to release Trae Agent as an open-source project to support the research community, with all resources available at https://github.com/bytedance/trae-agent.

Jiacheng Liu, Chang Zou, Yuanhuiyi Lyu et al.

Diffusion Transformers (DiT) have revolutionized high-fidelity image and video synthesis, yet their computational demands remain prohibitive for real-time applications. To solve this problem, feature caching has been proposed to accelerate diffusion models by caching the features in the previous timesteps and then reusing them in the following timesteps. However, at timesteps with significant intervals, the feature similarity in diffusion models decreases substantially, leading to a pronounced increase in errors introduced by feature caching, significantly harming the generation quality. To solve this problem, we propose TaylorSeer, which firstly shows that features of diffusion models at future timesteps can be predicted based on their values at previous timesteps. Based on the fact that features change slowly and continuously across timesteps, TaylorSeer employs a differential method to approximate the higher-order derivatives of features and predict features in future timesteps with Taylor series expansion. Extensive experiments demonstrate its significant effectiveness in both image and video synthesis, especially in high acceleration ratios. For instance, it achieves an almost lossless acceleration of 4.99$\times$ on FLUX and 5.00$\times$ on HunyuanVideo without additional training. On DiT, it achieves $3.41$ lower FID compared with previous SOTA at $4.53$$\times$ acceleration. %Our code is provided in the supplementary materials and will be made publicly available on GitHub. Our codes have been released in Github:https://github.com/Shenyi-Z/TaylorSeer

Xianda Chen, PakHin Tiu, Xu Han et al.

The continual evolution of autonomous driving technology requires car-following models that can adapt to diverse and dynamic traffic environments. Traditional learning-based models often suffer from performance degradation when encountering unseen traffic patterns due to a lack of continual learning capabilities. This paper proposes a novel car-following model based on continual learning that addresses this limitation. Our framework incorporates Elastic Weight Consolidation (EWC) and Memory Aware Synapses (MAS) techniques to mitigate catastrophic forgetting and enable the model to learn incrementally from new traffic data streams. We evaluate the performance of the proposed model on the Waymo and Lyft datasets which encompass various traffic scenarios. The results demonstrate that the continual learning techniques significantly outperform the baseline model, achieving 0\% collision rates across all traffic conditions. This research contributes to the advancement of autonomous driving technology by fostering the development of more robust and adaptable car-following models.

Zhihao Huang, Xi Qiu, Yukuo Ma et al.

Autoregressive models have achieved significant success in image generation. However, unlike the inherent hierarchical structure of image information in the spectral domain, standard autoregressive methods typically generate pixels sequentially in a fixed spatial order. To better leverage this spectral hierarchy, we introduce NextFrequency Image Generation (NFIG). NFIG is a novel framework that decomposes the image generation process into multiple frequency-guided stages. NFIG aligns the generation process with the natural image structure. It does this by first generating low-frequency components, which efficiently capture global structure with significantly fewer tokens, and then progressively adding higher-frequency details. This frequency-aware paradigm offers substantial advantages: it not only improves the quality of generated images but crucially reduces inference cost by efficiently establishing global structure early on. Extensive experiments on the ImageNet-256 benchmark validate NFIG's effectiveness, demonstrating superior performance (FID: 2.81) and a notable 1.25x speedup compared to the strong baseline VAR-d20. The source code is available at https://github.com/Pride-Huang/NFIG.

Yuhang Han, Xuyang Liu, Zihan Zhang et al.

The quadratic complexity of Multimodal Large Language Models (MLLMs) with respect to context length poses significant computational and memory challenges, hindering their real-world deployment. In the paper, we devise a ''filter-correlate-compress'' framework to accelerate the MLLM by systematically optimizing multimodal context length during prefilling. The framework first implements FiCoCo-V, a training-free method operating within the vision encoder. It employs a redundancy-based token discard mechanism that uses a novel integrated metric to accurately filter out redundant visual tokens. To mitigate information loss, the framework introduces a correlation-based information recycling mechanism that allows preserved tokens to selectively recycle information from correlated discarded tokens with a self-preserving compression, thereby preventing the dilution of their own core content. The framework's FiCoCo-L variant further leverages task-aware textual priors to perform token reduction directly within the LLM decoder. Extensive experiments demonstrate that the FiCoCo series effectively accelerates a range of MLLMs, achieves up to 14.7x FLOPs reduction with 93.6% performance retention. Our methods consistently outperform state-of-the-art training-free approaches, showcasing effectiveness and generalizability across model architectures, sizes, and tasks without requiring retraining. Code: https://github.com/kawhiiiileo/FiCoCo

Haitao Li, Yifan Chen, Yiran Hu et al.

Retrieval-augmented generation (RAG) has proven highly effective in improving large language models (LLMs) across various domains. However, there is no benchmark specifically designed to assess the effectiveness of RAG in the legal domain, which restricts progress in this area. To fill this gap, we propose LexRAG, the first benchmark to evaluate RAG systems for multi-turn legal consultations. LexRAG consists of 1,013 multi-turn dialogue samples and 17,228 candidate legal articles. Each sample is annotated by legal experts and consists of five rounds of progressive questioning. LexRAG includes two key tasks: (1) Conversational knowledge retrieval, requiring accurate retrieval of relevant legal articles based on multi-turn context. (2) Response generation, focusing on producing legally sound answers. To ensure reliable reproducibility, we develop LexiT, a legal RAG toolkit that provides a comprehensive implementation of RAG system components tailored for the legal domain. Additionally, we introduce an LLM-as-a-judge evaluation pipeline to enable detailed and effective assessment. Through experimental analysis of various LLMs and retrieval methods, we reveal the key limitations of existing RAG systems in handling legal consultation conversations. LexRAG establishes a new benchmark for the practical application of RAG systems in the legal domain, with its code and data available at https://github.com/CSHaitao/LexRAG.

Hanmo You, Zan Wang, Zishuo Dong et al.

Deep Learning (DL) has been widely adopted in diverse industrial domains, including autonomous driving, intelligent healthcare, and aided programming. Like traditional software, DL systems are also prone to faults, whose malfunctioning may expose users to significant risks. Consequently, numerous approaches have been proposed to address these issues. In this paper, we conduct a large-scale empirical study on 16 state-of-the-art DL model fixing approaches, spanning model-level, layer-level, and neuron-level categories, to comprehensively evaluate their performance. We assess not only their fixing effectiveness (their primary purpose) but also their impact on other critical properties, such as robustness, fairness, and backward compatibility. To ensure comprehensive and fair evaluation, we employ a diverse set of datasets, model architectures, and application domains within a uniform experimental setup for experimentation. We summarize several key findings with implications for both industry and academia. For example, model-level approaches demonstrate superior fixing effectiveness compared to others. No single approach can achieve the best fixing performance while improving accuracy and maintaining all other properties. Thus, academia should prioritize research on mitigating these side effects. These insights highlight promising directions for future exploration in this field.

Xuyang Liu, Ziming Wang, Junjie Chen et al.

Large vision-language models (LVLMs) excel at visual understanding, but face efficiency challenges due to quadratic complexity in processing long multi-modal contexts. While token compression can reduce computational costs, existing approaches are designed for single-view LVLMs and fail to consider the unique multi-view characteristics of high-resolution LVLMs with dynamic cropping. Existing methods treat all tokens uniformly, but our analysis reveals that global thumbnails can naturally guide the compression of local crops by providing holistic context for informativeness evaluation. In this paper, we first analyze dynamic cropping strategy, revealing both the complementary nature between thumbnails and crops, and the distinctive characteristics across different crops. Based on our observations, we propose "Global Compression Commander" (GlobalCom$^2$), a novel plug-and-play token compression framework for HR-LVLMs. GlobalCom$^2$ leverages thumbnail as the "commander" to guide the compression of local crops, adaptively preserving informative details while eliminating redundancy. Extensive experiments show that GlobalCom$^2$ maintains over 90% performance while compressing 90% visual tokens, reducing FLOPs and peak memory to 9.1% and 60%. Our code is available at https://github.com/xuyang-liu16/GlobalCom2.

Jiebin Yan, Jiale Rao, Junjie Chen et al.

Omnidirectional image quality assessment (OIQA) has been widely investigated in the past few years and achieved much success. However, most of existing studies are dedicated to solve the uniform distortion problem in OIQA, which has a natural gap with the non-uniform distortion problem, and their ability in capturing non-uniform distortion is far from satisfactory. To narrow this gap, in this paper, we propose a multitask auxiliary network for non-uniformly distorted omnidirectional images, where the parameters are optimized by jointly training the main task and other auxiliary tasks. The proposed network mainly consists of three parts: a backbone for extracting multiscale features from the viewport sequence, a multitask feature selection module for dynamically allocating specific features to different tasks, and auxiliary sub-networks for guiding the proposed model to capture local distortion and global quality change. Extensive experiments conducted on two large-scale OIQA databases demonstrate that the proposed model outperforms other state-of-the-art OIQA metrics, and these auxiliary sub-networks contribute to improve the performance of the proposed model. The source code is available at https://github.com/RJL2000/MTAOIQA.

Haitao Li, Jiaying Ye, Yiran Hu et al.

Legal case documents play a critical role in judicial proceedings. As the number of cases continues to rise, the reliance on manual drafting of legal case documents is facing increasing pressure and challenges. The development of large language models (LLMs) offers a promising solution for automating document generation. However, existing benchmarks fail to fully capture the complexities involved in drafting legal case documents in real-world scenarios. To address this gap, we introduce CaseGen, the benchmark for multi-stage legal case documents generation in the Chinese legal domain. CaseGen is based on 500 real case samples annotated by legal experts and covers seven essential case sections. It supports four key tasks: drafting defense statements, writing trial facts, composing legal reasoning, and generating judgment results. To the best of our knowledge, CaseGen is the first benchmark designed to evaluate LLMs in the context of legal case document generation. To ensure an accurate and comprehensive evaluation, we design the LLM-as-a-judge evaluation framework and validate its effectiveness through human annotations. We evaluate several widely used general-domain LLMs and legal-specific LLMs, highlighting their limitations in case document generation and pinpointing areas for potential improvement. This work marks a step toward a more effective framework for automating legal case documents drafting, paving the way for the reliable application of AI in the legal field. The dataset and code are publicly available at https://github.com/CSHaitao/CaseGen.

Jiebin Yan, Ziwen Tan, Yuming Fang et al.

The fast growing application of omnidirectional images calls for effective approaches for omnidirectional image quality assessment (OIQA). Existing OIQA methods have been developed and tested on homogeneously distorted omnidirectional images, but it is hard to transfer their success directly to the heterogeneously distorted omnidirectional images. In this paper, we conduct the largest study so far on OIQA, where we establish a large-scale database called OIQ-10K containing 10,000 omnidirectional images with both homogeneous and heterogeneous distortions. A comprehensive psychophysical study is elaborated to collect human opinions for each omnidirectional image, together with the spatial distributions (within local regions or globally) of distortions, and the head and eye movements of the subjects. Furthermore, we propose a novel multitask-derived adaptive feature-tailoring OIQA model named IQCaption360, which is capable of generating a quality caption for an omnidirectional image in a manner of textual template. Extensive experiments demonstrate the effectiveness of IQCaption360, which outperforms state-of-the-art methods by a significant margin on the proposed OIQ-10K database. The OIQ-10K database and the related source codes are available at https://github.com/WenJuing/IQCaption360.

Shiqi Kuang, Zhao Tian, Kaiwei Lin et al.

Issue resolution aims to automatically generate patches from given issue descriptions and has attracted significant attention with the rapid advancement of large language models (LLMs). However, due to the complexity of software issues and codebases, LLM-generated patches often fail to resolve corresponding issues. Although various advanced techniques have been proposed with carefully designed tools and workflows, they typically treat issue descriptions as direct inputs and largely overlook their quality (e.g., missing critical context or containing ambiguous information), which hinders LLMs from accurate understanding and resolution. To address this limitation, we draw on principles from software requirements engineering and propose REAgent, a requirement-driven LLM agent framework that introduces issue-oriented requirements as structured task specifications to better guide patch generation. Specifically, REAgent automatically constructs structured and information-rich issue-oriented requirements, identifies low-quality requirements, and iteratively refines them to improve patch correctness. We conduct comprehensive experiments on three widely used benchmarks using two advanced LLMs, comparing against five representative or state-of-the-art baselines. The results demonstrate that REAgent consistently outperforms all baselines, achieving an average improvement of 17.40% in terms of the number of successfully-resolved issues (% Resolved).

Weiya Li, Junjie Chen, Bei Li et al.

Machine translation has long been a central task in natural language processing. With the rapid advancement of large language models (LLMs), there has been remarkable progress in translation quality. However, fully realizing the translation potential of LLMs remains an open challenge. Recent studies have explored multi-agent systems to decompose complex translation tasks into collaborative subtasks, showing initial promise in enhancing translation quality through agent cooperation and specialization. Nevertheless, existing multi-agent translation frameworks largely neglect foundational insights from cognitive translation studies. These insights emphasize how human translators employ different cognitive strategies, such as balancing literal and free translation, refining expressions based on context, and iteratively evaluating outputs. To address this limitation, we propose a cognitively informed multi-agent framework called TACTIC, which stands for T ranslation A gents with Cognitive- T heoretic Interactive Collaboration. The framework comprises six functionally distinct agents that mirror key cognitive processes observed in human translation behavior. These include agents for drafting, refinement, evaluation, scoring, context reasoning, and external knowledge gathering. By simulating an interactive and theory-grounded translation workflow, TACTIC effectively leverages the full capacity of LLMs for high-quality translation. Experimental results on diverse language pairs from the FLORES-200 and WMT24 benchmarks show that our method consistently achieves state-of-the-art performance. Using DeepSeek-V3 as the base model, TACTIC surpasses GPT-4.1 by an average of +0.6 XCOMET and +1.18 COMETKIWI-23. Compared to DeepSeek-R1, it further improves by +0.84 XCOMET and +2.99 COMETKIWI-23. Code is available at https://github.com/weiyali126/TACTIC.

Xiangheng He, Junjie Chen, Björn W. Schuller

Multi-task learning (MTL) aims to improve the performance of a primary task by jointly learning with related auxiliary tasks. Traditional MTL methods select tasks randomly during training. However, both previous studies and our results suggest that such a random selection of tasks may not be helpful, and can even be harmful to performance. Therefore, new strategies for task selection and assignment in MTL need to be explored. This paper studies the multi-modal, multi-task dialogue act classification task, and proposes a method for selecting and assigning tasks based on non-stationary multi-armed bandits (MAB) with discounted Thompson Sampling (TS) using Gaussian priors. Our experimental results show that in different training stages, different tasks have different utility. Our proposed method can effectively identify the task utility, actively avoid useless or harmful tasks, and realise the task assignment during training. Our proposed method is significantly superior in terms of UAR and F1 to the single-task and multi-task baselines with p-values < 0.05. Further analysis of experiments indicates that for the dataset with the data imbalance problem, our proposed method has significantly higher stability and can obtain consistent and decent performance for minority classes. Our proposed method is superior to the current state-of-the-art model.

Chen Yang, Lin Yang, Ziqi Wang et al.

Recent advances in large language models (LLMs) have enabled promising performance in unit test generation through in-context learning (ICL). However, the quality of in-context examples significantly influences the effectiveness of generated tests-poorly structured or semantically unclear test examples often lead to suboptimal outputs. In this paper, we propose CLAST, a novel technique that systematically refines unit tests to improve their semantic clarity, thereby enhancing their utility as in-context examples. The approach decomposes complex tests into logically clearer ones and improves semantic clarity through a combination of program analysis and LLM-based rewriting. We evaluated CLAST on four open-source and three industrial projects. The results demonstrate that CLAST largely outperforms UTgen, the state-of-the-art refinement technique, in both preserving test effectiveness and enhancing semantic clarity. Specifically, CLAST fully retains the original effectiveness of unit tests, while UTgen reduces compilation success rate (CSR), pass rate (PR), test coverage (Cov), and mutation score (MS) by an average of 12.90%, 35.82%, 4.65%, and 5.07%, respectively. Over 85.33% of participants in our user study preferred the semantic clarity of CLAST-refined tests. Notably, incorporating CLAST-refined tests as examples effectively improves ICL-based unit test generation approaches such as RAGGen and TELPA, resulting in an average increase of 25.97% in CSR, 28.22% in PR, and 45.99% in Cov for generated tests, compared to incorporating UTgen-refined tests. The insights from the follow-up user study not only reinforce CLAST's potential impact in software testing practice but also illuminate avenues for future research.

Junjie Chen, Weilong Chen, Yifan Zuo et al.

Category-agnostic pose estimation aims to locate keypoints on query images according to a few annotated support images for arbitrary novel classes. Existing methods generally extract support features via heatmap pooling, and obtain interacted features from support and query via cross-attention. Hence, these works neglect to mine fine-grained and structure-aware (FGSA) features from both support and query images, which are crucial for pixel-level keypoint localization. To this end, we propose a novel yet concise framework, which recurrently mines FGSA features from both support and query images. Specifically, we design a FGSA mining module based on deformable attention mechanism. On the one hand, we mine fine-grained features by applying deformable attention head over multi-scale feature maps. On the other hand, we mine structure-aware features by offsetting the reference points of keypoints to their linked keypoints. By means of above module, we recurrently mine FGSA features from support and query images, and thus obtain better support features and query estimations. In addition, we propose to use mixup keypoints to pad various classes to a unified keypoint number, which could provide richer supervision than the zero padding used in existing works. We conduct extensive experiments and in-depth studies on large-scale MP-100 dataset, and outperform SOTA method dramatically (+3.2\%PCK@0.05). Code is avaiable at https://github.com/chenbys/FMMP.

Junjie Chen, Xuyang Liu, Subin Huang et al.

With the advent of large vision-language models (LVLMs) demonstrating increasingly human-like abilities, a pivotal question emerges: do different LVLMs interpret multimodal sarcasm differently, and can a single model grasp sarcasm from multiple perspectives like humans? To explore this, we introduce an analytical framework using systematically designed prompts on existing multimodal sarcasm datasets. Evaluating 12 state-of-the-art LVLMs over 2,409 samples, we examine interpretive variations within and across models, focusing on confidence levels, alignment with dataset labels, and recognition of ambiguous "neutral" cases. We further validate our findings on a diverse 100-sample mini-benchmark, incorporating multiple datasets, expanded prompt variants, and representative commercial LVLMs. Our findings reveal notable discrepancies -- across LVLMs and within the same model under varied prompts. While classification-oriented prompts yield higher internal consistency, models diverge markedly when tasked with interpretive reasoning. These results challenge binary labeling paradigms by highlighting sarcasm's subjectivity. We advocate moving beyond rigid annotation schemes toward multi-perspective, uncertainty-aware modeling, offering deeper insights into multimodal sarcasm comprehension. Our code and data are available at: https://github.com/CoderChen01/LVLMSarcasmAnalysis

Yanjie Jiang, Yian Huang, Guancheng Wang et al.

Large Language Models (LLMs) are increasingly deployed to resolve real-world GitHub issues. However, despite their potential, the specific failure modes of these models in complex repair tasks remain poorly understood. To characterize how LLM behavior diverges from human developer practices, this paper evaluates three state-of-the-art models, i.e., Claude 4.5 Sonnet, Gemini 3 Pro, and GPT-5, on the SWE-bench Verified dataset. We conduct a rigorous manual analysis of the symptoms and root causes underlying 243 failed attempts across 900 total trials. Our investigation first yields a unified failure taxonomy encompassing five distinct stages of the repair pipeline, within which we categorize typical failure symptoms and their prevalence. Secondly, our findings reveal that for all evaluated LLMs, strategy formulation and logic synthesis constitutes the most error-prone stage, followed by problem understanding, whereas localization exhibits the lowest failure rate. This suggests that LLMs may excel at fault localization, a task traditionally regarded as one of the most formidable challenges in automated program repair. Furthermore, we observe that robustness and operational costs (particularly in failure scenarios) vary significantly across different models. Finally, we uncover the root causes of these failures and propose actionable strategies to mitigate them. A particularly notable finding is that existing evaluation harnesses occasionally misjudge correct patches due to superficial discrepancies or hidden constraints. Collectively, our insights may provide promising directions for enhancing the effectiveness and reliability of LLM-based issue resolution.

Junjie Chen, Fei Wang, Zhihao Huang et al.

Human conversation involves continuous exchanges of speech and nonverbal cues such as head nods, gaze shifts, and facial expressions that convey attention and emotion. Modeling these bidirectional dynamics in 3D is essential for building expressive avatars and interactive robots. However, existing frameworks often treat talking and listening as independent processes or rely on non-causal full-sequence modeling, hindering temporal coherence across turns. We present TIMAR (Turn-level Interleaved Masked AutoRegression), a causal framework for 3D conversational head generation that models dialogue as interleaved audio-visual contexts. It fuses multimodal information within each turn and applies turn-level causal attention to accumulate conversational history, while a lightweight diffusion head predicts continuous 3D head dynamics that captures both coordination and expressive variability. Experiments on the DualTalk benchmark show that TIMAR reduces Fréchet Distance and MSE by 15-30% on the test set, and achieves similar gains on out-of-distribution data. The source code has been released at https://github.com/CoderChen01/towards-seamless-interaction.

Zhicun Yin, Junjie Chen, Ming Liu et al.

Blind facial image restoration is highly challenging due to unknown complex degradations and the sensitivity of humans to faces. Although existing methods introduce auxiliary information from generative priors or high-quality reference images, they still struggle with identity preservation problems, mainly due to improper feature introduction on detailed textures. In this paper, we focus on effectively incorporating appropriate features from high-quality reference images, presenting a novel blind facial image restoration method that considers reference selection, transfer, and reconstruction (RefSTAR). In terms of selection, we construct a reference selection (RefSel) module. For training the RefSel module, we construct a RefSel-HQ dataset through a mask generation pipeline, which contains annotating masks for 10,000 ground truth-reference pairs. As for the transfer, due to the trivial solution in vanilla cross-attention operations, a feature fusion paradigm is designed to force the features from the reference to be integrated. Finally, we propose a reference image reconstruction mechanism that further ensures the presence of reference image features in the output image. The cycle consistency loss is also redesigned in conjunction with the mask. Extensive experiments on various backbone models demonstrate superior performance, showing better identity preservation ability and reference feature transfer quality. Source code, dataset, and pre-trained models are available at https://github.com/yinzhicun/RefSTAR.

Junjie Chen, Hang Yu, Subin Huang et al.

Sarcasm in social media, frequently conveyed through the interplay of text and images, presents significant challenges for sentiment analysis and intention mining. Existing multi-modal sarcasm detection approaches have been shown to excessively depend on superficial cues within the textual modality, exhibiting limited capability to accurately discern sarcasm through subtle text-image interactions. To address this limitation, a novel framework, InterCLIP-MEP, is proposed. This framework integrates Interactive CLIP (InterCLIP), which employs an efficient training strategy to derive enriched cross-modal representations by embedding inter-modal information directly into each encoder, while using approximately 20.6$\times$ fewer trainable parameters compared with existing state-of-the-art (SOTA) methods. Furthermore, a Memory-Enhanced Predictor (MEP) is introduced, featuring a dynamic dual-channel memory mechanism that captures and retains valuable knowledge from test samples during inference, serving as a non-parametric classifier to enhance sarcasm detection robustness. Extensive experiments on MMSD, MMSD2.0, and DocMSU show that InterCLIP-MEP achieves SOTA performance, specifically improving accuracy by 1.08% and F1 score by 1.51% on MMSD2.0. Under distributional shift evaluation, it attains 73.96% accuracy, exceeding its memory-free variant by nearly 10% and the previous SOTA by over 15%, demonstrating superior stability and adaptability. The implementation of InterCLIP-MEP is publicly available at https://github.com/CoderChen01/InterCLIP-MEP.

Yan Liu, Zhijie Zhang, Li Niu et al.

Object detection has achieved promising success, but requires large-scale fully-annotated data, which is time-consuming and labor-extensive. Therefore, we consider object detection with mixed supervision, which learns novel object categories using weak annotations with the help of full annotations of existing base object categories. Previous works using mixed supervision mainly learn the class-agnostic objectness from fully-annotated categories, which can be transferred to upgrade the weak annotations to pseudo full annotations for novel categories. In this paper, we further transfer mask prior and semantic similarity to bridge the gap between novel categories and base categories. Specifically, the ability of using mask prior to help detect objects is learned from base categories and transferred to novel categories. Moreover, the semantic similarity between objects learned from base categories is transferred to denoise the pseudo full annotations for novel categories. Experimental results on three benchmark datasets demonstrate the effectiveness of our method over existing methods. Codes are available at https://github.com/bcmi/TraMaS-Weak-Shot-Object-Detection.

Junjie Chen, Li Niu, Liu Liu et al.

Recognizing fine-grained categories remains a challenging task, due to the subtle distinctions among different subordinate categories, which results in the need of abundant annotated samples. To alleviate the data-hungry problem, we consider the problem of learning novel categories from web data with the support of a clean set of base categories, which is referred to as weak-shot learning. In this setting, we propose a method called SimTrans to transfer pairwise semantic similarity from base categories to novel categories. Specifically, we firstly train a similarity net on clean data, and then leverage the transferred similarity to denoise web training data using two simple yet effective strategies. In addition, we apply adversarial loss on similarity net to enhance the transferability of similarity. Comprehensive experiments demonstrate the effectiveness of our weak-shot setting and our SimTrans method. Datasets and codes are available at https://github.com/bcmi/SimTrans-Weak-Shot-Classification.

Yuxin Liu, Fei Wang, Kun Li et al.

Image Deepfake Detection (IDD) separates manipulated images from authentic ones by spotting artifacts of synthesis or tampering. Although large vision-language models (LVLMs) offer strong image understanding, adapting them to IDD often demands costly fine-tuning and generalizes poorly to diverse, evolving manipulations. We propose the Semantic Consistent Evidence Pack (SCEP), a training-free LVLM framework that replaces whole-image inference with evidence-driven reasoning. SCEP mines a compact set of suspicious patch tokens that best reveal manipulation cues. It uses the vision encoder's CLS token as a global reference, clusters patch features into coherent groups, and scores patches with a fused metric combining CLS-guided semantic mismatch with frequency-and noise-based anomalies. To cover dispersed traces and avoid redundancy, SCEP samples a few high-confidence patches per cluster and applies grid-based NMS, producing an evidence pack that conditions a frozen LVLM for prediction. Experiments on diverse benchmarks show SCEP outperforms strong baselines without LVLM fine-tuning.

Jiashuo Tian, Dong Wang, Chen Yang et al.

False-positive bug reports represent a significant yet underexplored challenge in the development and maintenance of the Linux kernel. They occur when correct system behavior is mistakenly flagged as a defect, consuming developer effort without leading to actual code improvements. Such reports can mislead developers, waste debugging resources, and delay the resolution of real bugs. In this paper, we present the first comprehensive empirical study of false-positive bug reports in the Linux kernel. We manually construct a dataset of 2,006 bug reports comprising 1,509 genuine bugs and 497 false positives collected from Bugzilla and Syzkaller. Our analysis indicates that false positives demand effort comparable to real bugs, often requiring extended discussions and non-trivial closure time. They occur in several components, especially File Systems and Drivers, mainly due to external dependencies and semantic misunderstandings. To address this challenge, we evaluate large language models (LLMs) for automated false-positive bug report mitigation. Among various prompting strategies, retrieval-augmented generation (RAG) performs best, achieving 91% recall and an F1 score of 88%. These findings highlight the non-negligible cost of false positive bug reports and show the promise of LLMs for more efficient false positive mitigation in the Linux kernel.

Jiangnan Yang, Junjie Chen, Fei Wang et al.

Psychological counseling is a fundamentally multimodal cognitive process in which clinicians integrate verbal content with visual and vocal cues to infer clients' mental states and respond empathically. However, most existing language-model-based counseling systems operate on text alone and rely on implicit mental state inference. We introduce DELTA, a deliberative multi-agent framework that models counseling as a structured reasoning process over multimodal signals, separating evidence grounding, mental state abstraction, and response generation. DELTA further incorporates reinforcement learning guided by a distribution-level Emotion Attunement Score to encourage emotionally attuned responses. Experiments on a multimodal counseling benchmark show that DELTA improves both counseling quality and emotion attunement across models. Ablation and qualitative analyses suggest that explicit multimodal reasoning and structured mental state representations play complementary roles in supporting empathic human-AI interaction.

Zhiyuan Liu, Yicun Yang, Yaojie Zhang et al.

Autoregressive Models (ARMs) have long dominated the landscape of Large Language Models. Recently, a new paradigm has emerged in the form of diffusion-based Large Language Models (dLLMs), which generate text by iteratively denoising masked segments. This approach has shown significant advantages and potential. However, dLLMs suffer from high inference latency. Traditional ARM acceleration techniques, such as Key-Value caching, are incompatible with dLLMs due to their bidirectional attention mechanism. To address this specific challenge, our work begins with a key observation that dLLM inference involves a static prompt and a partially dynamic response, where most tokens remain stable across adjacent denoising steps. Based on this, we propose dLLM-Cache, a training-free adaptive caching framework that combines long-interval prompt caching with partial response updates guided by feature similarity. This design enables efficient reuse of intermediate computations without compromising model performance. Extensive experiments on representative dLLMs, including LLaDA 8B and Dream 7B, show that dLLM-Cache achieves up to 9.1 x speedup over standard inference without compromising output quality. Notably, our method brings dLLM inference latency close to that of ARMs under many settings. Codes are provided in the supplementary material and will be released publicly on GitHub.

Xuyang Liu, Zichen Wen, Shaobo Wang et al.

The advancement of large language models (LLMs) and multi-modal LLMs (MLLMs) has historically relied on scaling model parameters. However, as hardware limits constrain further model growth, the primary computational bottleneck has shifted to the quadratic cost of self-attention over increasingly long sequences by ultra-long text contexts, high-resolution images, and extended videos. In this position paper, \textbf{we argue that the focus of research for efficient artificial intelligence (AI) is shifting from model-centric compression to data-centric compression}. We position data-centric compression as the emerging paradigm, which improves AI efficiency by directly compressing the volume of data processed during model training or inference. To formalize this shift, we establish a unified framework for existing efficiency strategies and demonstrate why it constitutes a crucial paradigm change for long-context AI. We then systematically review the landscape of data-centric compression methods, analyzing their benefits across diverse scenarios. Finally, we outline key challenges and promising future research directions. Our work aims to provide a novel perspective on AI efficiency, synthesize existing efforts, and catalyze innovation to address the challenges posed by ever-increasing context lengths.

Shouyu Yin, Zhao Tian, Junjie Chen et al.

Code generation, which aims to automatically generate source code from given programming requirements, has the potential to substantially improve software development efficiency. With the rapid advancement of large language models (LLMs), LLM-based code generation has attracted widespread attention from both academia and industry. However, as programming requirements become increasingly complex, existing LLMs still exhibit notable performance limitations. To address this challenge, recent studies have proposed training-based curriculum reinforcement learning (CRL) strategies to improve LLM code generation performance. Despite their effectiveness, existing CRL approaches suffer from several limitations, including misaligned requirement difficulty perception, the absence of requirement difficulty optimization, and suboptimal curriculum sampling strategies. In CRL-based code generation, programming requirements serve as the sole input to the model, making their quality and difficulty critical to training effectiveness. Motivated by insights from software requirements engineering, we propose RECRL, a novel requirement-aware curriculum reinforcement learning framework for enhancing LLM-based code generation. RECRL automatically perceives model-specific requirement difficulty, optimizes challenging requirements to improve training data utilization, and employs an adaptive curriculum sampling strategy to construct training batches with smoothly varying difficulty. Extensive experiments on five state-of-the-art LLMs across five widely-used code generation benchmarks by comparing with five state-of-the-art baselines, demonstrate the significant effectiveness of RECRL. For example, RECRL achieves an average Pass@1 improvement of 1.23%-5.62% over all state-of-the-art baselines.

Zheng Chen, Jingkai Wang, Kai Liu et al.

This paper provides a review of the NTIRE 2025 challenge on real-world face restoration, highlighting the proposed solutions and the resulting outcomes. The challenge focuses on generating natural, realistic outputs while maintaining identity consistency. Its goal is to advance state-of-the-art solutions for perceptual quality and realism, without imposing constraints on computational resources or training data. The track of the challenge evaluates performance using a weighted image quality assessment (IQA) score and employs the AdaFace model as an identity checker. The competition attracted 141 registrants, with 13 teams submitting valid models, and ultimately, 10 teams achieved a valid score in the final ranking. This collaborative effort advances the performance of real-world face restoration while offering an in-depth overview of the latest trends in the field.

Haitao Li, Junjie Chen, Qingyao Ai et al.

The use of large language models (LLMs) as automated evaluation tools to assess the quality of generated natural language, known as LLMs-as-Judges, has demonstrated promising capabilities and is rapidly gaining widespread attention. However, when applied to pairwise comparisons of candidate responses, LLM-based evaluators often exhibit selection bias. Specifically, their judgments may become inconsistent when the option positions or ID tokens are swapped, compromising the effectiveness and fairness of the evaluation result. To address this challenge, we introduce CalibraEval, a novel label-free method for mitigating selection bias during inference. Specifically, CalibraEval reformulates debiasing as an optimization task aimed at adjusting observed prediction distributions to align with unbiased prediction distributions. To solve this optimization problem, we propose a non-parametric order-preserving algorithm (NOA). This algorithm leverages the partial order relationships between model prediction distributions, thereby eliminating the need for explicit labels and precise mathematical function modeling.Empirical evaluations of LLMs in multiple representative benchmarks demonstrate that CalibraEval effectively mitigates selection bias and improves performance compared to existing debiasing methods. This work marks a step toward building more robust and unbiased automated evaluation frameworks, paving the way for improved reliability in AI-driven assessments