Dong Zhang

Xin Zhang, Dong Zhang, Shimin Li et al.

Current speech large language models build upon discrete speech representations, which can be categorized into semantic tokens and acoustic tokens. However, existing speech tokens are not specifically designed for speech language modeling. To assess the suitability of speech tokens for building speech language models, we established the first benchmark, SLMTokBench. Our results indicate that neither semantic nor acoustic tokens are ideal for this purpose. Therefore, we propose SpeechTokenizer, a unified speech tokenizer for speech large language models. SpeechTokenizer adopts the Encoder-Decoder architecture with residual vector quantization (RVQ). Unifying semantic and acoustic tokens, SpeechTokenizer disentangles different aspects of speech information hierarchically across different RVQ layers. Furthermore, We construct a Unified Speech Language Model (USLM) leveraging SpeechTokenizer. Experiments show that SpeechTokenizer performs comparably to EnCodec in speech reconstruction and demonstrates strong performance on the SLMTokBench benchmark. Also, USLM outperforms VALL-E in zero-shot Text-to-Speech tasks. Code and models are available at https://github.com/ZhangXInFD/SpeechTokenizer/.

Maxime Gasse, Quentin Cappart, Jonas Charfreitag et al. · deepmind, utoronto

Combinatorial optimization is a well-established area in operations research and computer science. Until recently, its methods have focused on solving problem instances in isolation, ignoring that they often stem from related data distributions in practice. However, recent years have seen a surge of interest in using machine learning as a new approach for solving combinatorial problems, either directly as solvers or by enhancing exact solvers. Based on this context, the ML4CO aims at improving state-of-the-art combinatorial optimization solvers by replacing key heuristic components. The competition featured three challenging tasks: finding the best feasible solution, producing the tightest optimality certificate, and giving an appropriate solver configuration. Three realistic datasets were considered: balanced item placement, workload apportionment, and maritime inventory routing. This last dataset was kept anonymous for the contestants.

Yi Lin, Xiao Fang, Dong Zhang et al.

Recently, the advent of vision Transformer (ViT) has brought substantial advancements in 3D dataset benchmarks, particularly in 3D volumetric medical image segmentation (Vol-MedSeg). Concurrently, multi-layer perceptron (MLP) network has regained popularity among researchers due to their comparable results to ViT, albeit with the exclusion of the resource-intensive self-attention module. In this work, we propose a novel permutable hybrid network for Vol-MedSeg, named PHNet, which capitalizes on the strengths of both convolution neural networks (CNNs) and MLP. PHNet addresses the intrinsic isotropy problem of 3D volumetric data by employing a combination of 2D and 3D CNNs to extract local features. Besides, we propose an efficient multi-layer permute perceptron (MLPP) module that captures long-range dependence while preserving positional information. This is achieved through an axis decomposition operation that permutes the input tensor along different axes, thereby enabling the separate encoding of the positional information. Furthermore, MLPP tackles the resolution sensitivity issue of MLP in Vol-MedSeg with a token segmentation operation, which divides the feature into smaller tokens and processes them individually. Extensive experimental results validate that PHNet outperforms the state-of-the-art methods with lower computational costs on the widely-used yet challenging COVID-19-20 and Synapse benchmarks. The ablation study also demonstrates the effectiveness of PHNet in harnessing the strengths of both CNNs and MLP. The code is available on Github: \href{https://github.com/xiaofang007/PHNet}{https://github.com/xiaofang007/PHNet}.

Dong Zhang, Yi Lin, Hao Chen et al.

Over the past few years, the rapid development of deep learning technologies for computer vision has significantly improved the performance of medical image segmentation (MedISeg). However, the diverse implementation strategies of various models have led to an extremely complex MedISeg system, resulting in a potential problem of unfair result comparisons. In this paper, we collect a series of MedISeg tricks for different model implementation phases (i.e., pre-training model, data pre-processing, data augmentation, model implementation, model inference, and result post-processing), and experimentally explore the effectiveness of these tricks on consistent baselines. With the extensive experimental results on both the representative 2D and 3D medical image datasets, we explicitly clarify the effect of these tricks. Moreover, based on the surveyed tricks, we also open-sourced a strong MedISeg repository, where each component has the advantage of plug-and-play. We believe that this milestone work not only completes a comprehensive and complementary survey of the state-of-the-art MedISeg approaches, but also offers a practical guide for addressing the future medical image processing challenges including but not limited to small dataset, class imbalance learning, multi-modality learning, and domain adaptation. The code and training weights have been released at: https://github.com/hust-linyi/seg_trick.

Yu Quan, Dong Zhang, Liyan Zhang et al.

Visual feature pyramid has shown its superiority in both effectiveness and efficiency in a wide range of applications. However, the existing methods exorbitantly concentrate on the inter-layer feature interactions but ignore the intra-layer feature regulations, which are empirically proved beneficial. Although some methods try to learn a compact intra-layer feature representation with the help of the attention mechanism or the vision transformer, they ignore the neglected corner regions that are important for dense prediction tasks. To address this problem, in this paper, we propose a Centralized Feature Pyramid (CFP) for object detection, which is based on a globally explicit centralized feature regulation. Specifically, we first propose a spatial explicit visual center scheme, where a lightweight MLP is used to capture the globally long-range dependencies and a parallel learnable visual center mechanism is used to capture the local corner regions of the input images. Based on this, we then propose a globally centralized regulation for the commonly-used feature pyramid in a top-down fashion, where the explicit visual center information obtained from the deepest intra-layer feature is used to regulate frontal shallow features. Compared to the existing feature pyramids, CFP not only has the ability to capture the global long-range dependencies, but also efficiently obtain an all-round yet discriminative feature representation. Experimental results on the challenging MS-COCO validate that our proposed CFP can achieve the consistent performance gains on the state-of-the-art YOLOv5 and YOLOX object detection baselines.

Chunyan Wang, Dong Zhang, Liyan Zhang et al.

Thanks to the advantages of the friendly annotations and the satisfactory performance, Weakly-Supervised Semantic Segmentation (WSSS) approaches have been extensively studied. Recently, the single-stage WSSS was awakened to alleviate problems of the expensive computational costs and the complicated training procedures in multi-stage WSSS. However, results of such an immature model suffer from problems of background incompleteness and object incompleteness. We empirically find that they are caused by the insufficiency of the global object context and the lack of the local regional contents, respectively. Under these observations, we propose a single-stage WSSS model with only the image-level class label supervisions, termed as Weakly Supervised Feature Coupling Network (WS-FCN), which can capture the multi-scale context formed from the adjacent feature grids, and encode the fine-grained spatial information from the low-level features into the high-level ones. Specifically, a flexible context aggregation module is proposed to capture the global object context in different granular spaces. Besides, a semantically consistent feature fusion module is proposed in a bottom-up parameter-learnable fashion to aggregate the fine-grained local contents. Based on these two modules, WS-FCN lies in a self-supervised end-to-end training fashion. Extensive experimental results on the challenging PASCAL VOC 2012 and MS COCO 2014 demonstrate the effectiveness and efficiency of WS-FCN, which can achieve state-of-the-art results by 65.02\% and 64.22\% mIoU on PASCAL VOC 2012 val set and test set, 34.12\% mIoU on MS COCO 2014 val set, respectively. The code and weight have been released at:https://github.com/ChunyanWang1/ws-fcn.

Zihang Zhou, Ziqian Ren, Yukai Wu et al.

Functionality-correct repository setup aims to configure execution environments (e.g., dependencies, build scripts) to successfully execute a repository's documented features. It presents significant challenges due to diverse, repository-specific failures, including dependency incompatibilities, missing toolchains, incomplete installations, and verification-strategy mismatches. Existing LLM agents struggle to robustly resolve these issues, specifically failing to support (1) cross-repository experience transfer, (2) multi-step trial-and-repair under non-invertible state changes, and (3) robust verification of setup outcomes to distinguish setup-induced failures from repository bugs. To address this, we introduce SetupX, an experiential learning-based setup framework. First, we construct a Self-Evolving Experience Representation (XPU), a dual-modality knowledge unit encoding setup signals, textual guidance, executable actions to dynamically transfer verified environment fixes to unseen repositories. Second, we employ Experience-Augmented Speculative Execution backed by a LIFO Docker snapshot stack, enabling the agent to proactively trial fixes and safely roll back to known-good states. Third, we introduce a Prosecutor-Judge Verification Protocol that separates evidence collection from final judgment, enabling more reliable setup verification beyond superficial build-time metrics. Evaluation results on carefully-crafted benchmarks show SetupX achieves highest performance (e.g., 92% pass rate) and outperforms the strongest baseline by over 19%. Crucially, SetupX excels in complex multi-repository setup requiring coordinating multiple interconnected services across different containers. The code repository is available at https://github.com/OpenDataBox/SetupX.

Yuchen Shen, Dong Zhang, Zhihao Song et al.

Object detection in aerial images is a fundamental research topic in the geoscience and remote sensing domain. However, the advanced approaches on this topic mainly focus on designing the elaborate backbones or head networks but ignore neck networks. In this letter, we first underline the importance of the neck network in object detection from the perspective of information bottleneck. Then, to alleviate the information deficiency problem in the current approaches, we propose a global semantic network (GSNet), which acts as a bridge from the backbone network to the head network in a bidirectional global pattern. Compared to the existing approaches, our model can capture the rich and enhanced image features with less computational costs. Besides, we further propose a feature fusion refinement module (FRM) for different levels of features, which are suffering from the problem of semantic gap in feature fusion. To demonstrate the effectiveness and efficiency of our approach, experiments are carried out on two challenging and representative aerial image datasets (i.e., DOTA and HRSC2016). Experimental results in terms of accuracy and complexity validate the superiority of our method. The code has been open-sourced at GSNet.

Dong Zhang, Jinhui Tang, Kwang-Ting Cheng

Capturing the long-range dependencies has empirically proven to be effective on a wide range of computer vision tasks. The progressive advances on this topic have been made through the employment of the transformer framework with the help of the multi-head attention mechanism. However, the attention-based image patch interaction potentially suffers from problems of redundant interactions of intra-class patches and unoriented interactions of inter-class patches. In this paper, we propose a novel Graph Reasoning Transformer (GReaT) for image parsing to enable image patches to interact following a relation reasoning pattern. Specifically, the linearly embedded image patches are first projected into the graph space, where each node represents the implicit visual center for a cluster of image patches and each edge reflects the relation weight between two adjacent nodes. After that, global relation reasoning is performed on this graph accordingly. Finally, all nodes including the relation information are mapped back into the original space for subsequent processes. Compared to the conventional transformer, GReaT has higher interaction efficiency and a more purposeful interaction pattern. Experiments are carried out on the challenging Cityscapes and ADE20K datasets. Results show that GReaT achieves consistent performance gains with slight computational overheads on the state-of-the-art transformer baselines.

Zhaowei Li, Wei Wang, YiQing Cai et al.

Significant advancements has recently been achieved in the field of multi-modal large language models (MLLMs), demonstrating their remarkable capabilities in understanding and reasoning across diverse tasks. However, these models are often trained for specific tasks and rely on task-specific input-output formats, limiting their applicability to a broader range of tasks. This raises a fundamental question: Can we develop a unified approach to represent and handle different multi-modal tasks to maximize the generalizability of MLLMs? In this paper, we propose UnifiedMLLM, a comprehensive model designed to represent various tasks using a unified representation. Our model exhibits strong capabilities in comprehending the implicit intent of user instructions and preforming reasoning. In addition to generating textual responses, our model also outputs task tokens and grounding tokens, serving as indicators of task types and task granularity. These outputs are subsequently routed through the task router and directed to specific expert models for task completion. To train our model, we construct a task-specific dataset and an 100k multi-task dataset encompassing complex scenarios. Employing a three-stage training strategy, we equip our model with robust reasoning and task processing capabilities while preserving its generalization capacity and knowledge reservoir. Extensive experiments showcase the impressive performance of our unified representation approach across various tasks, surpassing existing methodologies. Furthermore, our approach exhibits exceptional scalability and generality. Our code, model, and dataset will be available at \url{https://github.com/lzw-lzw/UnifiedMLLM}.

Jeffry Wicaksana, Zengqiang Yan, Dong Zhang et al.

The purpose of federated learning is to enable multiple clients to jointly train a machine learning model without sharing data. However, the existing methods for training an image segmentation model have been based on an unrealistic assumption that the training set for each local client is annotated in a similar fashion and thus follows the same image supervision level. To relax this assumption, in this work, we propose a label-agnostic unified federated learning framework, named FedMix, for medical image segmentation based on mixed image labels. In FedMix, each client updates the federated model by integrating and effectively making use of all available labeled data ranging from strong pixel-level labels, weak bounding box labels, to weakest image-level class labels. Based on these local models, we further propose an adaptive weight assignment procedure across local clients, where each client learns an aggregation weight during the global model update. Compared to the existing methods, FedMix not only breaks through the constraint of a single level of image supervision, but also can dynamically adjust the aggregation weight of each local client, achieving rich yet discriminative feature representations. To evaluate its effectiveness, experiments have been carried out on two challenging medical image segmentation tasks, i.e., breast tumor segmentation and skin lesion segmentation. The results validate that our proposed FedMix outperforms the state-of-the-art method by a large margin.

Xiaomi LLM-Core Team, Dong Zhang, Gang Wang et al.

Existing audio language models typically rely on task-specific fine-tuning to accomplish particular audio tasks. In contrast, humans are able to generalize to new audio tasks with only a few examples or simple instructions. GPT-3 has shown that scaling next-token prediction pretraining enables strong generalization capabilities in text, and we believe this paradigm is equally applicable to the audio domain. By scaling MiMo-Audio's pretraining data to over one hundred million of hours, we observe the emergence of few-shot learning capabilities across a diverse set of audio tasks. We develop a systematic evaluation of these capabilities and find that MiMo-Audio-7B-Base achieves SOTA performance on both speech intelligence and audio understanding benchmarks among open-source models. Beyond standard metrics, MiMo-Audio-7B-Base generalizes to tasks absent from its training data, such as voice conversion, style transfer, and speech editing. MiMo-Audio-7B-Base also demonstrates powerful speech continuation capabilities, capable of generating highly realistic talk shows, recitations, livestreaming and debates. At the post-training stage, we curate a diverse instruction-tuning corpus and introduce thinking mechanisms into both audio understanding and generation. MiMo-Audio-7B-Instruct achieves open-source SOTA on audio understanding benchmarks (MMSU, MMAU, MMAR, MMAU-Pro), spoken dialogue benchmarks (Big Bench Audio, MultiChallenge Audio) and instruct-TTS evaluations, approaching or surpassing closed-source models. Model checkpoints and full evaluation suite are available at https://github.com/XiaomiMiMo/MiMo-Audio.

Xixi Jiang, Dong Zhang, Xiang Li et al.

Partially-supervised multi-organ medical image segmentation aims to develop a unified semantic segmentation model by utilizing multiple partially-labeled datasets, with each dataset providing labels for a single class of organs. However, the limited availability of labeled foreground organs and the absence of supervision to distinguish unlabeled foreground organs from the background pose a significant challenge, which leads to a distribution mismatch between labeled and unlabeled pixels. Although existing pseudo-labeling methods can be employed to learn from both labeled and unlabeled pixels, they are prone to performance degradation in this task, as they rely on the assumption that labeled and unlabeled pixels have the same distribution. In this paper, to address the problem of distribution mismatch, we propose a labeled-to-unlabeled distribution alignment (LTUDA) framework that aligns feature distributions and enhances discriminative capability. Specifically, we introduce a cross-set data augmentation strategy, which performs region-level mixing between labeled and unlabeled organs to reduce distribution discrepancy and enrich the training set. Besides, we propose a prototype-based distribution alignment method that implicitly reduces intra-class variation and increases the separation between the unlabeled foreground and background. This can be achieved by encouraging consistency between the outputs of two prototype classifiers and a linear classifier. Extensive experimental results on the AbdomenCT-1K dataset and a union of four benchmark datasets (including LiTS, MSD-Spleen, KiTS, and NIH82) demonstrate that our method outperforms the state-of-the-art partially-supervised methods by a considerable margin, and even surpasses the fully-supervised methods. The source code is publicly available at https://github.com/xjiangmed/LTUDA.

Zenan Shi, Haipeng Chen, Long Chen et al.

In this paper, we propose a novel image forgery detection paradigm for boosting the model learning capacity on both forgery-sensitive and genuine compact visual patterns. Compared to the existing methods that only focus on the discrepant-specific patterns (\eg, noises, textures, and frequencies), our method has a greater generalization. Specifically, we first propose a Discrepancy-Guided Encoder (DisGE) to extract forgery-sensitive visual patterns. DisGE consists of two branches, where the mainstream backbone branch is used to extract general semantic features, and the accessorial discrepant external attention branch is used to extract explicit forgery cues. Besides, a Double-Head Reconstruction (DouHR) module is proposed to enhance genuine compact visual patterns in different granular spaces. Under DouHR, we further introduce a Discrepancy-Aggregation Detector (DisAD) to aggregate these genuine compact visual patterns, such that the forgery detection capability on unknown patterns can be improved. Extensive experimental results on four challenging datasets validate the effectiveness of our proposed method against state-of-the-art competitors.

Pengyu Wang, Linyang Li, Ke Ren et al.

Widely applied large language models (LLMs) can generate human-like content, raising concerns about the abuse of LLMs. Therefore, it is important to build strong AI-generated text (AIGT) detectors. Current works only consider document-level AIGT detection, therefore, in this paper, we first introduce a sentence-level detection challenge by synthesizing a dataset that contains documents that are polished with LLMs, that is, the documents contain sentences written by humans and sentences modified by LLMs. Then we propose \textbf{Seq}uence \textbf{X} (Check) \textbf{GPT}, a novel method that utilizes log probability lists from white-box LLMs as features for sentence-level AIGT detection. These features are composed like \textit{waves} in speech processing and cannot be studied by LLMs. Therefore, we build SeqXGPT based on convolution and self-attention networks. We test it in both sentence and document-level detection challenges. Experimental results show that previous methods struggle in solving sentence-level AIGT detection, while our method not only significantly surpasses baseline methods in both sentence and document-level detection challenges but also exhibits strong generalization capabilities.

Yangjun Mao, Long Chen, Zhihong Jiang et al.

Distinctive Image Captioning (DIC) -- generating distinctive captions that describe the unique details of a target image -- has received considerable attention over the last few years. A recent DIC work proposes to generate distinctive captions by comparing the target image with a set of semantic-similar reference images, i.e., reference-based DIC (Ref-DIC). It aims to make the generated captions can tell apart the target and reference images. Unfortunately, reference images used by existing Ref-DIC works are easy to distinguish: these reference images only resemble the target image at scene-level and have few common objects, such that a Ref-DIC model can trivially generate distinctive captions even without considering the reference images. To ensure Ref-DIC models really perceive the unique objects (or attributes) in target images, we first propose two new Ref-DIC benchmarks. Specifically, we design a two-stage matching mechanism, which strictly controls the similarity between the target and reference images at object-/attribute- level (vs. scene-level). Secondly, to generate distinctive captions, we develop a strong Transformer-based Ref-DIC baseline, dubbed as TransDIC. It not only extracts visual features from the target image, but also encodes the differences between objects in the target and reference images. Finally, for more trustworthy benchmarking, we propose a new evaluation metric named DisCIDEr for Ref-DIC, which evaluates both the accuracy and distinctiveness of the generated captions. Experimental results demonstrate that our TransDIC can generate distinctive captions. Besides, it outperforms several state-of-the-art models on the two new benchmarks over different metrics.

Fengyun Wang, Dong Zhang, Hanwang Zhang et al.

Semantic Scene Completion (SSC) transforms an image of single-view depth and/or RGB 2D pixels into 3D voxels, each of whose semantic labels are predicted. SSC is a well-known ill-posed problem as the prediction model has to "imagine" what is behind the visible surface, which is usually represented by Truncated Signed Distance Function (TSDF). Due to the sensory imperfection of the depth camera, most existing methods based on the noisy TSDF estimated from depth values suffer from 1) incomplete volumetric predictions and 2) confused semantic labels. To this end, we use the ground-truth 3D voxels to generate a perfect visible surface, called TSDF-CAD, and then train a "cleaner" SSC model. As the model is noise-free, it is expected to focus more on the "imagination" of unseen voxels. Then, we propose to distill the intermediate "cleaner" knowledge into another model with noisy TSDF input. In particular, we use the 3D occupancy feature and the semantic relations of the "cleaner self" to supervise the counterparts of the "noisy self" to respectively address the above two incorrect predictions. Experimental results validate that our method improves the noisy counterparts with 3.1% IoU and 2.2% mIoU for measuring scene completion and SSC, and also achieves new state-of-the-art accuracy on the popular NYU dataset.

Chenchen Yang, Kexin Huang, Liwei Fan et al.

Speech conveys not only linguistic information but also rich non-verbal vocal events such as laughing and crying. While semantic transcription is well-studied, the precise localization of non-verbal events remains a critical yet under-explored challenge. Current methods suffer from insufficient task definitions with limited category coverage and ambiguous temporal granularity. They also lack standardized evaluation frameworks, hindering the development of downstream applications. To bridge this gap, we first develop a refined taxonomy of 21 vocal events, with a new categorization into discrete (standalone) versus continuous (mixed with speech) types. Based on the refined taxonomy, we introduce WESR-Bench, an expert-annotated evaluation set (900+ utterances) with a novel position-aware protocol that disentangles ASR errors from event detection, enabling precise localization measurement for both discrete and continuous events. We also build a strong baseline by constructing a 1,700+ hour corpus, and train specialized models, surpassing both open-source audio-language models and commercial APIs while preserving ASR quality. We anticipate that WESR will serve as a foundational resource for future research in modeling rich, real-world auditory scenes.

Yangjun Mao, Jun Xiao, Dong Zhang et al.

Distinctive Image Captioning (DIC) -- generating distinctive captions that describe the unique details of a target image -- has received considerable attention over the last few years. A recent DIC method proposes to generate distinctive captions by comparing the target image with a set of semantic-similar reference images, i.e., reference-based DIC (Ref-DIC). It aims to force the generated captions to distinguish between the target image and the reference image. To ensure Ref-DIC models really perceive the unique objects (or attributes) in target images, we propose two new Ref-DIC benchmarks and develop a Transformer-based Ref-DIC baseline TransDIC. The model only extracts visual features from the target image, but also encodes the differences between objects in the target and reference images. Taking one step further, we propose a stronger TransDIC++, which consists of an extra contrastive learning module to make full use of the reference images. This new module is model-agnostic, which can be easily incorporated into various Ref-DIC architectures. Finally, for more trustworthy benchmarking, we propose a new evaluation metric named DisCIDEr for Ref-DIC, which evaluates both the accuracy and distinctiveness of the generated captions. Experimental results demonstrate that our TransDIC++ can generate distinctive captions. Besides, it outperforms several state-of-the-art models on the two new benchmarks over different metrics.

Xiao Fang, Yi Lin, Dong Zhang et al.

Pre-trained large vision-language models (VLMs) like CLIP have revolutionized visual representation learning using natural language as supervisions, and demonstrated promising generalization ability. In this work, we propose ViP, a novel visual symptom-guided prompt learning framework for medical image analysis, which facilitates general knowledge transfer from CLIP. ViP consists of two key components: a visual symptom generator (VSG) and a dual-prompt network. Specifically, VSG aims to extract explicable visual symptoms from pre-trained large language models, while the dual-prompt network utilizes these visual symptoms to guide the training on two learnable prompt modules, i.e., context prompt and merge prompt, which effectively adapts our framework to medical image analysis via large VLMs. Extensive experimental results demonstrate that ViP can outperform state-of-the-art methods on two challenging datasets.

Meng Zhao, Yifan Hu, Ruixuan Jiang et al.

Background: To develop an artificial intelligence system that can accurately identify acute non-traumatic intracranial hemorrhage (ICH) etiology based on non-contrast CT (NCCT) scans and investigate whether clinicians can benefit from it in a diagnostic setting. Materials and Methods: The deep learning model was developed with 1868 eligible NCCT scans with non-traumatic ICH collected between January 2011 and April 2018. We tested the model on two independent datasets (TT200 and SD 98) collected after April 2018. The model's diagnostic performance was compared with clinicians's performance. We further designed a simulated study to compare the clinicians's performance with and without the deep learning system augmentation. Results: The proposed deep learning system achieved area under the receiver operating curve of 0.986 (95% CI 0.967-1.000) on aneurysms, 0.952 (0.917-0.987) on hypertensive hemorrhage, 0.950 (0.860-1.000) on arteriovenous malformation (AVM), 0.749 (0.586-0.912) on Moyamoya disease (MMD), 0.837 (0.704-0.969) on cavernous malformation (CM), and 0.839 (0.722-0.959) on other causes in TT200 dataset. Given a 90% specificity level, the sensitivities of our model were 97.1% and 90.9% for aneurysm and AVM diagnosis, respectively. The model also shows an impressive generalizability in an independent dataset SD98. The clinicians achieve significant improvements in the sensitivity, specificity, and accuracy of diagnoses of certain hemorrhage etiologies with proposed system augmentation. Conclusions: The proposed deep learning algorithms can be an effective tool for early identification of hemorrhage etiologies based on NCCT scans. It may also provide more information for clinicians for triage and further imaging examination selection.

Yuchen Shen, Dong Zhang, Zhao Zhang et al.

In recent years, instance segmentation has garnered significant attention across various applications. However, training a fully-supervised instance segmentation model requires costly both instance-level and pixel-level annotations. In contrast, weakly-supervised instance segmentation methods, such as those using image-level class labels or point labels, often struggle to satisfy the accuracy and recall requirements of practical scenarios. In this paper, we propose a novel paradigm called Synthetic Instance Segmentation (SISeg). SISeg achieves instance segmentation results by leveraging image masks generated by existing semantic segmentation models, and it is highly efficient as we do not require additional training for semantic segmentation or the use of instance-level image annotations. In other words, the proposed model does not need extra manpower or higher computational expenses. Specifically, we first obtain a semantic segmentation mask of the input image via an existent semantic segmentation model. Then, we calculate a displacement field vector for each pixel based on the segmentation mask, which can indicate representations belonging to the same class but different instances, i.e., obtaining the instance-level object information. Finally, the instance segmentation results are refined by a learnable category-agnostic object boundary branch. Extensive experimental results on two challenging datasets highlight the effectiveness of SISeg in achieving competitive results when compared to state-of-the-art methods, especially fully-supervised methods. The code will be released at: SISeg

Zhaowei Li, Qi Xu, Dong Zhang et al.

Multi-modal large language models have demonstrated impressive performance across various tasks in different modalities. However, existing multi-modal models primarily emphasize capturing global information within each modality while neglecting the importance of perceiving local information across modalities. Consequently, these models lack the ability to effectively understand the fine-grained details of input data, limiting their performance in tasks that require a more nuanced understanding. To address this limitation, there is a compelling need to develop models that enable fine-grained understanding across multiple modalities, thereby enhancing their applicability to a wide range of tasks. In this paper, we propose GroundingGPT, a language enhanced multi-modal grounding model. Beyond capturing global information like other multi-modal models, our proposed model excels at tasks demanding a detailed understanding of local information within the input. It demonstrates precise identification and localization of specific regions in images or moments in videos. To achieve this objective, we design a diversified dataset construction pipeline, resulting in a multi-modal, multi-granularity dataset for model training. The code, dataset, and demo of our model can be found at https: //github.com/lzw-lzw/GroundingGPT.

Renqiang Luo, Dong Zhang, Yupeng Gao et al.

Semantic understanding of popularity bias is a crucial yet underexplored challenge in recommender systems, where popular items are often favored at the expense of niche content. Most existing debiasing methods treat the semantic understanding of popularity bias as a matter of diversity enhancement or long-tail coverage, neglecting the deeper semantic layer that embodies the causal origins of the bias itself. Consequently, such shallow interpretations limit both their debiasing effectiveness and recommendation accuracy. In this paper, we propose FairLRM, a novel framework that bridges the gap in the semantic understanding of popularity bias with Recommendation via Large Language Model (RecLLM). FairLRM decomposes popularity bias into item-side and user-side components, using structured instruction-based prompts to enhance the model's comprehension of both global item distributions and individual user preferences. Unlike traditional methods that rely on surface-level features such as "diversity" or "debiasing", FairLRM improves the model's ability to semantically interpret and address the underlying bias. Through empirical evaluation, we show that FairLRM significantly enhances both fairness and recommendation accuracy, providing a more semantically aware and trustworthy approach to enhance the semantic understanding of popularity bias. The implementation is available at https://github.com/LuoRenqiang/FairLRM.

Li Guo, Anas M. Tahir, Dong Zhang et al.

The increasing demand for medical imaging has surpassed the capacity of available radiologists, leading to diagnostic delays and potential misdiagnoses. Artificial intelligence (AI) techniques, particularly in automatic medical report generation (AMRG), offer a promising solution to this dilemma. This review comprehensively examines AMRG methods from 2021 to 2024. It (i) presents solutions to primary challenges in this field, (ii) explores AMRG applications across various imaging modalities, (iii) introduces publicly available datasets, (iv) outlines evaluation metrics, (v) identifies techniques that significantly enhance model performance, and (vi) discusses unresolved issues and potential future research directions. This paper aims to provide a comprehensive understanding of the existing literature and inspire valuable future research.

Dong Zhang, Zhaowei Li, Shimin Li et al.

Speech language models have significantly advanced in generating realistic speech, with neural codec language models standing out. However, the integration of human feedback to align speech outputs to human preferences is often neglected. This paper addresses this gap by first analyzing the distribution gap in codec language models, highlighting how it leads to discrepancies between the training and inference phases, which negatively affects performance. Then we explore leveraging learning from human feedback to bridge the distribution gap. We introduce SpeechAlign, an iterative self-improvement strategy that aligns speech language models to human preferences. SpeechAlign involves constructing a preference codec dataset contrasting golden codec tokens against synthetic tokens, followed by preference optimization to improve the codec language model. This cycle of improvement is carried out iteratively to steadily convert weak models to strong ones. Through both subjective and objective evaluations, we show that SpeechAlign can bridge the distribution gap and facilitating continuous self-improvement of the speech language model. Moreover, SpeechAlign exhibits robust generalization capabilities and works for smaller models. Code and models will be available at https://github.com/0nutation/SpeechGPT.

LLM-Core Xiaomi, Bingquan Xia, Bowen Shen et al. · pku

We present MiMo-7B, a large language model born for reasoning tasks, with optimization across both pre-training and post-training stages. During pre-training, we enhance the data preprocessing pipeline and employ a three-stage data mixing strategy to strengthen the base model's reasoning potential. MiMo-7B-Base is pre-trained on 25 trillion tokens, with additional Multi-Token Prediction objective for enhanced performance and accelerated inference speed. During post-training, we curate a dataset of 130K verifiable mathematics and programming problems for reinforcement learning, integrating a test-difficulty-driven code-reward scheme to alleviate sparse-reward issues and employing strategic data resampling to stabilize training. Extensive evaluations show that MiMo-7B-Base possesses exceptional reasoning potential, outperforming even much larger 32B models. The final RL-tuned model, MiMo-7B-RL, achieves superior performance on mathematics, code and general reasoning tasks, surpassing the performance of OpenAI o1-mini. The model checkpoints are available at https://github.com/xiaomimimo/MiMo.

Ziliang Gan, Yu Lu, Dong Zhang et al.

In recent years, multimodal benchmarks for general domains have guided the rapid development of multimodal models on general tasks. However, the financial field has its peculiarities. It features unique graphical images (e.g., candlestick charts, technical indicator charts) and possesses a wealth of specialized financial knowledge (e.g., futures, turnover rate). Therefore, benchmarks from general fields often fail to measure the performance of multimodal models in the financial domain, and thus cannot effectively guide the rapid development of large financial models. To promote the development of large financial multimodal models, we propose MME-Finance, an bilingual open-ended and practical usage-oriented Visual Question Answering (VQA) benchmark. The characteristics of our benchmark are finance and expertise, which include constructing charts that reflect the actual usage needs of users (e.g., computer screenshots and mobile photography), creating questions according to the preferences in financial domain inquiries, and annotating questions by experts with 10+ years of experience in the financial industry. Additionally, we have developed a custom-designed financial evaluation system in which visual information is first introduced in the multi-modal evaluation process. Extensive experimental evaluations of 19 mainstream MLLMs are conducted to test their perception, reasoning, and cognition capabilities. The results indicate that models performing well on general benchmarks cannot do well on MME-Finance; for instance, the top-performing open-source and closed-source models obtain 65.69 (Qwen2VL-72B) and 63.18 (GPT-4o), respectively. Their performance is particularly poor in categories most relevant to finance, such as candlestick charts and technical indicator charts. In addition, we propose a Chinese version, which helps compare performance of MLLMs under a Chinese context.

Bin Wang, Wenjie Yu, Yilu Zhong et al.

Large language models (LLMs) for code generation are becoming integral to modern software development, but their real-world prevalence and security impact remain poorly understood. We present the first large-scale empirical study of AI-generated code (AIGCode) in the wild. We build a high-precision detection pipeline and a representative benchmark to distinguish AIGCode from human-written code, and apply them to (i) development commits from the top 1,000 GitHub repositories (2022-2025) and (ii) 7,000+ recent CVE-linked code changes. This lets us label commits, files, and functions along a human/AI axis and trace how AIGCode moves through projects and vulnerability life cycles. Our measurements show three ecological patterns. First, AIGCode is already a substantial fraction of new code, but adoption is structured: AI concentrates in glue code, tests, refactoring, documentation, and other boilerplate, while core logic and security-critical configurations remain mostly human-written. Second, adoption has security consequences: some CWE families are overrepresented in AI-tagged code, and near-identical insecure templates recur across unrelated projects, suggesting "AI-induced vulnerabilities" propagated by shared models rather than shared maintainers. Third, in human-AI edit chains, AI introduces high-throughput changes while humans act as security gatekeepers; when review is shallow, AI-introduced defects persist longer, remain exposed on network-accessible surfaces, and spread to more files and repositories. We will open-source the complete dataset and release analysis artifacts and fine-grained documentation of our methodology and findings.

Bin Wang, Hui Li, Liyang Zhang et al.

Sensitive information leakage in code repositories has emerged as a critical security challenge. Traditional detection methods that rely on regular expressions, fingerprint features, and high-entropy calculations often suffer from high false-positive rates. This not only reduces detection efficiency but also significantly increases the manual screening burden on developers. Recent advances in large language models (LLMs) and multi-agent collaborative architectures have demonstrated remarkable potential for tackling complex tasks, offering a novel technological perspective for sensitive information detection. In response to these challenges, we propose Argus, a multi-agent collaborative framework for detecting sensitive information. Argus employs a three-tier detection mechanism that integrates key content, file context, and project reference relationships to effectively reduce false positives and enhance overall detection accuracy. To comprehensively evaluate Argus in real-world repository environments, we developed two new benchmarks, one to assess genuine leak detection capabilities and another to evaluate false-positive filtering performance. Experimental results show that Argus achieves up to 94.86% accuracy in leak detection, with a precision of 96.36%, recall of 94.64%, and an F1 score of 0.955. Moreover, the analysis of 97 real repositories incurred a total cost of only 2.2$. All code implementations and related datasets are publicly available at https://github.com/TheBinKing/Argus-Guard for further research and application.

Prabin Bohara, Pralhad Kumar Shrestha, Arpan Rai et al.

Accurate automatic brain tumor segmentation in Low and Middle-Income (LMIC) countries is challenging due to the lack of defined national imaging protocols, diverse imaging data, extensive use of low-field Magnetic Resonance Imaging (MRI) scanners and limited health-care resources. As part of the Brain Tumor Segmentation (BraTS) Africa 2025 Challenge, we applied topology refinement to the state-of-the-art segmentation models like nnU-Net, MedNeXt, and a combination of both. Since the BraTS-Africa dataset has low MRI image quality, we incorporated the BraTS 2025 challenge data of pre-treatment adult glioma (Task 1) to pre-train the segmentation model and use it to fine-tune on the BraTS-Africa dataset. We added an extra topology refinement module to address the issue of deformation in prediction that arose due to topological error. With the introduction of this module, we achieved a better Normalized Surface Distance (NSD) of 0.810, 0.829, and 0.895 on Surrounding Non-Enhancing FLAIR Hyperintensity (SNFH) , Non-Enhancing Tumor Core (NETC) and Enhancing tumor (ET).

Tian Dong, Yanjun Chen, Shoufeng Zhang et al.

AI infra has become a shared execution layer for model training, deployment, and agent orchestration. Because many projects reimplement similar model-centric workflows, a vulnerability disclosed in one repository can recur as a variant in another repository with a related design. Yet the prevalence and detectability of these variants remain poorly understood. This paper presents a measurement study of vulnerability variants in AI infra. Analyzing 688 GitHub repositories and 251 publicly disclosed vulnerabilities, we find that AI infra projects frequently share overlapping functionality and recurrent vulnerable patterns, creating a concrete basis for cross-repository variants. Building on this finding, we study how to automatically identify such variants from known disclosures. We propose INFRASCOPE, a reference-driven multi-agent framework that extracts transferable vulnerability semantics from known cases and uses them to locate and validate variants in new repositories. Evaluating INFRASCOPE on 20 real-world AI infra repositories, we uncover over 20 vulnerabilities, including 11 acknowledged cases and 4 cases that have been assigned CVEs so far.

Yi Lin, Zeyu Wang, Dong Zhang et al.

Nuclei segmentation is a fundamental prerequisite in the digital pathology workflow. The development of automated methods for nuclei segmentation enables quantitative analysis of the wide existence and large variances in nuclei morphometry in histopathology images. However, manual annotation of tens of thousands of nuclei is tedious and time-consuming, which requires significant amount of human effort and domain-specific expertise. To alleviate this problem, in this paper, we propose a weakly-supervised nuclei segmentation method that only requires partial point labels of nuclei. Specifically, we propose a novel boundary mining framework for nuclei segmentation, named BoNuS, which simultaneously learns nuclei interior and boundary information from the point labels. To achieve this goal, we propose a novel boundary mining loss, which guides the model to learn the boundary information by exploring the pairwise pixel affinity in a multiple-instance learning manner. Then, we consider a more challenging problem, i.e., partial point label, where we propose a nuclei detection module with curriculum learning to detect the missing nuclei with prior morphological knowledge. The proposed method is validated on three public datasets, MoNuSeg, CPM, and CoNIC datasets. Experimental results demonstrate the superior performance of our method to the state-of-the-art weakly-supervised nuclei segmentation methods. Code: https://github.com/hust-linyi/bonus.

Yitian Gong, Luozhijie Jin, Ruifan Deng et al.

Speech codecs serve as bridges between speech signals and large language models. An ideal codec for speech language models should not only preserve acoustic information but also capture rich semantic information. However, existing speech codecs struggle to balance high-quality audio reconstruction with ease of modeling by language models. In this study, we analyze the limitations of previous codecs in balancing semantic richness and acoustic fidelity. We propose XY-Tokenizer, a novel codec that mitigates the conflict between semantic and acoustic capabilities through multi-stage, multi-task learning. Experimental results demonstrate that XY-Tokenizer achieves performance in both semantic and acoustic tasks comparable to that of state-of-the-art codecs operating at similar bitrates, even though those existing codecs typically excel in only one aspect. Specifically, XY-Tokenizer achieves strong text alignment, surpassing distillation-based semantic modeling methods such as SpeechTokenizer and Mimi, while maintaining a speaker similarity score of 0.83 between reconstructed and original audio. The reconstruction performance of XY-Tokenizer is comparable to that of BigCodec, the current state-of-the-art among acoustic-only codecs, which achieves a speaker similarity score of 0.84 at a similar bitrate. Code and models are available at https://github.com/gyt1145028706/XY-Tokenizer.

Yun Zhu, Dong Zhang, Yi Lin et al.

Medical image segmentation demands the aggregation of global and local feature representations, posing a challenge for current methodologies in handling both long-range and short-range feature interactions. Recently, vision mamba (ViM) models have emerged as promising solutions for addressing model complexities by excelling in long-range feature iterations with linear complexity. However, existing ViM approaches overlook the importance of preserving short-range local dependencies by directly flattening spatial tokens and are constrained by fixed scanning patterns that limit the capture of dynamic spatial context information. To address these challenges, we introduce a simple yet effective method named context clustering ViM (CCViM), which incorporates a context clustering module within the existing ViM models to segment image tokens into distinct windows for adaptable local clustering. Our method effectively combines long-range and short-range feature interactions, thereby enhancing spatial contextual representations for medical image segmentation tasks. Extensive experimental evaluations on diverse public datasets, i.e., Kumar, CPM17, ISIC17, ISIC18, and Synapse demonstrate the superior performance of our method compared to current state-of-the-art methods. Our code can be found at https://github.com/zymissy/CCViM.

Pingcheng Dong, Yonghao Tan, Dong Zhang et al.

Non-linear functions are prevalent in Transformers and their lightweight variants, incurring substantial and frequently underestimated hardware costs. Previous state-of-the-art works optimize these operations by piece-wise linear approximation and store the parameters in look-up tables (LUT), but most of them require unfriendly high-precision arithmetics such as FP/INT 32 and lack consideration of integer-only INT quantization. This paper proposed a genetic LUT-Approximation algorithm namely GQA-LUT that can automatically determine the parameters with quantization awareness. The results demonstrate that GQA-LUT achieves negligible degradation on the challenging semantic segmentation task for both vanilla and linear Transformer models. Besides, proposed GQA-LUT enables the employment of INT8-based LUT-Approximation that achieves an area savings of 81.3~81.7% and a power reduction of 79.3~80.2% compared to the high-precision FP/INT 32 alternatives. Code is available at https:// github.com/PingchengDong/GQA-LUT.

Xiaomi LLM-Core Team, Zihao Yue, Zhenru Lin et al. · pku

We open-source MiMo-VL-7B-SFT and MiMo-VL-7B-RL, two powerful vision-language models delivering state-of-the-art performance in both general visual understanding and multimodal reasoning. MiMo-VL-7B-RL outperforms Qwen2.5-VL-7B on 35 out of 40 evaluated tasks, and scores 59.4 on OlympiadBench, surpassing models with up to 78B parameters. For GUI grounding applications, it sets a new standard with 56.1 on OSWorld-G, even outperforming specialized models such as UI-TARS. Our training combines four-stage pre-training (2.4 trillion tokens) with Mixed On-policy Reinforcement Learning (MORL) integrating diverse reward signals. We identify the importance of incorporating high-quality reasoning data with long Chain-of-Thought into pre-training stages, and the benefits of mixed RL despite challenges in simultaneous multi-domain optimization. We also contribute a comprehensive evaluation suite covering 50+ tasks to promote reproducibility and advance the field. The model checkpoints and full evaluation suite are available at https://github.com/XiaomiMiMo/MiMo-VL.

Yang Liu, Tongfei Shen, Dong Zhang et al.

The growing importance of multi-modal humor detection within affective computing correlates with the expanding influence of short-form video sharing on social media platforms. In this paper, we propose a novel two-branch hierarchical model for short-form video humor detection (SVHD), named Comment-aided Video-Language Alignment (CVLA) via data-augmented multi-modal contrastive pre-training. Notably, our CVLA not only operates on raw signals across various modal channels but also yields an appropriate multi-modal representation by aligning the video and language components within a consistent semantic space. The experimental results on two humor detection datasets, including DY11k and UR-FUNNY, demonstrate that CVLA dramatically outperforms state-of-the-art and several competitive baseline approaches. Our dataset, code and model release at https://github.com/yliu-cs/CVLA.

Dong Zhang, Zhaowei Li, Pengyu Wang et al.

Human communication is a complex and diverse process that not only involves multiple factors such as language, commonsense, and cultural backgrounds but also requires the participation of multimodal information, such as speech. Large Language Model (LLM)-based multi-agent systems have demonstrated promising performance in simulating human society. Can we leverage LLM-based multi-agent systems to simulate human communication? However, current LLM-based multi-agent systems mainly rely on text as the primary medium. In this paper, we propose SpeechAgents, a multi-modal LLM based multi-agent system designed for simulating human communication. SpeechAgents utilizes multi-modal LLM as the control center for individual agent and employes multi-modal signals as the medium for exchanged messages among agents. Additionally, we propose Multi-Agent Tuning to enhance the multi-agent capabilities of LLM without compromising general abilities. To strengthen and evaluate the effectiveness of human communication simulation, we build the Human-Communication Simulation Benchmark. Experimental results demonstrate that SpeechAgents can simulate human communication dialogues with consistent content, authentic rhythm, and rich emotions and demonstrate excellent scalability even with up to 25 agents, which can apply to tasks such as drama creation and audio novels generation. Code and models will be open-sourced at https://github. com/0nutation/SpeechAgents

Dongjie Huo, Junhui Wang, Chao Gao et al.

Mobile robots operating in human-centered environments must generate not only collision-free paths but also trajectories that follow local behavioral conventions. Conventional costmap-based navigation emphasizes geometric feasibility and often overlooks such requirements, which can result in socially inappropriate behaviors. This paper presents NORM-Nav, a zero-shot framework that integrates natural language behavioral constraints into costmap-based planning. An LLM parses each instruction into structured constraints and grounds them using real-time vision--LiDAR perception. These constraints are encoded as multi-layer costmaps that represent geometric, semantic, directional, and velocity cues and are directly compatible with standard grid-based planners. Simulation and real-world experiments indicate that NORM-Nav improves task success rates and produces trajectories closer to human references than representative baselines. The project website is available at https://ei-nav.github.io/NORM-Nav.

Xinghao Wang, Pengyu Wang, Bo Wang et al.

Large language models (LLMs) have revolutionized numerous applications, yet their deployment remains challenged by memory constraints on local devices. While scaling laws have enhanced LLM capabilities, the primary bottleneck has shifted from \textit{capability} to \textit{availability}, emphasizing the need for efficient memory management. Traditional compression methods, such as quantization, often require predefined compression ratios and separate compression processes for each setting, complicating deployment in variable memory environments. In this paper, we introduce \textbf{BitStack}, a novel, training-free weight compression approach that enables megabyte-level trade-offs between memory usage and model performance. By leveraging weight decomposition, BitStack can dynamically adjust the model size with minimal transmission between running memory and storage devices. Our approach iteratively decomposes weight matrices while considering the significance of each parameter, resulting in an approximately 1-bit per parameter residual block in each decomposition iteration. These blocks are sorted and stacked in storage as basic transmission units, with different quantities loaded based on current memory availability. Extensive experiments across a wide range of tasks demonstrate that, despite offering fine-grained size control, BitStack consistently matches or surpasses strong quantization baselines, particularly at extreme compression ratios. To the best of our knowledge, this is the first decomposition-based method that effectively bridges the gap to practical compression techniques like quantization. Code is available at https://github.com/xinghaow99/BitStack.

Yuhan Guo, Cong Guo, Aiwen Sun et al.

Multimodal large-scale models have significantly advanced the development of web agents, enabling perception and interaction with digital environments akin to human cognition. In this paper, we argue that web agents must first acquire sufficient knowledge to effectively engage in cognitive reasoning. Therefore, we decompose a web agent's capabilities into two essential stages: knowledge content learning and cognitive processes. To formalize this, we propose Web-CogKnowledge Framework, categorizing knowledge as Factual, Conceptual, and Procedural. In this framework, knowledge content learning corresponds to the agent's processes of Memorizing and Understanding, which rely on the first two knowledge types, representing the "what" of learning. Conversely, cognitive processes correspond to Exploring, grounded in Procedural knowledge, defining the "how" of reasoning and action. To facilitate knowledge acquisition, we construct the Web-CogDataset, a structured resource curated from 14 real-world websites, designed to systematically instill core knowledge necessary for web agent. This dataset serves as the agent's conceptual grounding-the "nouns" upon which comprehension is built-as well as the basis for learning how to reason and act. Building on this foundation, we operationalize these processes through a novel knowledge-driven Chain-of-Thought (CoT) reasoning framework, developing and training our proposed agent, the Web-CogReasoner. Extensive experimentation reveals its significant superiority over existing models, especially in generalizing to unseen tasks where structured knowledge is decisive. To enable rigorous evaluation, we introduce the Web-CogBench, a comprehensive evaluation suite designed to assess and compare agent performance across the delineated knowledge domains and cognitive capabilities. Our code and data is open sourced at https://github.com/Gnonymous/Web-CogReasoner

Bijay Adhikari, Pratibha Kulung, Jakesh Bohaju et al.

Automating brain tumor segmentation using deep learning methods is an ongoing challenge in medical imaging. Multiple lingering issues exist including domain-shift and applications in low-resource settings which brings a unique set of challenges including scarcity of data. As a step towards solving these specific problems, we propose Convolutional adapter-inspired Parameter-efficient Fine-tuning (PEFT) of MedNeXt architecture. To validate our idea, we show our method performs comparable to full fine-tuning with the added benefit of reduced training compute using BraTS-2021 as pre-training dataset and BraTS-Africa as the fine-tuning dataset. BraTS-Africa consists of a small dataset (60 train / 35 validation) from the Sub-Saharan African population with marked shift in the MRI quality compared to BraTS-2021 (1251 train samples). We first show that models trained on BraTS-2021 dataset do not generalize well to BraTS-Africa as shown by 20% reduction in mean dice on BraTS-Africa validation samples. Then, we show that PEFT can leverage both the BraTS-2021 and BraTS-Africa dataset to obtain mean dice of 0.8 compared to 0.72 when trained only on BraTS-Africa. Finally, We show that PEFT (0.80 mean dice) results in comparable performance to full fine-tuning (0.77 mean dice) which may show PEFT to be better on average but the boxplots show that full finetuning results is much lesser variance in performance. Nevertheless, on disaggregation of the dice metrics, we find that the model has tendency to oversegment as shown by high specificity (0.99) compared to relatively low sensitivity(0.75). The source code is available at https://github.com/CAMERA-MRI/SPARK2024/tree/main/PEFT_MedNeXt

Jun Zhan, Mingyang Han, Yuxuan Xie et al.

Spoken language models (SLMs) have emerged as a unified paradigm for speech understanding and generation, enabling natural human machine interaction. However, while most progress has focused on semantic accuracy and instruction following, the ability of SLMs to adapt their speaking style based on spoken instructions has received limited attention. We introduce Voice Style Adaptation (VSA), a new task that examines whether SLMs can modify their speaking style, such as timbre, prosody, or persona following natural language spoken commands. To study this task, we present VStyle, a bilingual (Chinese & English) benchmark covering four categories of speech generation: acoustic attributes, natural language instruction, role play, and implicit empathy. We also introduce the Large Audio Language Model as a Judge (LALM as a Judge) framework, which progressively evaluates outputs along textual faithfulness, style adherence, and naturalness, ensuring reproducible and objective assessment. Experiments on commercial systems and open source SLMs demonstrate that current models face clear limitations in controllable style adaptation, highlighting both the novelty and challenge of this task. By releasing VStyle and its evaluation toolkit, we aim to provide the community with a foundation for advancing human centered spoken interaction. The dataset and code are publicly available at \href{https://junzhan2000.github.io/VStyle.github.io/}{project's homepage}.

Han Zhu, Wei Kang, Liyong Guo et al.

Generating spoken dialogue is more challenging than monologue text-to-speech (TTS) due to the need for realistic turn-taking and distinct speaker timbres. Existing spoken dialogue generation models, being auto-regressive, suffer from slow and unstable inference. To overcome these limitations, we introduce ZipVoice-Dialog, a non-autoregressive zero-shot spoken dialogue generation model built upon flow matching. Key designs include: 1) speaker-turn embeddings for precise speaker turn-taking; 2) a curriculum learning strategy for stable speech-text alignment; 3) specialized strategies to enable stereo dialogue generation. Additionally, recognizing the lack of open-source large-scale spoken dialogue datasets, we curated OpenDialog, a 6.8k-hour spoken dialogue dataset from in-the-wild speech data. Furthermore, we established a benchmark to comprehensively evaluate various models. Experimental results demonstrate that ZipVoice-Dialog achieves superior performance in intelligibility, speaker turn-taking accuracy, speaker similarity, and inference speed. Our codes, model checkpoints, demo samples, and the OpenDialog dataset are all publicly available at https://github.com/k2-fsa/ZipVoice.

Quanwei Tang, Sophia Yat Mei Lee, Junshuang Wu et al.

Recent advancements in retrieval-augmented generation (RAG) have enhanced large language models in question answering by integrating external knowledge. However, challenges persist in achieving global understanding and aligning responses with human ethical and quality preferences. To address these issues, we propose GraphMPA, a comprehensive graph-based framework with mode-seeking preference alignment. Our approach constructs a hierarchical document graph using a general similarity measurement, mimicking human cognitive processes for information understanding and synthesis. Additionally, we introduce mode-seeking preference optimization to better align model outputs with human preferences through probability-matching constraints. Extensive experiments on six datasets demonstrate the effectiveness of our \href{https://github.com/tangquanwei/GraphMPA}{GraphMPA}.

Yi Lin, Dong Zhang, Xiao Fang et al.

Medical image segmentation is a pivotal task within the realms of medical image analysis and computer vision. While current methods have shown promise in accurately segmenting major regions of interest, the precise segmentation of boundary areas remains challenging. In this study, we propose a novel network architecture named CTO, which combines Convolutional Neural Networks (CNNs), Vision Transformer (ViT) models, and explicit edge detection operators to tackle this challenge. CTO surpasses existing methods in terms of segmentation accuracy and strikes a better balance between accuracy and efficiency, without the need for additional data inputs or label injections. Specifically, CTO adheres to the canonical encoder-decoder network paradigm, with a dual-stream encoder network comprising a mainstream CNN stream for capturing local features and an auxiliary StitchViT stream for integrating long-range dependencies. Furthermore, to enhance the model's ability to learn boundary areas, we introduce a boundary-guided decoder network that employs binary boundary masks generated by dedicated edge detection operators to provide explicit guidance during the decoding process. We validate the performance of CTO through extensive experiments conducted on seven challenging medical image segmentation datasets, namely ISIC 2016, PH2, ISIC 2018, CoNIC, LiTS17, and BTCV. Our experimental results unequivocally demonstrate that CTO achieves state-of-the-art accuracy on these datasets while maintaining competitive model complexity. The codes have been released at: https://github.com/xiaofang007/CTO.

Yonghao Tan, Pingcheng Dong, Yongkun Wu et al.

DNN accelerators, significantly advanced by model compression and specialized dataflow techniques, have marked considerable progress. However, the frequent access of high-precision partial sums (PSUMs) leads to excessive memory demands in architectures utilizing input/weight stationary dataflows. Traditional compression strategies have typically overlooked PSUM quantization, which may account for 69% of power consumption. This study introduces a novel Additive Partial Sum Quantization (APSQ) method, seamlessly integrating PSUM accumulation into the quantization framework. A grouping strategy that combines APSQ with PSUM quantization enhanced by a reconfigurable architecture is further proposed. The APSQ performs nearly lossless on NLP and CV tasks across BERT, Segformer, and EfficientViT models while compressing PSUMs to INT8. This leads to a notable reduction in energy costs by 28-87%. Extended experiments on LLaMA2-7B demonstrate the potential of APSQ for large language models. Code is available at https://github.com/Yonghao-Tan/APSQ.

Jun Zhan, Junqi Dai, Jiasheng Ye et al.

We introduce AnyGPT, an any-to-any multimodal language model that utilizes discrete representations for the unified processing of various modalities, including speech, text, images, and music. AnyGPT can be trained stably without any alterations to the current large language model (LLM) architecture or training paradigms. Instead, it relies exclusively on data-level preprocessing, facilitating the seamless integration of new modalities into LLMs, akin to the incorporation of new languages. We build a multimodal text-centric dataset for multimodal alignment pre-training. Utilizing generative models, we synthesize the first large-scale any-to-any multimodal instruction dataset. It consists of 108k samples of multi-turn conversations that intricately interweave various modalities, thus equipping the model to handle arbitrary combinations of multimodal inputs and outputs. Experimental results demonstrate that AnyGPT is capable of facilitating any-to-any multimodal conversation while achieving performance comparable to specialized models across all modalities, proving that discrete representations can effectively and conveniently unify multiple modalities within a language model. Demos are shown in https://junzhan2000.github.io/AnyGPT.github.io/

Mohtady Barakat, Omar Salah, Ahmed Yasser et al.

Gliomas are placing an increasingly clinical burden on Sub-Saharan Africa (SSA). In the region, the median survival for patients remains under two years, and access to diagnostic imaging is extremely limited. These constraints highlight an urgent need for automated tools that can extract the maximum possible information from each available scan, tools that are specifically trained on local data, rather than adapted from high-income settings where conditions are vastly different. We utilize the Brain Tumor Segmentation (BraTS) Africa 2025 Challenge dataset, an expert annotated collection of glioma MRIs. Our objectives are: (i) establish a strong baseline with nnUNet on this dataset, and (ii) explore whether the celebrated "grokking" phenomenon an abrupt, late training jump from memorization to superior generalization can be triggered to push performance without extra labels. We evaluate two training regimes. The first is a fast, budget-conscious approach that limits optimization to just a few epochs, reflecting the constrained GPU resources typically available in African institutions. Despite this limitation, nnUNet achieves strong Dice scores: 92.3% for whole tumor (WH), 86.6% for tumor core (TC), and 86.3% for enhancing tumor (ET). The second regime extends training well beyond the point of convergence, aiming to trigger a grokking-driven performance leap. With this approach, we were able to achieve grokking and enhanced our results to higher Dice scores: 92.2% for whole tumor (WH), 90.1% for tumor core (TC), and 90.2% for enhancing tumor (ET).