Chao Xu

Pan Zhang, Xiaoyi Dong, Bin Wang et al. · pku

We propose InternLM-XComposer, a vision-language large model that enables advanced image-text comprehension and composition. The innovative nature of our model is highlighted by three appealing properties: 1) Interleaved Text-Image Composition: InternLM-XComposer can effortlessly generate coherent and contextual articles that seamlessly integrate images, providing a more engaging and immersive reading experience. Simply provide a writing instruction, and our system will generate the corresponding manuscript. It can intelligently identify the areas in the text where images would enhance the content and automatically insert the most appropriate visual candidates. 2) Comprehension with Rich Multilingual Knowledge: The text-image comprehension is empowered by training on an extensive multi-modal multilingual database with carefully crafted strategies, resulting in a deep understanding of visual content. 3) State-of-the-art Performance: Our model consistently achieves state-of-the-art results across various mainstream benchmarks for vision-language foundational models, including MME Benchmark, MMBench, MMBench-CN, Seed-Bench, CCBench (Chinese Cultural Benchmark), QBench and Tiny LVLM. Owing to the absence of established metrics for quantitatively assessing text-image composition, we have devised a robust evaluation procedure that comprises both human and GPT4-Vision (GPT4-V) to ensure reliability. Notably, our InternLM-XComposer achieves competitive text-image composition scores compared to public solutions, including GPT4-V and GPT3.5. Collectively, InternLM-XComposer seamlessly blends advanced text-image comprehension and composition, revolutionizing vision-language interaction and offering new insights and opportunities. The InternLM-XComposer model series are publicly available at https://github.com/InternLM/InternLM-XComposer.

Ruoxi Shi, Hansheng Chen, Zhuoyang Zhang et al. · stanford

We report Zero123++, an image-conditioned diffusion model for generating 3D-consistent multi-view images from a single input view. To take full advantage of pretrained 2D generative priors, we develop various conditioning and training schemes to minimize the effort of finetuning from off-the-shelf image diffusion models such as Stable Diffusion. Zero123++ excels in producing high-quality, consistent multi-view images from a single image, overcoming common issues like texture degradation and geometric misalignment. Furthermore, we showcase the feasibility of training a ControlNet on Zero123++ for enhanced control over the generation process. The code is available at https://github.com/SUDO-AI-3D/zero123plus.

Jiefeng Li, Siyuan Bian, Chao Xu et al. · tencent-ai

3D human pose estimation from a monocular video has recently seen significant improvements. However, most state-of-the-art methods are kinematics-based, which are prone to physically implausible motions with pronounced artifacts. Current dynamics-based methods can predict physically plausible motion but are restricted to simple scenarios with static camera view. In this work, we present D&D (Learning Human Dynamics from Dynamic Camera), which leverages the laws of physics to reconstruct 3D human motion from the in-the-wild videos with a moving camera. D&D introduces inertial force control (IFC) to explain the 3D human motion in the non-inertial local frame by considering the inertial forces of the dynamic camera. To learn the ground contact with limited annotations, we develop probabilistic contact torque (PCT), which is computed by differentiable sampling from contact probabilities and used to generate motions. The contact state can be weakly supervised by encouraging the model to generate correct motions. Furthermore, we propose an attentive PD controller that adjusts target pose states using temporal information to obtain smooth and accurate pose control. Our approach is entirely neural-based and runs without offline optimization or simulation in physics engines. Experiments on large-scale 3D human motion benchmarks demonstrate the effectiveness of D&D, where we exhibit superior performance against both state-of-the-art kinematics-based and dynamics-based methods. Code is available at https://github.com/Jeffsjtu/DnD

Yehui Tang, Kai Han, Jianyuan Guo et al.

Light-weight convolutional neural networks (CNNs) are specially designed for applications on mobile devices with faster inference speed. The convolutional operation can only capture local information in a window region, which prevents performance from being further improved. Introducing self-attention into convolution can capture global information well, but it will largely encumber the actual speed. In this paper, we propose a hardware-friendly attention mechanism (dubbed DFC attention) and then present a new GhostNetV2 architecture for mobile applications. The proposed DFC attention is constructed based on fully-connected layers, which can not only execute fast on common hardware but also capture the dependence between long-range pixels. We further revisit the expressiveness bottleneck in previous GhostNet and propose to enhance expanded features produced by cheap operations with DFC attention, so that a GhostNetV2 block can aggregate local and long-range information simultaneously. Extensive experiments demonstrate the superiority of GhostNetV2 over existing architectures. For example, it achieves 75.3% top-1 accuracy on ImageNet with 167M FLOPs, significantly suppressing GhostNetV1 (74.5%) with a similar computational cost. The source code will be available at https://github.com/huawei-noah/Efficient-AI-Backbones/tree/master/ghostnetv2_pytorch and https://gitee.com/mindspore/models/tree/master/research/cv/ghostnetv2.

Hao-Shu Fang, Jiefeng Li, Hongyang Tang et al.

Accurate whole-body multi-person pose estimation and tracking is an important yet challenging topic in computer vision. To capture the subtle actions of humans for complex behavior analysis, whole-body pose estimation including the face, body, hand and foot is essential over conventional body-only pose estimation. In this paper, we present AlphaPose, a system that can perform accurate whole-body pose estimation and tracking jointly while running in realtime. To this end, we propose several new techniques: Symmetric Integral Keypoint Regression (SIKR) for fast and fine localization, Parametric Pose Non-Maximum-Suppression (P-NMS) for eliminating redundant human detections and Pose Aware Identity Embedding for jointly pose estimation and tracking. During training, we resort to Part-Guided Proposal Generator (PGPG) and multi-domain knowledge distillation to further improve the accuracy. Our method is able to localize whole-body keypoints accurately and tracks humans simultaneously given inaccurate bounding boxes and redundant detections. We show a significant improvement over current state-of-the-art methods in both speed and accuracy on COCO-wholebody, COCO, PoseTrack, and our proposed Halpe-FullBody pose estimation dataset. Our model, source codes and dataset are made publicly available at https://github.com/MVIG-SJTU/AlphaPose.

Haoran Geng, Helin Xu, Chengyang Zhao et al. · berkeley

For years, researchers have been devoted to generalizable object perception and manipulation, where cross-category generalizability is highly desired yet underexplored. In this work, we propose to learn such cross-category skills via Generalizable and Actionable Parts (GAParts). By identifying and defining 9 GAPart classes (lids, handles, etc.) in 27 object categories, we construct a large-scale part-centric interactive dataset, GAPartNet, where we provide rich, part-level annotations (semantics, poses) for 8,489 part instances on 1,166 objects. Based on GAPartNet, we investigate three cross-category tasks: part segmentation, part pose estimation, and part-based object manipulation. Given the significant domain gaps between seen and unseen object categories, we propose a robust 3D segmentation method from the perspective of domain generalization by integrating adversarial learning techniques. Our method outperforms all existing methods by a large margin, no matter on seen or unseen categories. Furthermore, with part segmentation and pose estimation results, we leverage the GAPart pose definition to design part-based manipulation heuristics that can generalize well to unseen object categories in both the simulator and the real world. Our dataset, code, and demos are available on our project page.

Bin Wang, Chao Xu, Xiaomeng Zhao et al.

Document content analysis has been a crucial research area in computer vision. Despite significant advancements in methods such as OCR, layout detection, and formula recognition, existing open-source solutions struggle to consistently deliver high-quality content extraction due to the diversity in document types and content. To address these challenges, we present MinerU, an open-source solution for high-precision document content extraction. MinerU leverages the sophisticated PDF-Extract-Kit models to extract content from diverse documents effectively and employs finely-tuned preprocessing and postprocessing rules to ensure the accuracy of the final results. Experimental results demonstrate that MinerU consistently achieves high performance across various document types, significantly enhancing the quality and consistency of content extraction. The MinerU open-source project is available at https://github.com/opendatalab/MinerU.

Minghua Liu, Ruoxi Shi, Linghao Chen et al. · stanford

Recent advancements in open-world 3D object generation have been remarkable, with image-to-3D methods offering superior fine-grained control over their text-to-3D counterparts. However, most existing models fall short in simultaneously providing rapid generation speeds and high fidelity to input images - two features essential for practical applications. In this paper, we present One-2-3-45++, an innovative method that transforms a single image into a detailed 3D textured mesh in approximately one minute. Our approach aims to fully harness the extensive knowledge embedded in 2D diffusion models and priors from valuable yet limited 3D data. This is achieved by initially finetuning a 2D diffusion model for consistent multi-view image generation, followed by elevating these images to 3D with the aid of multi-view conditioned 3D native diffusion models. Extensive experimental evaluations demonstrate that our method can produce high-quality, diverse 3D assets that closely mirror the original input image. Our project webpage: https://sudo-ai-3d.github.io/One2345plus_page.

Conghui He, Zhenjiang Jin, Chao Xu et al.

The rise in popularity of ChatGPT and GPT-4 has significantly accelerated the development of large models, leading to the creation of numerous impressive large language models(LLMs) and multimodal large language models (MLLMs). These cutting-edge models owe their remarkable performance to high-quality data. However, the details of the training data used in leading paradigms are often kept confidential. This lack of transparency, coupled with the scarcity of open-source data, impedes further developments within the community. As a response, this paper presents "Wan Juan", a large-scale multimodal dataset composed of both Chinese and English data, collected from a wide range of web sources. The dataset incorporates text, image-text, and video modalities, with a total volume exceeding 2TB. It was utilized in the training of InternLM, a model that demonstrated significant advantages in multi-dimensional evaluations when compared to models of a similar scale. All data can be accessed at https://opendatalab.org.cn/WanJuan1.0.

Ning Ding, Yixing Xu, Yehui Tang et al.

Domain Adaptation aims to transfer the knowledge learned from a labeled source domain to an unlabeled target domain whose data distributions are different. However, the training data in source domain required by most of the existing methods is usually unavailable in real-world applications due to privacy preserving policies. Recently, Source-Free Domain Adaptation (SFDA) has drawn much attention, which tries to tackle domain adaptation problem without using source data. In this work, we propose a novel framework called SFDA-DE to address SFDA task via source Distribution Estimation. Firstly, we produce robust pseudo-labels for target data with spherical k-means clustering, whose initial class centers are the weight vectors (anchors) learned by the classifier of pretrained model. Furthermore, we propose to estimate the class-conditioned feature distribution of source domain by exploiting target data and corresponding anchors. Finally, we sample surrogate features from the estimated distribution, which are then utilized to align two domains by minimizing a contrastive adaptation loss function. Extensive experiments show that the proposed method achieves state-of-the-art performance on multiple DA benchmarks, and even outperforms traditional DA methods which require plenty of source data.

Minghua Liu, Chong Zeng, Xinyue Wei et al. · stanford

Open-world 3D reconstruction models have recently garnered significant attention. However, without sufficient 3D inductive bias, existing methods typically entail expensive training costs and struggle to extract high-quality 3D meshes. In this work, we introduce MeshFormer, a sparse-view reconstruction model that explicitly leverages 3D native structure, input guidance, and training supervision. Specifically, instead of using a triplane representation, we store features in 3D sparse voxels and combine transformers with 3D convolutions to leverage an explicit 3D structure and projective bias. In addition to sparse-view RGB input, we require the network to take input and generate corresponding normal maps. The input normal maps can be predicted by 2D diffusion models, significantly aiding in the guidance and refinement of the geometry's learning. Moreover, by combining Signed Distance Function (SDF) supervision with surface rendering, we directly learn to generate high-quality meshes without the need for complex multi-stage training processes. By incorporating these explicit 3D biases, MeshFormer can be trained efficiently and deliver high-quality textured meshes with fine-grained geometric details. It can also be integrated with 2D diffusion models to enable fast single-image-to-3D and text-to-3D tasks. Project page: https://meshformer3d.github.io

Yicheng Zou, Dongsheng Zhu, Lin Zhu et al.

We introduce Intern-S1-Pro, the first one-trillion-parameter scientific multimodal foundation model. Scaling to this unprecedented size, the model delivers a comprehensive enhancement across both general and scientific domains. Beyond stronger reasoning and image-text understanding capabilities, its intelligence is augmented with advanced agent capabilities. Simultaneously, its scientific expertise has been vastly expanded to master over 100 specialized tasks across critical science fields, including chemistry, materials, life sciences, and earth sciences. Achieving this massive scale is made possible by the robust infrastructure support of XTuner and LMDeploy, which facilitates highly efficient Reinforcement Learning (RL) training at the 1-trillion parameter level while ensuring strict precision consistency between training and inference. By seamlessly integrating these advancements, Intern-S1-Pro further fortifies the fusion of general and specialized intelligence, working as a Specializable Generalist, demonstrating its position in the top tier of open-source models for general capabilities, while outperforming proprietary models in the depth of specialized scientific tasks.

Minghua Liu, Chao Xu, Haian Jin et al.

Single image 3D reconstruction is an important but challenging task that requires extensive knowledge of our natural world. Many existing methods solve this problem by optimizing a neural radiance field under the guidance of 2D diffusion models but suffer from lengthy optimization time, 3D inconsistency results, and poor geometry. In this work, we propose a novel method that takes a single image of any object as input and generates a full 360-degree 3D textured mesh in a single feed-forward pass. Given a single image, we first use a view-conditioned 2D diffusion model, Zero123, to generate multi-view images for the input view, and then aim to lift them up to 3D space. Since traditional reconstruction methods struggle with inconsistent multi-view predictions, we build our 3D reconstruction module upon an SDF-based generalizable neural surface reconstruction method and propose several critical training strategies to enable the reconstruction of 360-degree meshes. Without costly optimizations, our method reconstructs 3D shapes in significantly less time than existing methods. Moreover, our method favors better geometry, generates more 3D consistent results, and adheres more closely to the input image. We evaluate our approach on both synthetic data and in-the-wild images and demonstrate its superiority in terms of both mesh quality and runtime. In addition, our approach can seamlessly support the text-to-3D task by integrating with off-the-shelf text-to-image diffusion models.

Yuchuan Tian, Hanting Chen, Tianyu Guo et al.

Convolutional Neural Networks (CNNs) are hard to deploy on edge devices due to its high computation and storage complexities. As a common practice for model compression, network pruning consists of two major categories: unstructured and structured pruning, where unstructured pruning constantly performs better. However, unstructured pruning presents a structured pattern at high pruning rates, which limits its performance. To this end, we propose a Rank-based PruninG (RPG) method to maintain the ranks of sparse weights in an adversarial manner. In each step, we minimize the low-rank approximation error for the weight matrices using singular value decomposition, and maximize their distance by pushing the weight matrices away from its low rank approximation. This rank-based optimization objective guides sparse weights towards a high-rank topology. The proposed method is conducted in a gradual pruning fashion to stabilize the change of rank during training. Experimental results on various datasets and different tasks demonstrate the effectiveness of our algorithm in high sparsity. The proposed RPG outperforms the state-of-the-art performance by 1.13% top-1 accuracy on ImageNet in ResNet-50 with 98% sparsity. The codes are available at https://github.com/huawei-noah/Efficient-Computing/tree/master/Pruning/RPG and https://gitee.com/mindspore/models/tree/master/research/cv/RPG.

Yang Liu, Cheng Yu, Lei Shang et al.

Recent advancement in personalized image generation have unveiled the intriguing capability of pre-trained text-to-image models on learning identity information from a collection of portrait images. However, existing solutions are vulnerable in producing truthful details, and usually suffer from several defects such as (i) The generated face exhibit its own unique characteristics, \ie facial shape and facial feature positioning may not resemble key characteristics of the input, and (ii) The synthesized face may contain warped, blurred or corrupted regions. In this paper, we present FaceChain, a personalized portrait generation framework that combines a series of customized image-generation model and a rich set of face-related perceptual understanding models (\eg, face detection, deep face embedding extraction, and facial attribute recognition), to tackle aforementioned challenges and to generate truthful personalized portraits, with only a handful of portrait images as input. Concretely, we inject several SOTA face models into the generation procedure, achieving a more efficient label-tagging, data-processing, and model post-processing compared to previous solutions, such as DreamBooth ~\cite{ruiz2023dreambooth} , InstantBooth ~\cite{shi2023instantbooth} , or other LoRA-only approaches ~\cite{hu2021lora} . Besides, based on FaceChain, we further develop several applications to build a broader playground for better showing its value, including virtual try on and 2D talking head. We hope it can grow to serve the burgeoning needs from the communities. Note that this is an ongoing work that will be consistently refined and improved upon. FaceChain is open-sourced under Apache-2.0 license at \url{https://github.com/modelscope/facechain}.

Xuhai Chen, Jiangning Zhang, Chao Xu et al.

Most of the existing blind image Super-Resolution (SR) methods assume that the blur kernels are space-invariant. However, the blur involved in real applications are usually space-variant due to object motion, out-of-focus, etc., resulting in severe performance drop of the advanced SR methods. To address this problem, we firstly introduce two new datasets with out-of-focus blur, i.e., NYUv2-BSR and Cityscapes-BSR, to support further researches of blind SR with space-variant blur. Based on the datasets, we design a novel Cross-MOdal fuSion network (CMOS) that estimate both blur and semantics simultaneously, which leads to improved SR results. It involves a feature Grouping Interactive Attention (GIA) module to make the two modalities interact more effectively and avoid inconsistency. GIA can also be used for the interaction of other features because of the universality of its structure. Qualitative and quantitative experiments compared with state-of-the-art methods on above datasets and real-world images demonstrate the superiority of our method, e.g., obtaining PSNR/SSIM by +1.91/+0.0048 on NYUv2-BSR than MANet.

Yuchuan Tian, Yuchen Liang, Jiacheng Sun et al.

Large language models (LLMs) excel at generation but dominant autoregressive (AR) decoding is inherently sequential, creating a throughput bottleneck. Diffusion Language Models (DLMs)--especially block-wise variants--enable parallel generation and intra-block bidirectional reasoning, yet training large DLMs from scratch is costly and wastes the knowledge in mature AR checkpoints. Prior "adaptation" attempts either modify logits or randomly grow attention masks to full-sequence diffusion, or simply transplant AR weights into a block-diffusion recipe, leaving a fundamental mismatch between AR causality and block-wise bidirectionality unaddressed. We reframe adaptation as a intra-paradigm path from AR to Block-Diffusion by viewing AR as Block-Diffusion with blocksize=1. Concretely, we design the pathway of adaptation as follows: we use a context-causal attention mask (causal in context, bidirectional only within the active block), an efficient parallel adaptation procedure, an auxiliary AR loss to maximize data utilization and retain pretrained knowledge, and gradual increment of the generation block size. The recipe integrates cleanly with masked block-diffusion and maintains train-inference consistency. Built on these components, NBDiff-7B (Base and Instruct) could inherit the long-context modeling and reasoning capabilities, and achieve state-of-the-art performance among the 7B-class DLMs, delivering strong gains on general-knowledge, math, and code benchmarks over strong baselines. These results demonstrate that principled AR-to-block-diffusion adaptation is an effective and compute-efficient alternative to training DLMs from scratch. Codes: https://github.com/YuchuanTian/NBDiff.

Chao Xu, Jiangning Zhang, Miao Hua et al.

This paper presents a novel Region-Aware Face Swapping (RAFSwap) network to achieve identity-consistent harmonious high-resolution face generation in a local-global manner: \textbf{1)} Local Facial Region-Aware (FRA) branch augments local identity-relevant features by introducing the Transformer to effectively model misaligned cross-scale semantic interaction. \textbf{2)} Global Source Feature-Adaptive (SFA) branch further complements global identity-relevant cues for generating identity-consistent swapped faces. Besides, we propose a \textit{Face Mask Predictor} (FMP) module incorporated with StyleGAN2 to predict identity-relevant soft facial masks in an unsupervised manner that is more practical for generating harmonious high-resolution faces. Abundant experiments qualitatively and quantitatively demonstrate the superiority of our method for generating more identity-consistent high-resolution swapped faces over SOTA methods, \eg, obtaining 96.70 ID retrieval that outperforms SOTA MegaFS by 5.87$\uparrow$.

Yushuo Chen, Zerong Zheng, Zhe Li et al.

We present a novel pipeline for learning high-quality triangular human avatars from multi-view videos. Recent methods for avatar learning are typically based on neural radiance fields (NeRF), which is not compatible with traditional graphics pipeline and poses great challenges for operations like editing or synthesizing under different environments. To overcome these limitations, our method represents the avatar with an explicit triangular mesh extracted from an implicit SDF field, complemented by an implicit material field conditioned on given poses. Leveraging this triangular avatar representation, we incorporate physics-based rendering to accurately decompose geometry and texture. To enhance both the geometric and appearance details, we further employ a 2D UNet as the network backbone and introduce pseudo normal ground-truth as additional supervision. Experiments show that our method can learn triangular avatars with high-quality geometry reconstruction and plausible material decomposition, inherently supporting editing, manipulation or relighting operations.

Chao Zhang, Kunlun Li, Chao Xu et al.

Perfect Doppler compensation and synchronization is nontrivial due to multi-path Doppler effects and Einstein's theory of relativity in the space-air-ground-integrated networks (SAGINs). Hence, by considering the residual Doppler and the synchronization delay, this paper investigates the bit-error-rate (BER) performance attained under time-varying correlated Shadowed-Rician SAGIN channels. First, a practical SAGIN model is harnessed, encompassing correlated Shadowed-Rician channels, the Snell's law-based path loss, atmospheric absorption, the line-of-sight Doppler compensation, elliptical satellite orbits, and Einstein's theory of relativity. Then, a specific correlation coefficient between the pilot and data symbols is derived in the context of correlated Shadowed-Rician channels. By exploiting this correlation coefficient, the channel distribution is mimicked by a bi-variate Gamma distribution. Then, a closed-form BER formula is derived under employing least-square channel estimation and equalization for 16-QAM. Our analytical results indicate for a 300-km-altitude LEO that 1) the period of realistic elliptical orbits is around 0.8 seconds longer than that of the idealized circular orbits; and 2) the relativistic delay is lower than 1 microsecond over a full LEO pass (from rise to set). Our numerical results for the L bands quantify the effects of: 1) the residual Doppler; 2) atmospheric shadowing; 3) synchronization errors; and 4) pilot overhead.

Peng Yu, Chao Xu, Albert Bifet et al.

Decision trees are well-known due to their ease of interpretability. To improve accuracy, we need to grow deep trees or ensembles of trees. These are hard to interpret, offsetting their original benefits. Shapley values have recently become a popular way to explain the predictions of tree-based machine learning models. It provides a linear weighting to features independent of the tree structure. The rise in popularity is mainly due to TreeShap, which solves a general exponential complexity problem in polynomial time. Following extensive adoption in the industry, more efficient algorithms are required. This paper presents a more efficient and straightforward algorithm: Linear TreeShap. Like TreeShap, Linear TreeShap is exact and requires the same amount of memory.

Hao Wang, Licheng Pan, Qingsong Wen et al.

Autocorrelation is a defining characteristic of time-series data, where each observation is statistically dependent on its predecessors. In the context of deep time-series forecasting, autocorrelation arises in both the input history and the label sequences, presenting two central research challenges: (1) designing neural architectures that model autocorrelation in history sequences, and (2) devising learning objectives that model autocorrelation in label sequences. Recent studies have made strides in tackling these challenges, but a systematic survey examining both aspects remains lacking. To bridge this gap, this paper provides a comprehensive review of deep time-series forecasting from the perspective of autocorrelation modeling. In contrast to existing surveys, this work makes two distinctive contributions. First, it proposes a novel taxonomy that encompasses recent literature on both model architectures and learning objectives -- whereas prior surveys neglect or inadequately discuss the latter aspect. Second, it offers a thorough analysis of the motivations, insights, and progression of the surveyed literature from a unified, autocorrelation-centric perspective, providing a holistic overview of the evolution of deep time-series forecasting. The full list of papers and resources is available at https://github.com/Master-PLC/Awesome-TSF-Papers.

Shuying Gan, Marie Siew, Chao Xu et al.

Mobile edge computing (MEC) is a promising paradigm to meet the quality of service (QoS) requirements of latency-sensitive IoT applications. However, attackers may eavesdrop on the offloading decisions to infer the edge server's (ES's) queue information and users' usage patterns, thereby incurring the pattern privacy (PP) issue. Therefore, we propose an offloading strategy which jointly minimizes the latency, ES's energy consumption, and task dropping rate, while preserving PP. Firstly, we formulate the dynamic computation offloading procedure as a Markov decision process (MDP). Next, we develop a Differential Privacy Deep Q-learning based Offloading (DP-DQO) algorithm to solve this problem while addressing the PP issue by injecting noise into the generated offloading decisions. This is achieved by modifying the deep Q-network (DQN) with a Function-output Gaussian process mechanism. We provide a theoretical privacy guarantee and a utility guarantee (learning error bound) for the DP-DQO algorithm and finally, conduct simulations to evaluate the performance of our proposed algorithm by comparing it with greedy and DQN-based algorithms.

Zhe Chen, Weiyun Wang, Hao Tian et al.

In this report, we introduce InternVL 1.5, an open-source multimodal large language model (MLLM) to bridge the capability gap between open-source and proprietary commercial models in multimodal understanding. We introduce three simple improvements: (1) Strong Vision Encoder: we explored a continuous learning strategy for the large-scale vision foundation model -- InternViT-6B, boosting its visual understanding capabilities, and making it can be transferred and reused in different LLMs. (2) Dynamic High-Resolution: we divide images into tiles ranging from 1 to 40 of 448$\times$448 pixels according to the aspect ratio and resolution of the input images, which supports up to 4K resolution input. (3) High-Quality Bilingual Dataset: we carefully collected a high-quality bilingual dataset that covers common scenes, document images, and annotated them with English and Chinese question-answer pairs, significantly enhancing performance in OCR- and Chinese-related tasks. We evaluate InternVL 1.5 through a series of benchmarks and comparative studies. Compared to both open-source and proprietary models, InternVL 1.5 shows competitive performance, achieving state-of-the-art results in 8 of 18 benchmarks. Code has been released at https://github.com/OpenGVLab/InternVL.

Jiefeng Li, Siyuan Bian, Chao Xu et al.

Recovering whole-body mesh by inferring the abstract pose and shape parameters from visual content can obtain 3D bodies with realistic structures. However, the inferring process is highly non-linear and suffers from image-mesh misalignment, resulting in inaccurate reconstruction. In contrast, 3D keypoint estimation methods utilize the volumetric representation to achieve pixel-level accuracy but may predict unrealistic body structures. To address these issues, this paper presents a novel hybrid inverse kinematics solution, HybrIK, that integrates the merits of 3D keypoint estimation and body mesh recovery in a unified framework. HybrIK directly transforms accurate 3D joints to body-part rotations via twist-and-swing decomposition. The swing rotations are analytically solved with 3D joints, while the twist rotations are derived from visual cues through neural networks. To capture comprehensive whole-body details, we further develop a holistic framework, HybrIK-X, which enhances HybrIK with articulated hands and an expressive face. HybrIK-X is fast and accurate by solving the whole-body pose with a one-stage model. Experiments demonstrate that HybrIK and HybrIK-X preserve both the accuracy of 3D joints and the realistic structure of the parametric human model, leading to pixel-aligned whole-body mesh recovery. The proposed method significantly surpasses the state-of-the-art methods on various benchmarks for body-only, hand-only, and whole-body scenarios. Code and results can be found at https://jeffli.site/HybrIK-X/

Yuxuan Zhang, Chao Xu, Howard H. Yang et al.

This paper proposes a client selection (CS) method to tackle the communication bottleneck of federated learning (FL) while concurrently coping with FL's data heterogeneity issue. Specifically, we first analyze the effect of CS in FL and show that FL training can be accelerated by adequately choosing participants to diversify the training dataset in each round of training. Based on this, we leverage data profiling and determinantal point process (DPP) sampling techniques to develop an algorithm termed Federated Learning with DPP-based Participant Selection (FL-DP$^3$S). This algorithm effectively diversifies the participants' datasets in each round of training while preserving their data privacy. We conduct extensive experiments to examine the efficacy of our proposed method. The results show that our scheme attains a faster convergence rate, as well as a smaller communication overhead than several baselines.

Haojie Liu, Daoxun Xia, Wei Jiang et al.

Visible-infrared person re-identification (VI-ReID) is a challenging and essential task, which aims to retrieve a set of person images over visible and infrared camera views. In order to mitigate the impact of large modality discrepancy existing in heterogeneous images, previous methods attempt to apply generative adversarial network (GAN) to generate the modality-consisitent data. However, due to severe color variations between the visible domain and infrared domain, the generated fake cross-modality samples often fail to possess good qualities to fill the modality gap between synthesized scenarios and target real ones, which leads to sub-optimal feature representations. In this work, we address cross-modality matching problem with Aligned Grayscale Modality (AGM), an unified dark-line spectrum that reformulates visible-infrared dual-mode learning as a gray-gray single-mode learning problem. Specifically, we generate the grasycale modality from the homogeneous visible images. Then, we train a style tranfer model to transfer infrared images into homogeneous grayscale images. In this way, the modality discrepancy is significantly reduced in the image space. In order to reduce the remaining appearance discrepancy, we further introduce a multi-granularity feature extraction network to conduct feature-level alignment. Rather than relying on the global information, we propose to exploit local (head-shoulder) features to assist person Re-ID, which complements each other to form a stronger feature descriptor. Comprehensive experiments implemented on the mainstream evaluation datasets include SYSU-MM01 and RegDB indicate that our method can significantly boost cross-modality retrieval performance against the state of the art methods.

Chao Xu, Ang Li, Linghao Chen et al.

Open-world 3D generation has recently attracted considerable attention. While many single-image-to-3D methods have yielded visually appealing outcomes, they often lack sufficient controllability and tend to produce hallucinated regions that may not align with users' expectations. In this paper, we explore an important scenario in which the input consists of one or a few unposed 2D images of a single object, with little or no overlap. We propose a novel method, SpaRP, to reconstruct a 3D textured mesh and estimate the relative camera poses for these sparse-view images. SpaRP distills knowledge from 2D diffusion models and finetunes them to implicitly deduce the 3D spatial relationships between the sparse views. The diffusion model is trained to jointly predict surrogate representations for camera poses and multi-view images of the object under known poses, integrating all information from the input sparse views. These predictions are then leveraged to accomplish 3D reconstruction and pose estimation, and the reconstructed 3D model can be used to further refine the camera poses of input views. Through extensive experiments on three datasets, we demonstrate that our method not only significantly outperforms baseline methods in terms of 3D reconstruction quality and pose prediction accuracy but also exhibits strong efficiency. It requires only about 20 seconds to produce a textured mesh and camera poses for the input views. Project page: https://chaoxu.xyz/sparp.

Jiazheng Xing, Chao Xu, Mengmeng Wang et al.

Applying large-scale vision-language pre-trained models like CLIP to few-shot action recognition (FSAR) can significantly enhance both performance and efficiency. While several studies have recognized this advantage, most of them resort to full-parameter fine-tuning to make CLIP's visual encoder adapt to the FSAR data, which not only costs high computations but also overlooks the potential of the visual encoder to engage in temporal modeling and focus on targeted semantics directly. To tackle these issues, we introduce MA-FSAR, a framework that employs the Parameter-Efficient Fine-Tuning (PEFT) technique to enhance the CLIP visual encoder in terms of action-related temporal and semantic representations. Our solution involves a Fine-grained Multimodal Adaptation, which is different from the previous attempts of PEFT in regular action recognition. Specifically, we first insert a Global Temporal Adaptation that only receives the class token to capture global motion cues efficiently. Then these outputs integrate with visual tokens to enhance local temporal dynamics by a Local Multimodal Adaptation, which incorporates text features unique to the FSAR support set branch to highlight fine-grained semantics related to actions. In addition to these token-level designs, we propose a prototype-level text-guided construction module to further enrich the temporal and semantic characteristics of video prototypes. Extensive experiments demonstrate our superior performance in various tasks using minor trainable parameters.

Zheng Cai, Maosong Cao, Haojiong Chen et al. · pku

The evolution of Large Language Models (LLMs) like ChatGPT and GPT-4 has sparked discussions on the advent of Artificial General Intelligence (AGI). However, replicating such advancements in open-source models has been challenging. This paper introduces InternLM2, an open-source LLM that outperforms its predecessors in comprehensive evaluations across 6 dimensions and 30 benchmarks, long-context modeling, and open-ended subjective evaluations through innovative pre-training and optimization techniques. The pre-training process of InternLM2 is meticulously detailed, highlighting the preparation of diverse data types including text, code, and long-context data. InternLM2 efficiently captures long-term dependencies, initially trained on 4k tokens before advancing to 32k tokens in pre-training and fine-tuning stages, exhibiting remarkable performance on the 200k ``Needle-in-a-Haystack" test. InternLM2 is further aligned using Supervised Fine-Tuning (SFT) and a novel Conditional Online Reinforcement Learning from Human Feedback (COOL RLHF) strategy that addresses conflicting human preferences and reward hacking. By releasing InternLM2 models in different training stages and model sizes, we provide the community with insights into the model's evolution.

Ning Ding, Yehui Tang, Zhongqian Fu et al.

The upsurge in pre-trained large models started by ChatGPT has swept across the entire deep learning community. Such powerful models demonstrate advanced generative ability and multimodal understanding capability, which quickly set new state of the arts on a variety of benchmarks. The pre-trained LLM usually plays the role as a universal AI model that can conduct various tasks like article analysis and image comprehension. However, due to the prohibitively high memory and computational cost of implementing such a large model, the conventional models (such as CNN and ViT) are still essential for many visual perception tasks. In this paper, we propose to enhance the representation ability of ordinary vision models on perception tasks (e.g. image classification) by taking advantage of the off-the-shelf large pre-trained models. We present a new learning framework, dubbed GPT4Image, where the knowledge of the large pre-trained models are extracted to help CNNs and ViTs learn better representations and achieve higher performance. Firstly, we curate a high quality description set by prompting a multimodal LLM to generate descriptions for training images. Then, these detailed descriptions are fed into a pre-trained encoder to extract text embeddings that encodes the rich semantics of images. During training, text embeddings will serve as extra supervising signal and be aligned with image representations learned by vision models. The alignment process helps vision models achieve better performance with the aid of pre-trained LLMs. We conduct extensive experiments to verify the effectiveness of the proposed algorithm on various visual perception tasks for heterogeneous model architectures.

Peng Yu, Yike Chen, Chao Xu et al.

In the context of the Classification and Regression Trees (CART) algorithm, the efficient splitting of categorical features using standard criteria like GINI and Entropy is well-established. However, using the Mean Absolute Error (MAE) criterion for categorical features has traditionally relied on various numerical encoding methods. This paper demonstrates that unsupervised numerical encoding methods are not viable for the MAE criteria. Furthermore, we present a novel and efficient splitting algorithm that addresses the challenges of handling categorical features with the MAE criterion. Our findings underscore the limitations of existing approaches and offer a promising solution to enhance the handling of categorical data in CART algorithms.

Bo Zou, Chao Xu

High-quality expert chain-of-thought (CoT) data is one of the core bottlenecks in large language model (LLM) post-training. Existing data production methods each have structural limitations: crowdsourced annotation lacks deep reasoning paths; expert solo writing is constrained by the "expert blind spot" -- experts structurally skip reasoning steps they consider obvious; RLHF only produces preference signals rather than reasoning chains. This paper proposes the BC Protocol -- a structured dual-expert elicitation method for LLM post-training data production. The method carefully pairs a domain expert (crystallized intelligence) with a knowledge engineer (fluid intelligence), systematically externalizing the expert's implicit judgments as natural language reasoning chains. We introduce the Participant Aptitude Model, which defines six participant characteristic dimensions that affect elicitation quality. "Calibrated Ignorance" is an original concept proposed in this paper. We further propose "Selection-over-Prescription" as a methodological principle: for implicit knowledge elicitation tasks, investing quality-control resources in personnel selection yields a higher return than investing the same resources in process design. In a controlled experiment in the narrative fiction domain, we directly compared CoT produced by BC Protocol dual dialogue (Group A, (n=20)) against CoT written independently by the same domain expert (Group B, (n=20)). Three cross-vendor judge models -- GPT-4o, Claude Opus 4.5, and Gemini 2.5 Pro -- conducted blind evaluation across five dimensions (600 ratings total). Results show that the BC Protocol achieves an overwhelming advantage in "naturalness of reasoning process" (Group A mean 4.80 vs. Group B mean 1.30, (p=2.4\times10^{-8}), Cliff's (δ=1.0)).

Jiaxu Liu, Song Chen, Shengze Cai et al.

In this paper, we investigate a distributed aggregative optimization problem in a network, where each agent has its own local cost function which depends not only on the local state variable but also on an aggregated function of state variables from all agents. To accelerate the optimization process, we combine heavy ball and Nesterov's accelerated methods with distributed aggregative gradient tracking, and propose two novel algorithms named DAGT-HB and DAGT-NES for solving the distributed aggregative optimization problem. We analyse that the DAGT-HB and DAGT-NES algorithms can converge to an optimal solution at a global $\mathbf{R}-$linear convergence rate when the objective function is smooth and strongly convex, and when the parameters (e.g., step size and momentum coefficients) are selected within certain ranges. A numerical experiment on the optimal placement problem is given to verify the effectiveness and superiority of our proposed algorithms.

Qing Miao, Xiaohe Wu, Chao Xu et al.

Learning with noisy labels (LNL) has been extensively studied, with existing approaches typically following a framework that alternates between clean sample selection and semi-supervised learning (SSL). However, this approach has a limitation: the clean set selected by the Deep Neural Network (DNN) classifier, trained through self-training, inevitably contains noisy samples. This mixture of clean and noisy samples leads to misguidance in DNN training during SSL, resulting in impaired generalization performance due to confirmation bias caused by error accumulation in sample selection. To address this issue, we propose a method called Collaborative Sample Selection (CSS), which leverages the large-scale pre-trained model CLIP. CSS aims to remove the mixed noisy samples from the identified clean set. We achieve this by training a 2-Dimensional Gaussian Mixture Model (2D-GMM) that combines the probabilities from CLIP with the predictions from the DNN classifier. To further enhance the adaptation of CLIP to LNL, we introduce a co-training mechanism with a contrastive loss in semi-supervised learning. This allows us to jointly train the prompt of CLIP and the DNN classifier, resulting in improved feature representation, boosted classification performance of DNNs, and reciprocal benefits to our Collaborative Sample Selection. By incorporating auxiliary information from CLIP and utilizing prompt fine-tuning, we effectively eliminate noisy samples from the clean set and mitigate confirmation bias during training. Experimental results on multiple benchmark datasets demonstrate the effectiveness of our proposed method in comparison with the state-of-the-art approaches.

Song Chen, Shengze Cai, Tehuan Chen et al.

In this paper, we propose a novel nonlinear observer based on neural networks, called neural observer, for observation tasks of linear time-invariant (LTI) systems and uncertain nonlinear systems. In particular, the neural observer designed for uncertain systems is inspired by the active disturbance rejection control, which can measure the uncertainty in real-time. The stability analysis (e.g., exponential convergence rate) of LTI and uncertain nonlinear systems (involving neural observers) are presented and guaranteed, where it is shown that the observation problems can be solved only using the linear matrix inequalities (LMIs). Also, it is revealed that the observability and controllability of the system matrices are required to demonstrate the existence of solutions of LMIs. Finally, the effectiveness of neural observers is verified on three simulation cases, including the X-29A aircraft model, the nonlinear pendulum, and the four-wheel steering vehicle.

Yunhe Wang, Kai Han, Huiling Zhen et al.

The paradigm of Large Language Models (LLMs) is currently defined by auto-regressive (AR) architectures, which generate text through a sequential ``brick-by-brick'' process. Despite their success, AR models are inherently constrained by a causal bottleneck that limits global structural foresight and iterative refinement. Diffusion Language Models (DLMs) offer a transformative alternative, conceptualizing text generation as a holistic, bidirectional denoising process akin to a sculptor refining a masterpiece. However, the potential of DLMs remains largely untapped as they are frequently confined within AR-legacy infrastructures and optimization frameworks. In this Perspective, we identify ten fundamental challenges ranging from architectural inertia and gradient sparsity to the limitations of linear reasoning that prevent DLMs from reaching their ``GPT-4 moment''. We propose a strategic roadmap organized into four pillars: foundational infrastructure, algorithmic optimization, cognitive reasoning, and unified multimodal intelligence. By shifting toward a diffusion-native ecosystem characterized by multi-scale tokenization, active remasking, and latent thinking, we can move beyond the constraints of the causal horizon. We argue that this transition is essential for developing next-generation AI capable of complex structural reasoning, dynamic self-correction, and seamless multimodal integration.

Marie Siew, Shikhar Sharma, Zekai Li et al.

In edge computing, users' service profiles are migrated due to user mobility. Reinforcement learning (RL) frameworks have been proposed to do so, often trained on simulated data. However, existing RL frameworks overlook occasional server failures, which although rare, impact latency-sensitive applications like autonomous driving and real-time obstacle detection. Nevertheless, these failures (rare events), being not adequately represented in historical training data, pose a challenge for data-driven RL algorithms. As it is impractical to adjust failure frequency in real-world applications for training, we introduce FIRE, a framework that adapts to rare events by training a RL policy in an edge computing digital twin environment. We propose ImRE, an importance sampling-based Q-learning algorithm, which samples rare events proportionally to their impact on the value function. FIRE considers delay, migration, failure, and backup placement costs across individual and shared service profiles. We prove ImRE's boundedness and convergence to optimality. Next, we introduce novel deep Q-learning (ImDQL) and actor critic (ImACRE) versions of our algorithm to enhance scalability. We extend our framework to accommodate users with varying risk tolerances. Through trace driven experiments, we show that FIRE reduces costs compared to vanilla RL and the greedy baseline in the event of failures.

Jiaming Liang, Chao Xu, Shengze Cai

Flow visualization technologies such as particle tracking velocimetry (PTV) are broadly used in understanding the all-pervasiveness three-dimensional (3D) turbulent flow from nature and industrial processes. Despite the advances in 3D acquisition techniques, the developed motion estimation algorithms in particle tracking remain great challenges of large particle displacements, dense particle distributions and high computational cost. By introducing a novel deep neural network based on recurrent Graph Optimal Transport, called GotFlow3D, we present an end-to-end solution to learn the 3D fluid flow motion from double-frame particle sets. The proposed network constructs two graphs in the geometric and feature space and further enriches the original particle representations with the fused intrinsic and extrinsic features learnt from a graph neural network. The extracted deep features are subsequently utilized to make optimal transport plans indicating the correspondences of particle pairs, which are then iteratively and adaptively retrieved to guide the recurrent flow learning. Experimental evaluations, including assessments on numerical experiments and validations on real-world experiments, demonstrate that the proposed GotFlow3D achieves state-of-the-art performance against both recently-developed scene flow learners and particle tracking algorithms, with impressive accuracy, robustness and generalization ability, which can provide deeper insight into the complex dynamics of broad physical and biological systems.

Dongyang Liu, Renrui Zhang, Longtian Qiu et al. · stanford, tsinghua

We propose SPHINX-X, an extensive Multimodality Large Language Model (MLLM) series developed upon SPHINX. To improve the architecture and training efficiency, we modify the SPHINX framework by removing redundant visual encoders, bypassing fully-padded sub-images with skip tokens, and simplifying multi-stage training into a one-stage all-in-one paradigm. To fully unleash the potential of MLLMs, we assemble a comprehensive multi-domain and multimodal dataset covering publicly available resources in language, vision, and vision-language tasks. We further enrich this collection with our curated OCR intensive and Set-of-Mark datasets, extending the diversity and generality. By training over different base LLMs including TinyLlama1.1B, InternLM2-7B, LLaMA2-13B, and Mixtral8x7B, we obtain a spectrum of MLLMs that vary in parameter size and multilingual capabilities. Comprehensive benchmarking reveals a strong correlation between the multi-modal performance with the data and parameter scales. Code and models are released at https://github.com/Alpha-VLLM/LLaMA2-Accessory

Bo Zou, Chao Xu

Large language models (LLMs) face a dual challenge in creative capability evaluation: existing benchmarks (e.g., Story Cloze Test, HellaSwag) measure models' discriminative ability over narrative continuation using multiple-choice recognition paradigms, rather than directly measuring creative generation capability; rubric-based scoring and LLM-as-Judge methods rely on subjective dimension assessment or natural language model outputs, and cannot provide objective, automated scoring mechanisms. This paper proposes QUIET (Quality Understanding via Interlocked Evaluation Testing), a diagnostic benchmark for LLM creative capability based on multi-blank cascaded story cloze. QUIET sets N blanks (10-20) in a story with complete structure, with each blank accompanied by an explicit content constraint, and cascade dependency relationships between blanks -- the content filled into earlier blanks constrains the feasible solution space for later blanks. The evaluated model (or human participants) fills all blanks in open-ended generation mode; the results are scored by an information-theoretic automated scoring protocol without human grading. The scoring protocol directly operationalizes the "calibrated surprise" theoretical framework (Zou & Xu, 2026a). For each blank k, a composite score is computed: score = satisfy * (1 + lambda * surprise), where lambda = 1.0. Here, "satisfy" measures how well the blank filling satisfies the content constraint (objective logical reasoning judgment, not subjective aesthetic scoring), and "surprise" measures the degree of surprise given that the constraint is satisfied. Creative answers that do not satisfy the constraint score zero; answers that satisfy the constraint but are mediocre score low; answers that satisfy the constraint and are surprising score high.

Bo Zou, Chao Xu

This paper provides an empirical implementation of the creative quality metric proposed in Calibrated Surprise (Zou & Xu, 2026a). The question this paper addresses is: does this mathematical claim hold at the engineering level? To make the answer as general as possible, we deliberately choose the strictest engineering conditions: low data cost and a small base model. Training data comes from approximately 100 expert chain-of-thought (CoT) annotations produced by the BC Protocol (Zou & Xu, 2026b). We also identify a data bias: most publicly available alignment datasets are skewed toward craft-related knowledge, while audience modeling and reality-logic coverage are systematically weak. We use the term Creative Quality Alignment (CQA) to describe this class of engineering methods. We also offer a supporting theoretical observation: in an LLM with a single conditional distribution architecture, calibrating the appreciation side automatically transfers to the generation side via architectural duality. This is the structural reason why ~100 CoT examples are sufficient -- not a purely empirical observation like LIMA (Zhou et al., 2023).

Zeyu Cai, Yuliang Xiu, Renke Wang et al.

Fitting an underlying body model to 3D clothed human assets has been extensively studied, yet most approaches focus on either single-modal inputs such as point clouds or multi-view images alone, often requiring a known metric scale. This constraint is frequently impractical, especially for AI-generated assets where scale distortion is common. We propose OmniFit, a method that can seamlessly handle diverse multi-modal inputs, including full scans, partial depth observations, and image captures, while remaining scale-agnostic for both real and synthetic assets. Our key innovation is a simple yet effective conditional transformer decoder that directly maps surface points to dense body landmarks, which are then used for SMPL-X parameter fitting. In addition, an optional plug-and-play image adapter incorporates visual cues to compensate for missing geometric information. We further introduce a dedicated scale predictor that rescales subjects to canonical body proportions. OmniFit substantially outperforms state-of-the-art methods by 57.1 to 80.9 percent across daily and loose clothing scenarios. To the best of our knowledge, it is the first body fitting method to surpass multi-view optimization baselines and the first to achieve millimeter-level accuracy on the CAPE and 4D-DRESS benchmarks.

Jiaxu Liu, Song Chen, Shengze Cai et al.

The vanilla fractional order gradient descent may oscillatively converge to a region around the global minimum instead of converging to the exact minimum point, or even diverge, in the case where the objective function is strongly convex. To address this problem, a novel adaptive fractional order gradient descent (AFOGD) method and a novel adaptive fractional order accelerated gradient descent (AFOAGD) method are proposed in this paper. Inspired by the quadratic constraints and Lyapunov stability analysis from robust control theory, we establish a linear matrix inequality to analyse the convergence of our proposed algorithms. We prove that the proposed algorithms can achieve R-linear convergence when the objective function is $\textbf{L-}$smooth and $\textbf{m-}$strongly-convex. Several numerical simulations are demonstrated to verify the effectiveness and superiority of our proposed algorithms.

Yingte Shu, Yuchuan Tian, Chao Xu et al.

Diffusion language models (DLMs) have recently emerged as a strong alternative to autoregressive models by enabling parallel text generation. To improve inference efficiency and KV-cache compatibility, prior work commonly adopts block-based diffusion, decoding tokens block by block. However, this paradigm suffers from a structural limitation that we term Boundary-Induced Context Truncation (BICT): undecoded tokens near block boundaries are forced to commit without access to nearby future context, even when such context could substantially reduce uncertainty. This limitation degrades decoding certainty and generation quality, especially for tasks requiring precise reasoning, such as mathematical problem solving and code generation. We propose Deferred Commitment Decoding (DCD), a novel, training-free decoding strategy that mitigates this issue. DCD maintains a certainty-aware sliding window over masked tokens, resolving low-uncertainty tokens early while deferring high-uncertainty tokens until sufficient contextual evidence becomes available. Extensive experiments across multiple diffusion language models, benchmarks, and caching configurations show that DCD improves generation accuracy by 1.73% with comparable time on average compared to fixed block-based diffusion methods, with the most significant improvement reaching 16.5%. These results demonstrate that deferring token commitment based on uncertainty is a simple yet effective principle for improving both the quality and efficiency of diffusion language model decoding.

Jiazheng Xing, Chao Xu, Hangjie Yuan et al.

Multimodal Large Language Models (MLLMs) have propelled the field of few-shot action recognition (FSAR). However, preliminary explorations in this area primarily focus on generating captions to form a suboptimal feature->caption->feature pipeline and adopt metric learning solely within the visual space. In this paper, we propose FSAR-LLaVA, the first end-to-end method to leverage MLLMs (such as Video-LLaVA) as a multimodal knowledge base for directly enhancing FSAR. First, at the feature level, we leverage the MLLM's multimodal decoder to extract spatiotemporally and semantically enriched representations, which are then decoupled and enhanced by our Multimodal Feature-Enhanced Module into distinct visual and textual features that fully exploit their semantic knowledge for FSAR. Next, we leverage the versatility of MLLMs to craft input prompts that flexibly adapt to diverse scenarios, and use their aligned outputs to drive our designed Composite Task-Oriented Prototype Construction, effectively bridging the distribution gap between meta-train and meta-test sets. Finally, to enable multimodal features to guide metric learning jointly, we introduce a training-free Multimodal Prototype Matching Metric that adaptively selects the most decisive cues and efficiently leverages the decoupled feature representations produced by MLLMs. Extensive experiments demonstrate superior performance across various tasks with minimal trainable parameters.

Yuchuan Tian, Zhijun Tu, Hanting Chen et al.

Diffusion Transformers (DiTs) introduce the transformer architecture to diffusion tasks for latent-space image generation. With an isotropic architecture that chains a series of transformer blocks, DiTs demonstrate competitive performance and good scalability; but meanwhile, the abandonment of U-Net by DiTs and their following improvements is worth rethinking. To this end, we conduct a simple toy experiment by comparing a U-Net architectured DiT with an isotropic one. It turns out that the U-Net architecture only gain a slight advantage amid the U-Net inductive bias, indicating potential redundancies within the U-Net-style DiT. Inspired by the discovery that U-Net backbone features are low-frequency-dominated, we perform token downsampling on the query-key-value tuple for self-attention that bring further improvements despite a considerable amount of reduction in computation. Based on self-attention with downsampled tokens, we propose a series of U-shaped DiTs (U-DiTs) in the paper and conduct extensive experiments to demonstrate the extraordinary performance of U-DiT models. The proposed U-DiT could outperform DiT-XL/2 with only 1/6 of its computation cost. Codes are available at https://github.com/YuchuanTian/U-DiT.

Linke Ouyang, Yuan Qu, Hongbin Zhou et al.

Document content extraction is a critical task in computer vision, underpinning the data needs of large language models (LLMs) and retrieval-augmented generation (RAG) systems. Despite recent progress, current document parsing methods have not been fairly and comprehensively evaluated due to the narrow coverage of document types and the simplified, unrealistic evaluation procedures in existing benchmarks. To address these gaps, we introduce OmniDocBench, a novel benchmark featuring high-quality annotations across nine document sources, including academic papers, textbooks, and more challenging cases such as handwritten notes and densely typeset newspapers. OmniDocBench supports flexible, multi-level evaluations--ranging from an end-to-end assessment to the task-specific and attribute--based analysis using 19 layout categories and 15 attribute labels. We conduct a thorough evaluation of both pipeline-based methods and end-to-end vision-language models, revealing their strengths and weaknesses across different document types. OmniDocBench sets a new standard for the fair, diverse, and fine-grained evaluation in document parsing. Dataset and code are available at https://github.com/opendatalab/OmniDocBench.

Chao Xu, Xu Zhang, Zihang Luo et al.

Reducing collective communication latency is a critical goal for large model training and inference in both academia and industry. Many-to-many communications, such as AllGather and AlltoAll (dispatch), are core components of modern parallelization strategies. State-of-the-art implementations of these communications rely on unicast-based writes and transmit duplicate copies of the same data across physical links for multiple receivers. This redundant transmission congests network bottlenecks and degrades end-to-end latency. We present MultiWrite, a novel many-to-many transmission semantic that eliminates redundant packets to directly reduce operator latency. MultiWrite adopts multicast principles while addressing critical limitations of traditional multicast for AI workloads. These limitations include heavy management plane overhead and ecosystem compatibility issues. We implement MultiWrite on Ascend NPUs. Long-term stress tests demonstrate that our MultiWrite-based operators achieve up to 33% latency reduction on commercially deployed devices.

Juncheng Chen, Tiancheng Lai, Xingpeng Wang et al.

Time-of-Flight (ToF) cameras possess compact design and high measurement precision to be applied to various robot tasks. However, their limited sensing range restricts deployment in large-scale scenarios. Depth completion has emerged as a potential solution to expand the sensing range of ToF cameras, but existing research lacks dedicated datasets and struggles to generalize to ToF measurements. In this paper, we propose a full-stack framework that enables depth completion in large-scale scenarios for short-range ToF cameras. First, we construct a multi-sensor platform with a reconstruction-based pipeline to collect real-world ToF samples with dense large-scale ground truth, yielding the first LArge-ScalE scenaRio ToF depth completion dataset (LASER-ToF). Second, we propose a sensor-aware depth completion network that incorporates a novel 3D branch with a 3D-2D Joint Propagation Pooling (JPP) module and Multimodal Cross-Covariance Attention (MXCA), enabling effective modeling of long-range relationships and efficient 3D-2D fusion under non-uniform ToF depth sparsity. Moreover, our network can utilize the sparse point cloud from visual SLAM as a supplement to ToF depth to further improve prediction accuracy. Experiments show that our method achieves an 8.6% lower mean absolute error than the second-best method, while maintaining lightweight design to support onboard deployment. Finally, to verify the system's applicability on real robots, we deploy proposed method on a quadrotor at a 10Hz runtime, enabling reliable large-scale mapping and long-range planning in challenging environments for short-range ToF cameras.