Explore · 每周精选
发现最受关注的研究论文,追踪研究趋势,订阅感兴趣的期刊与关键词。
Mengting Chen, Zhengrui Chen, Yongchao Du +16 authors
A commercial-scale virtual try-on system achieves high success rates, photorealistic results, and real-time performance through integrated system design and multi-stage training.
50 篇论文 · 按点赞排序
Inclusion AI, Tiwei Bie, Haoxing Chen +15 authors
LLaDA2.0-Uni is a unified discrete diffusion language model that integrates multimodal understanding and generation through a semantic discrete tokenizer, MoE-based backbone, and diffusion decoder, achieving performance comparable to specialized vision-language models while enabling efficient inference and high-fidelity image generation.
Pengcheng Wang, Jerry Huang, Jiarui Yao +7 authors
AgentSPEX is a domain-specific language and framework for creating structured, modular, and interpretable large language model agent workflows with explicit control flow and state management.
Jinghui Lu, Jiayi Guan, Zhijian Huang +47 authors
OneVL presents a unified vision-language-action framework that improves latent chain-of-thought reasoning for autonomous driving by integrating language and visual world model supervision for faster, more accurate trajectory prediction.
Chenxi Zhao, Chen Zhu, Xiaokun Feng +6 authors
Researchers extend MeanFlow generation from class labels to text inputs by integrating powerful LLM-based text encoders, overcoming limitations of few-step refinement through enhanced semantic feature representation.
Guanting Dong, Junting Lu, Junjie Huang +17 authors
Agent-World introduces a self-evolving training framework that advances general agent intelligence through autonomous environment discovery and continuous learning across diverse real-world scenarios.
Xiangyang Luo, Xiaozhe Xin, Tao Feng +3 authors
CoInteract presents an end-to-end framework for human-object interaction video synthesis using a Diffusion Transformer backbone with specialized modules for structural stability and physical plausibility.
Yilei Jiang, Jinyuan Hu, Qianyin Xiao +8 authors
OpenGame is an open-source agentic framework for end-to-end web game creation that uses specialized code models and evaluation benchmarks to overcome challenges in interactive application development.
Chuanyu Qin, Chenxu Yang, Qingyi Si +6 authors
Mixed-policy reinforcement learning approach using near-future policy optimization to accelerate convergence and improve performance by balancing trajectory quality and variance.
Meng Yu, Lei Sun, Jianhao Zeng +2 authors
Diffusion probabilistic models suffer from SNR-timestep bias during inference, which is addressed through a differential correction method that processes frequency components separately, improving generation quality across multiple models with minimal computational cost.
Xiaofan Li, Ming Yang, Zhiyuan Ma +9 authors
A novel reinforcement learning approach for large language models that addresses the exploration-exploitation trade-off through perplexity-based sample partitioning and bidirectional reward allocation mechanisms.
Qwen Team
Qwen3.5-Omni is a large-scale multimodal model with hundreds of billions of parameters that excels in audio-visual understanding and generation, featuring advanced architectures and novel capabilities like Audio-Visual Vibe Coding.
Ido Galil, Moshe Kimhi, Ran El-Yaniv
Deep neural networks exhibit catastrophic vulnerability to minimal parameter bit flips across multiple domains, which can be identified and mitigated through targeted protection strategies.
Venus Team, Sunhao Dai, Yong Deng +10 authors
DR-Venus-4B is a 4-billion-parameter deep research agent trained entirely on open data using agentic supervised fine-tuning and reinforcement learning with turn-level rewards to achieve superior performance on research benchmarks while maintaining edge-scale deployment advantages.
Ying Zeng, Miaosen Luo, Guangyuan Li +10 authors
SmartPhotoCrafter automates photographic image editing by combining image quality comprehension with targeted enhancement, using a reasoning-to-generation approach that eliminates the need for explicit human instructions.
Haoyu Wu, Jiwen Yu, Yingtian Zou +1 authors
MultiWorld is a unified framework for multi-agent multi-view world modeling that achieves accurate multi-agent control while maintaining multi-view consistency through specialized modules for condition handling and global state encoding.
Chang Nie, Chaoyou Fu, Yifan Zhang +2 authors
A novel personalized multimodal language model framework called PersonaVLM is introduced that enables long-term personalization through memory retention, multi-turn reasoning, and response alignment capabilities.
Chuanyu Qin, Chenxu Yang, Qingyi Si +6 authors
EasyVideoR1 presents an efficient reinforcement learning framework for video understanding that improves training throughput, supports diverse video tasks, and enables joint image-video training with comprehensive evaluation across multiple benchmarks.
Yutian Chen, Shi Guo, Renbiao Jin +7 authors
AnyRecon enables scalable 3D reconstruction from arbitrary sparse inputs using diffusion models with persistent scene memory and geometry-aware conditioning for improved geometric consistency.
Xiaojie Xu, Zhengyuan Lin, Kang He +5 authors
WorldMark establishes a standardized benchmark for evaluating interactive video generation models with unified controls, identical scenarios, and comprehensive evaluation metrics across multiple model architectures.
Qingyang Zhang, Xinke Kong, Haitao Wu +7 authors
TEMPO is a test-time training framework that alternates policy refinement with critic recalibration to sustain performance improvements in language models without diversity collapse.
Boyu Chen, Yi Chen, Lu Qiu +3 authors
UniT enables human-to-humanoid transfer by creating a unified visual-language representation that bridges kinematic differences through cross-reconstruction mechanisms and shared latent spaces.
Xiaohua Wang, Muzhao Tian, Yuqi Zeng +20 authors
Reward hacking in aligned language models stems from optimizing expressive policies against compressed reward signals, leading to systematic misalignment behaviors that generalize beyond initial shortcuts.
Xirui Li, Ming Li, Derry Xu +4 authors
An automated pipeline generates diverse, verified environments for claw-like agents from natural language descriptions, enabling large-scale benchmark construction and continuous evaluation.
Xianming Li, Zongxi Li, Tsz-fung Andrew Lee +3 authors
ShadowPEFT is a parameter-efficient fine-tuning framework that performs layer-level refinement through depth-shared shadow modules, offering competitive performance with reduced computational overhead compared to traditional low-rank adaptation methods.
Kanzhi Cheng, Zehao Li, Zheng Ma +11 authors
An open-source framework for mobile agent training that synthesizes task instructions and trajectories through scalable pipelines and policy-switching strategies, achieving superior performance on AndroidWorld benchmarks.
Kwan Yun, Changmin Lee, Ayeong Jeong +3 authors
StyleID presents a human perception-aware dataset and evaluation framework for facial identity preservation under stylization, featuring two datasets derived from psychometric experiments and calibrated semantic encoders that improve correlation with human judgments.
Valentin Gabeur, Shangbang Long, Songyou Peng +22 authors
Image generation pretraining enables vision models to develop strong visual understanding capabilities, achieving state-of-the-art performance on diverse vision tasks through lightweight instruction-tuning while maintaining generation abilities.
Zhiyuan Peng, Wei Tao, Xin Yin +3 authors
Large language models struggle to generate logically correct GUI applications, prompting the development of PlayEval benchmark and PlayCoder framework that uses multi-agent approaches to improve functional correctness through iterative repair.
Uday Allu, Sonu Kedia, Tanmay Odapally +1 authors
Web Retrieval-Aware Chunking (W-RAC) introduces a cost-efficient framework for web document processing that reduces LLM token usage and hallucination risks through structured content representation and retrieval-aware grouping decisions.
北京市昌平区探索星信息技术及软件开发工作室
京ICP备2026059466号