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。
B. Y. Yan, Chaofan Li, Hongjin Qian +2 authors
GAM, a novel framework that employs JIT compilation principles, improves memory efficiency and task completion by leveraging a lightweight memorizer and researcher in conjunction with reinforcement learning.
50 篇论文 · 按点赞排序
Nicolas Carion, Laura Gustafson, Yuan-Ting Hu +35 authors
Segment Anything Model 3 achieves state-of-the-art performance in promptable concept segmentation and tracking by leveraging a unified model architecture with decoupled recognition and localization.
Jiaru Zou, Xiyuan Yang, Ruizhong Qiu +10 authors
LatentMAS enables efficient, lossless collaboration among LLM agents in latent space, improving performance and reducing computational costs compared to text-based methods.
Valentin Khrulkov, Andrey Galichin, Denis Bashkirov +5 authors
GigaEvo is an open-source framework for LLM-guided evolutionary computation, offering modular and concurrent tools for research and experimentation in solving complex optimization problems.
Yikun Wang, Zuyan Liu, Ziyi Wang +3 authors
GeoVista, an agentic model integrating tool invocation and reinforcement learning, achieves high geolocalization performance on GeoBench, outperforming open-source models and matching closed-source models.
Kaichen Zhang, Keming Wu, Zuhao Yang +5 authors
OpenMMReasoner, a two-stage training approach combining supervised fine-tuning and reinforcement learning, enhances multimodal reasoning performance through rigorous data curation and improved training strategies.
Jiayi Zhang, Yiran Peng, Fanqi Kong +12 authors
AutoEnv and AutoEnv-36 provide a standardized framework and dataset for evaluating cross-environment learning in agents, highlighting the challenges and limitations of existing learning methods.
Vladislav Pedashenko, Laida Kushnareva, Yana Khassan Nibal +5 authors
The study explores intrinsic dimension in large language models through cross-encoder analysis, linguistic features, and sparse autoencoders, revealing its independence from entropy, genre-specific stratification, and causal features related to text type.
Artem Chervyakov, Ulyana Isaeva, Anton Emelyanov +15 authors
Mera Multi is an open multimodal evaluation framework for Russian-spoken architectures, addressing the lack of such benchmarks with 18 newly constructed tasks and a methodology to prevent benchmark leakage.
Zehong Ma, Longhui Wei, Shuai Wang +2 authors
The frequency-DeCoupled pixel diffusion framework improves image generation efficiency and quality by separating high-frequency details and low-frequency semantics, achieving superior performance compared to existing pixel diffusion models.
Rulin Shao, Akari Asai, Shannon Zejiang Shen +18 authors
Reinforcement Learning with Evolving Rubrics (RLER) enables training of deep research models for long-form tasks, outperforming existing models and proprietary systems while being more cost-effective.
Anglin Liu, Rundong Xue, Xu R. Cao +5 authors
MedSAM-3, a text-promptable medical segmentation model fine-tuned on SAM 3 architecture, achieves superior performance across various medical imaging modalities using semantic conceptual labels and multimodal large language models.
Kevin Qinghong Lin, Siyuan Hu, Linjie Li +4 authors
A framework leveraging Computer-Use Agents as judges to assist coding-oriented language models in designing efficient and functional GUIs.
Xiaoyue Mi, Wenqing Yu, Jiesong Lian +9 authors
PRFL optimizes video generation preferences in latent space, improving alignment with human preferences while reducing memory consumption and training time.
Inferix Team, Tianyu Feng, Yizeng Han +13 authors
Inferix is a next-generation inference engine designed for immersive world synthesis using semi-autoregressive decoding, combining diffusion and autoregressive methods for high-quality, real-time video generation and interaction.
Jiaqi Liu, Kaiwen Xiong, Peng Xia +6 authors
Agent0-VL, a self-evolving vision-language agent, incorporates tool usage into both reasoning and self-evaluation, enabling continual improvement through evidence-grounded analysis and reinforcement learning.
Wei He, Kai Han, Hang Zhou +4 authors
ROOT, a robust optimizer, enhances training stability and convergence for large language models by addressing dimensional fragility and outlier noise through adaptive Newton iterations and proximal optimization.
Chang Gao, Chujie Zheng, Xiong-Hui Chen +7 authors
Soft Adaptive Policy Optimization (SAPO) enhances the stability and performance of reinforcement learning in large language models by adaptively attenuating off-policy updates with a smooth, temperature-controlled gate, leading to improved training stability and performance.
Jiaming Zhang, Shengming Cao, Rui Li +8 authors
SteadyDancer, an Image-to-Video framework, ensures first-frame identity preservation and precise motion control through harmonized conditions, adaptive pose representation, and hierarchical training objectives.
Tian Ye, Song Fei, Lei Zhu
UltraFlux, a Flux-based DiT trained on a 4K dataset, addresses failures in diffusion transformers at 4K resolution through enhanced positional encoding, improved VAE compression, gradient rebalancing, and aesthetic curriculum learning, achieving superior performance compared to existing models.
Yusuf Dalva, Guocheng Gordon Qian, Maya Goldenberg +5 authors
Canvas-to-Image is a unified framework that encodes diverse control signals into a composite canvas image for high-fidelity multimodal image generation, outperforming existing methods in various benchmarks.
Tengxiao Liu, Zifeng Wang, Jin Miao +12 authors
Budget-aware methods improve the scaling of tool-augmented agents by providing continuous budget awareness and adaptive planning, leading to better cost-performance trade-offs.
Yuwei Niu, Weiyang Jin, Jiaqi Liao +7 authors
UniSandbox evaluates Unified Multimodal Models, revealing a gap between understanding and generation, and identifies Chain-of-Thought and self-training as means to bridge this gap.
Jiatao Gu, Ying Shen, Tianrong Chen +6 authors
STARFlow-V, a normalizing flow-based video generator, offers end-to-end learning, robust causal prediction, and high-quality video generation with practical sampling efficiency.
Zhoujie Fu, Xianfang Zeng, Jinghong Lan +9 authors
iMontage repurposes pre-trained video models to generate high-quality, diverse image sets with natural transitions and enhanced dynamics through a unified framework and tailored adaptation strategy.
Gongfan Fang, Xinyin Ma, Xinchao Wang
Video generative models can interpret and execute visual instructions embedded within frames, enhancing controllability in image-to-video generation.
GigaWorld Team, Angen Ye, Boyuan Wang +22 authors
GigaWorld-0 is a unified world model framework that integrates video generation and 3D modeling to produce high-quality, diverse, and physically plausible VLA data, enabling strong real-world performance in embodied AI without real-world training.
Zhenyi Shen, Junru Lu, Lin Gui +4 authors
SSA, a unified training framework for sparse attention in LLMs, achieves state-of-the-art performance by aligning sparse attention with full attention, improving long-context processing and extrapolation.
Yiming Qin, Bomin Wei, Jiaxin Ge +4 authors
Chain-of-Visual-Thought (COVT) enables Vision-Language Models to reason through visual tokens, improving their performance on perceptual tasks by capturing dense visual information.
Bing Wu, Chang Zou, Changlin Li +78 authors
HunyuanVideo 1.5 is a lightweight video generation model with state-of-the-art visual quality and motion coherence, using a DiT architecture with SSTA and an efficient video super-resolution network.
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