Yiyu Wang

Yiyu Wang, Jungang Xu, Yingfei Sun

CNN-LSTM based architectures have played an important role in image captioning, but limited by the training efficiency and expression ability, researchers began to explore the CNN-Transformer based models and achieved great success. Meanwhile, almost all recent works adopt Faster R-CNN as the backbone encoder to extract region-level features from given images. However, Faster R-CNN needs a pre-training on an additional dataset, which divides the image captioning task into two stages and limits its potential applications. In this paper, we build a pure Transformer-based model, which integrates image captioning into one stage and realizes end-to-end training. Firstly, we adopt SwinTransformer to replace Faster R-CNN as the backbone encoder to extract grid-level features from given images; Then, referring to Transformer, we build a refining encoder and a decoder. The refining encoder refines the grid features by capturing the intra-relationship between them, and the decoder decodes the refined features into captions word by word. Furthermore, in order to increase the interaction between multi-modal (vision and language) features to enhance the modeling capability, we calculate the mean pooling of grid features as the global feature, then introduce it into refining encoder to refine with grid features together, and add a pre-fusion process of refined global feature and generated words in decoder. To validate the effectiveness of our proposed model, we conduct experiments on MSCOCO dataset. The experimental results compared to existing published works demonstrate that our model achieves new state-of-the-art performances of 138.2% (single model) and 141.0% (ensemble of 4 models) CIDEr scores on `Karpathy' offline test split and 136.0% (c5) and 138.3% (c40) CIDEr scores on the official online test server. Trained models and source code will be released.

Xuyang Liu, Yiyu Wang, Junpeng Ma et al.

Video large language models (VideoLLM) excel at video understanding, but face efficiency challenges due to the quadratic complexity of abundant visual tokens. Our systematic analysis of token compression methods for VideoLLMs reveals two critical issues: (i) overlooking distinctive visual signals across frames, leading to information loss; (ii) suffering from implementation constraints, causing incompatibility with modern architectures or efficient operators. To address these challenges, we distill three design principles for VideoLLM token compression and propose a plug-and-play inference acceleration framework "Video Compression Commander" (VidCom2). By quantifying each frame's uniqueness, VidCom2 adaptively adjusts compression intensity across frames, effectively preserving essential information while reducing redundancy in video sequences. Extensive experiments across various VideoLLMs and benchmarks demonstrate the superior performance and efficiency of our VidCom2. With only 25% visual tokens, VidCom2 achieves 99.6% of the original performance on LLaVA-OV while reducing 70.8% of the LLM generation latency. Notably, our Frame Compression Adjustment strategy is compatible with other token compression methods to further improve their performance. Our code is available at https://github.com/xuyang-liu16/VidCom2.

Yuhang Han, Yuyang Wu, Zhengbo Jiao et al.

Reinforcement Learning from Verifiable Rewards (RLVR) has substantially enhanced the reasoning capabilities of large language models in abstract reasoning tasks. However, its application to Large Vision-Language Models (LVLMs) remains constrained by a structural representational bottleneck. Existing approaches generally lack explicit modeling and effective utilization of visual information, preventing visual representations from being tightly coupled with the reinforcement learning optimization process and thereby limiting further improvements in multimodal reasoning performance. To address this limitation, we propose KAWHI (Key-Region Aligned Weighted Harmonic Incentive), a plug-and-play reward reweighting mechanism that explicitly incorporates structured visual information into uniform reward policy optimization methods (e.g., GRPO and GSPO). The method adaptively localizes semantically salient regions through hierarchical geometric aggregation, identifies vision-critical attention heads via structured attribution, and performs paragraph-level credit reallocation to align spatial visual evidence with semantically decisive reasoning steps. Extensive empirical evaluations on diverse reasoning benchmarks substantiate KAWHI as a general-purpose enhancement module, consistently improving the performance of various uniform reward optimization methods. Project page: KAWHI (https://kawhiiiileo.github.io/KAWHI_PAGE/)

Jifang Wang, Xue Yang, Longyue Wang et al.

Conditional image generation has gained significant attention for its ability to personalize content. However, the field faces challenges in developing task-agnostic, reliable, and explainable evaluation metrics. This paper introduces CIGEval, a unified agentic framework for comprehensive evaluation of conditional image generation tasks. CIGEval utilizes large multimodal models (LMMs) as its core, integrating a multi-functional toolbox and establishing a fine-grained evaluation framework. Additionally, we synthesize evaluation trajectories for fine-tuning, empowering smaller LMMs to autonomously select appropriate tools and conduct nuanced analyses based on tool outputs. Experiments across seven prominent conditional image generation tasks demonstrate that CIGEval (GPT-4o version) achieves a high correlation of 0.4625 with human assessments, closely matching the inter-annotator correlation of 0.47. Moreover, when implemented with 7B open-source LMMs using only 2.3K training trajectories, CIGEval surpasses the previous GPT-4o-based state-of-the-art method. Case studies on GPT-4o image generation highlight CIGEval's capability in identifying subtle issues related to subject consistency and adherence to control guidance, indicating its great potential for automating evaluation of image generation tasks with human-level reliability.

Xinying Lin, Xuyang Liu, Yiyu Wang et al.

Video large language models (VideoLLMs) show strong capability in video understanding, yet long-context inference is still dominated by massive redundant visual tokens in the prefill stage. We revisit token compression for VideoLLMs under a tight budget and identify a key bottleneck, namely insufficient spatio-temporal information coverage. Existing methods often introduce discontinuous coverage through coarse per-frame allocation or scene segmentation, and token merging can further misalign spatio-temporal coordinates under MRoPE-style discrete (t,h,w) bindings. To address these issues, we propose V-CAST (Video Curvature-Aware Spatio-Temporal Pruning), a training-free, plug-and-play pruning policy for long-context video inference. V-CAST casts token compression as a trajectory approximation problem and introduces a curvature-guided temporal allocation module that routes per-frame token budgets to semantic turns and event boundaries. It further adopts a dual-anchor spatial selection mechanism that preserves high-entropy visual evidence without attention intervention, while keeping retained tokens at their original coordinates to maintain positional alignment. Extensive experiments across multiple VideoLLMs of different architectures and scales demonstrate that V-CAST achieves 98.6% of the original performance, outperforms the second-best method by +1.1% on average, and reduces peak memory and total latency to 86.7% and 86.4% of vanilla Qwen3-VL-8B-Instruct.

Zhenran Xu, Xue Yang, Yiyu Wang et al.

We introduce ComfyUI-Copilot, a large language model-powered plugin designed to enhance the usability and efficiency of ComfyUI, an open-source platform for AI-driven art creation. Despite its flexibility and user-friendly interface, ComfyUI can present challenges to newcomers, including limited documentation, model misconfigurations, and the complexity of workflow design. ComfyUI-Copilot addresses these challenges by offering intelligent node and model recommendations, along with automated one-click workflow construction. At its core, the system employs a hierarchical multi-agent framework comprising a central assistant agent for task delegation and specialized worker agents for different usages, supported by our curated ComfyUI knowledge bases to streamline debugging and deployment. We validate the effectiveness of ComfyUI-Copilot through both offline quantitative evaluations and online user feedback, showing that it accurately recommends nodes and accelerates workflow development. Additionally, use cases illustrate that ComfyUI-Copilot lowers entry barriers for beginners and enhances workflow efficiency for experienced users. The ComfyUI-Copilot installation package and a demo video are available at https://github.com/AIDC-AI/ComfyUI-Copilot.

Yiyu Wang, Xuyang Liu, Xiyan Gui et al.

Streaming Video Large Language Models (VideoLLMs) have demonstrated impressive performance across various video understanding tasks, but they face significant challenges in real-time deployment due to the high computational cost of processing dense visual tokens from continuous video streams. In streaming video scenarios, the primary bottleneck lies in the Vision Transformer (ViT) encoding stage, where redundant processing of temporally similar frames leads to inefficiency. Additionally, inflated token sequences during LLM pre-filling further exacerbate latency and memory overhead. To address these challenges, we propose \textbf{S}treaming \textbf{T}oken \textbf{C}ompression (\textbf{STC}), a plug-and-play hierarchical framework that seamlessly integrates into existing streaming VideoLLMs, optimizing both ViT encoding and LLM pre-filling stages to accelerate processing. STC introduces two token-level accelerators: \textbf{STC-Cacher}, which reduces ViT encoding overhead by caching and reusing features from temporally similar frames, and \textbf{STC-Pruner}, which compresses the visual token sequence before it enters the LLM, preserving only the most salient tokens based on both spatial and temporal relevance. Extensive experiments on four baseline streaming VideoLLMs across five benchmarks demonstrate that STC outperforms other compression methods. Notably, STC retains up to \textbf{99\%} of accuracy on the ReKV framework while reducing ViT encoding latency and LLM pre-filling latency by \textbf{24.5\%} and \textbf{45.3\%}.

Chenfei Liao, Wensong Wang, Zichen Wen et al.

Recent endeavors to accelerate inference in Multimodal Large Language Models (MLLMs) have primarily focused on visual token compression. The effectiveness of these methods is typically assessed by measuring the accuracy drop on established benchmarks, comparing model performance before and after compression. However, these benchmarks are originally designed to assess the perception and reasoning capabilities of MLLMs, rather than to evaluate compression techniques. As a result, directly applying them to visual token compression introduces a task mismatch. Strikingly, our investigation reveals that simple image downsampling consistently outperforms many advanced compression methods across multiple widely used benchmarks. Through extensive experiments, we make the following observations: (i) Current benchmarks are noisy for the visual token compression task. (ii) Down-sampling is able to serve as a data filter to evaluate the difficulty of samples in the visual token compression task. Motivated by these findings, we introduce VTC-Bench, an evaluation framework that incorporates a data filtering mechanism to denoise existing benchmarks, thereby enabling fairer and more accurate assessment of visual token compression methods. All data and code are available at https://github.com/Chenfei-Liao/VTC-Bench.

Kun Jing, Luoyu Chen, Jungang Xu et al.

Neural architecture search (NAS) is a promising approach for automatically designing neural network architectures. However, the architecture estimation of NAS is computationally expensive and time-consuming because of training multiple architectures from scratch. Although existing zero-shot NAS methods use training-free proxies to accelerate the architecture estimation, their effectiveness, stability, and generality are still lacking. We present a novel training-free estimation proxy called weighted response correlation (WRCor). WRCor utilizes correlation coefficient matrices of responses across different input samples to calculate the proxy scores of estimated architectures, which can measure their expressivity and generalizability. Experimental results on proxy evaluation demonstrate that WRCor and its voting proxies are more efficient estimation strategies than existing proxies. We also apply them with different search strategies in architecture search. Experimental results on architecture search show that our zero-shot NAS algorithm outperforms most existing NAS algorithms in different search spaces. Our NAS algorithm can discover an architecture with a 22.1% test error on the ImageNet-1k dataset within 4 GPU hours. All codes are publicly available at https://github.com/kunjing96/ZSNAS-WRCor.git.

Hongwei Zhang, Qiqiang Zhong, Jiangxia Cao et al.

Modeling ultra-long user sequences involves a difficult trade-off between efficiency and effectiveness. While current paradigms rely on either item-specific search or item-agnostic compression, we propose UxSID, a framework exploring a third path: semantic-group shared interest memory. By utilizing Semantic IDs (SIDs) and a dual-level attention strategy, UxSID captures target-aware preferences without the heavy cost of item-specific models. This end-to-end architecture balances computational parsimony with semantic awareness, achieving state-of-the-art performance and a 0.337% revenue lift in large-scale advertising A/B test.

Xuyang Liu, Zichen Wen, Shaobo Wang et al.

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

Zhenran Xu, Longyue Wang, Jifang Wang et al.

Virtual film production requires intricate decision-making processes, including scriptwriting, virtual cinematography, and precise actor positioning and actions. Motivated by recent advances in automated decision-making with language agent-based societies, this paper introduces FilmAgent, a novel LLM-based multi-agent collaborative framework for end-to-end film automation in our constructed 3D virtual spaces. FilmAgent simulates various crew roles, including directors, screenwriters, actors, and cinematographers, and covers key stages of a film production workflow: (1) idea development transforms brainstormed ideas into structured story outlines; (2) scriptwriting elaborates on dialogue and character actions for each scene; (3) cinematography determines the camera setups for each shot. A team of agents collaborates through iterative feedback and revisions, thereby verifying intermediate scripts and reducing hallucinations. We evaluate the generated videos on 15 ideas and 4 key aspects. Human evaluation shows that FilmAgent outperforms all baselines across all aspects and scores 3.98 out of 5 on average, showing the feasibility of multi-agent collaboration in filmmaking. Further analysis reveals that FilmAgent, despite using the less advanced GPT-4o model, surpasses the single-agent o1, showing the advantage of a well-coordinated multi-agent system. Lastly, we discuss the complementary strengths and weaknesses of OpenAI's text-to-video model Sora and our FilmAgent in filmmaking.

Lingfeng Ming, Bo Zeng, Chenyang Lyu et al.

Large Language Models (LLMs) have achieved remarkable progress in recent years; however, their excellent performance is still largely limited to major world languages, primarily English. Many LLMs continue to face challenges with multilingual tasks, especially when it comes to low-resource languages. To address this issue, we introduced Marco-LLM: Massive multilingual training for cross-lingual enhancement LLM. We have collected a substantial amount of multilingual data for several low-resource languages and conducted extensive continual pre-training using the Qwen2 models. This effort has resulted in a multilingual LLM named Marco-LLM. Through comprehensive evaluations on various multilingual benchmarks, including MMMLU, AGIEval, Belebele, Flores-200, XCOPA and many others, Marco-LLM has demonstrated substantial improvements over state-of-the-art LLMs. Furthermore, Marco-LLM achieved substantial enhancements in any-to-any machine translation tasks, showing the effectiveness of our multilingual LLM. Marco-LLM is a pioneering multilingual LLM designed to not only perform exceptionally well in multilingual tasks, including low-resource languages, but also maintain strong performance in English and other major languages, closing the performance gap between high- and low-resource language capabilities. By bridging languages, this effort demonstrates our dedication to ensuring LLMs work accurately across various languages.

Junjie Chen, Xuyang Liu, Zichen Wen et al.

Large vision-language models (LVLMs) have demonstrated remarkable capabilities in multimodal understanding tasks. However, the increasing demand for high-resolution image and long-video understanding results in substantial token counts, leading to reduced inference efficiency. Token compression offers a direct solution by reducing the number of tokens to be processed, thereby improving computational efficiency. Through extensive analysis, we identify two critical limitations in existing inner-LLM token compression methods: positional bias and incompatibility with efficient operators, which hinder their practical deployment for LVLM acceleration. This paper presents the first approach from a token variation perspective, revealing that visual token variations within LLMs exhibit task-agnostic properties. We propose Variation-aware Vision Token Dropping (\textit{i.e.}, \textbf{V$^2$Drop}), which progressively removes visual tokens with minimal variation during LVLM inference, thereby enhancing computational efficiency. Extensive experiments across multiple models and benchmarks demonstrate that our V$^2$Drop is able to maintain \textbf{94.0\%} and \textbf{98.6\%} of the original model performance for image and video understanding tasks respectively, while reducing LLM generation latency by \textbf{31.5\%} and \textbf{74.2\%}. When combined with efficient operators, V$^2$Drop further reduces GPU peak memory usage.

Yiyu Wang, Hao Luo, Jungang Xu et al.

Supervised image captioning approaches have made great progress, but it is challenging to collect high-quality human-annotated image-text data. Recently, large-scale vision and language models (e.g., CLIP) and large-scale generative language models (e.g., GPT-2) have shown strong performances in various tasks, which also provide some new solutions for image captioning with web paired data, unpaired data or even text-only data. Among them, the mainstream solution is to project image embeddings into the text embedding space with the assistance of consistent representations between image-text pairs from the CLIP model. However, the current methods still face several challenges in adapting to the diversity of data configurations in a unified solution, accurately estimating image-text embedding bias, and correcting unsatisfactory prediction results in the inference stage. This paper proposes a new Text data-centric approach with Interactive Prompts for image Captioning, named TIPCap. 1) We consider four different settings which gradually reduce the dependence on paired data. 2) We construct a mapping module driven by multivariate Gaussian distribution to mitigate the modality gap, which is applicable to the above four different settings. 3) We propose a prompt interaction module that can incorporate optional prompt information before generating captions. Extensive experiments show that our TIPCap outperforms other weakly or unsupervised image captioning methods and achieves a new state-of-the-art performance on two widely used datasets, i.e., MS-COCO and Flickr30K.

Zichen Wen, Yiyu Wang, Chenfei Liao et al.

In an era where AI is evolving from a passive tool into an active and adaptive companion, we introduce AI for Service (AI4Service), a new paradigm that enables proactive and real-time assistance in daily life. Existing AI services remain largely reactive, responding only to explicit user commands. We argue that a truly intelligent and helpful assistant should be capable of anticipating user needs and taking actions proactively when appropriate. To realize this vision, we propose Alpha-Service, a unified framework that addresses two fundamental challenges: Know When to intervene by detecting service opportunities from egocentric video streams, and Know How to provide both generalized and personalized services. Inspired by the von Neumann computer architecture and based on AI glasses, Alpha-Service consists of five key components: an Input Unit for perception, a Central Processing Unit for task scheduling, an Arithmetic Logic Unit for tool utilization, a Memory Unit for long-term personalization, and an Output Unit for natural human interaction. As an initial exploration, we implement Alpha-Service through a multi-agent system deployed on AI glasses. Case studies, including a real-time Blackjack advisor, a museum tour guide, and a shopping fit assistant, demonstrate its ability to seamlessly perceive the environment, infer user intent, and provide timely and useful assistance without explicit prompts.

Zhenran Xu, Yiyu Wang, Xue Yang et al.

AI-generated content has evolved from monolithic models to modular workflows, particularly on platforms like ComfyUI, enabling customization in creative pipelines. However, crafting effective workflows requires great expertise to orchestrate numerous specialized components, presenting a steep learning curve for users. To address this challenge, we introduce ComfyUI-R1, the first large reasoning model for automated workflow generation. Starting with our curated dataset of 4K workflows, we construct long chain-of-thought (CoT) reasoning data, including node selection, workflow planning, and code-level workflow representation. ComfyUI-R1 is trained through a two-stage framework: (1) CoT fine-tuning for cold start, adapting models to the ComfyUI domain; (2) reinforcement learning for incentivizing reasoning capability, guided by a fine-grained rule-metric hybrid reward, ensuring format validity, structural integrity, and node-level fidelity. Experiments show that our 7B-parameter model achieves a 97\% format validity rate, along with high pass rate, node-level and graph-level F1 scores, significantly surpassing prior state-of-the-art methods that employ leading closed-source models such as GPT-4o and Claude series. Further analysis highlights the critical role of the reasoning process and the advantage of transforming workflows into code. Qualitative comparison reveals our strength in synthesizing intricate workflows with diverse nodes, underscoring the potential of long CoT reasoning in AI art creation.

Taohan Weng, Kaibing Hu, Henan Liu et al.

Ultrasound is crucial in modern medicine but faces challenges like operator dependence, image noise, and real-time scanning, hindering AI integration. While large multimodal models excel in other medical imaging areas, they struggle with ultrasound's complexities. To address this, we introduce Dolphin v1.0 (V1) and its reasoning-augmented version, Dolphin R1-the first large-scale multimodal ultrasound foundation models unifying diverse clinical tasks in a single vision-language framework.To tackle ultrasound variability and noise, we curated a 2-million-scale multimodal dataset, combining textbook knowledge, public data, synthetic samples, and general corpora. This ensures robust perception, generalization, and clinical adaptability.The Dolphin series employs a three-stage training strategy: domain-specialized pretraining, instruction-driven alignment, and reinforcement-based refinement. Dolphin v1.0 delivers reliable performance in classification, detection, regression, and report generation. Dolphin R1 enhances diagnostic inference, reasoning transparency, and interpretability through reinforcement learning with ultrasound-specific rewards.Evaluated on U2-Bench across eight ultrasound tasks, Dolphin R1 achieves a U2-score of 0.5835-over twice the second-best model (0.2968) setting a new state of the art. Dolphin v1.0 also performs competitively, validating the unified framework. Comparisons show reasoning-enhanced training significantly improves diagnostic accuracy, consistency, and interpretability, highlighting its importance for high-stakes medical AI.

Eli Sennesh, Zulqarnain Khan, Yiyu Wang et al.

Neuroimaging studies produce gigabytes of spatio-temporal data for a small number of participants and stimuli. Rarely do researchers attempt to model and examine how individual participants vary from each other -- a question that should be addressable even in small samples given the right statistical tools. We propose Neural Topographic Factor Analysis (NTFA), a probabilistic factor analysis model that infers embeddings for participants and stimuli. These embeddings allow us to reason about differences between participants and stimuli as signal rather than noise. We evaluate NTFA on data from an in-house pilot experiment, as well as two publicly available datasets. We demonstrate that inferring representations for participants and stimuli improves predictive generalization to unseen data when compared to previous topographic methods. We also demonstrate that the inferred latent factor representations are useful for downstream tasks such as multivoxel pattern analysis and functional connectivity.

Yiyu Wang, Jungang Xu, Yingfei Sun et al.

Image captioning is a challenging task and attracting more and more attention in the field of Artificial Intelligence, and which can be applied to efficient image retrieval, intelligent blind guidance and human-computer interaction, etc. In this paper, we present a survey on advances in image captioning based on Deep Learning methods, including Encoder-Decoder structure, improved methods in Encoder, improved methods in Decoder, and other improvements. Furthermore, we discussed future research directions.