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。
Pengqi Lu
Deep diffusion transformers face structural instability at extreme depths due to mean-dominated collapse triggered by mean mode screaming, which is mitigated through mean-variance split residuals that maintain stable training while preserving performance.
50 篇论文 · 按点赞排序
Mind Lab, Song Cao, Vic Cao +59 authors
MinT is a managed infrastructure system that enables efficient low-rank adaptation training and serving by keeping base models resident and moving lightweight adapter revisions, scaling across multiple dimensions including large model architectures, reduced storage requirements, and distributed policy management.
Haiwen Diao, Penghao Wu, Hanming Deng +55 authors
Unified vision-language models treat understanding and generation as integrated processes rather than separate tasks, demonstrating strong performance across multiple multimodal capabilities including image synthesis and action reasoning.
Yafu Li, Runzhe Zhan, Haoran Zhang +25 authors
A systematic approach transforms post-trained reasoning models into rigorous olympiad-level solvers through reverse-perplexity curriculum, two-stage reinforcement learning, and test-time scaling, achieving gold-medal performance on mathematical and physics competitions.
Yining Chen, Jihao Zhao, Bo Tang +3 authors
MemPrivacy enables privacy-preserving personalized memory in edge-cloud environments by using type-aware placeholders to protect sensitive data while maintaining semantic integrity for effective memory operations.
Alan Arazi, Eilam Shapira, Shoham Grunblat +8 authors
Multimodal tabular learning benchmarks reveal that task-specific embedding tuning improves performance over frozen pretrained embeddings, particularly when modalities provide complementary predictive signals.
Jingdi Lei, Di Zhang, Junxian Li +7 authors
A lightweight memory mechanism called δ-mem enhances large language models by augmenting a frozen attention backbone with a compact associative memory state that provides low-rank corrections to attention computations.
Bing Zhao, Chenfei Wu, Deqing Li +72 authors
Qwen-Image-2.0 is an advanced image generation model that combines high-fidelity synthesis with precise editing capabilities through a unified framework using Qwen3-VL as condition encoder and Multimodal Diffusion Transformer for joint modeling.
Zhengxi Lu, Zhiyuan Yao, Zhuowen Han +8 authors
SDAR enhances reinforcement learning for multi-turn agent training by integrating self-distillation through a sigmoid gate that selectively strengthens positive token-level guidance while mitigating negative teacher rejections.
Yuchao Gu, Guian Fang, Yuxin Jiang +4 authors
AnyFlow introduces a novel any-step video diffusion distillation framework that improves upon consistency distillation by optimizing full ODE sampling trajectories through flow-map transition learning and backward simulation techniques.
Zhen Fang, Wenxuan Huang, Yu Zeng +8 authors
Flow-OPD addresses limitations in Flow Matching text-to-image models through a two-stage alignment approach combining on-policy distillation and manifold anchor regularization, achieving significant improvements in generation quality and alignment metrics.
Min Zhao, Hongzhou Zhu, Kaiwen Zheng +6 authors
A novel causal consistency distillation method enables efficient frame-wise video generation with reduced latency and improved quality compared to existing chunk-wise approaches.
Haoyi Zhu, Haozhe Liu, Yuyang Zhao +6 authors
SANA-WM is an efficient 2.6B-parameter world model that generates high-fidelity 720p videos with precise camera control, achieving industrial-level quality while significantly reducing computational requirements through hybrid attention, dual-camera branches, two-stage generation, and robust annotation pipelines.
Zhaowei Wang, Lishu Luo, Haodong Duan +9 authors
Long-context continued pre-training enhances vision-language models' ability to handle extended documents while maintaining performance across diverse contexts through strategic data mixture design.
Kaixing Yang, Jiashu Zhu, Xulong Tang +7 authors
MACE-Dance is a music-driven dance video generation framework that combines cascaded Mixture-of-Experts with diffusion models and specialized training strategies to achieve high-quality visual appearance and realistic human motion.
Guijin Son, Seungone Kim, Catherine Arnett +65 authors
A new 439-problem mathematical benchmark created by mathematicians demonstrates significant gaps in advanced reasoning capabilities of leading language models, particularly in identifying ill-posed problems.
Gaotang Li, Bhavana Dalvi Mishra, Zifeng Wang +9 authors
Deep research agents trained using RubricEM framework demonstrate superior performance on long-form research tasks through rubric-guided reinforcement learning with stage-aware planning and reflection-based meta-policy evolution.
Xiyu Ren, Zhaowei Wang, Yiming Du +11 authors
A new benchmark evaluates memory capabilities in vision-language models through multi-session conversations, revealing limitations of both long-context and memory-augmented approaches.
Tara Bogavelli, Gabrielle Gauthier Melançon, Katrina Stankiewicz +10 authors
EVA-Bench presents a comprehensive evaluation framework for voice agents that simulates realistic conversations and measures performance across multiple voice-specific failure modes using novel accuracy and experience metrics.
Siyin Wang, Junhao Shi, Zhaoyang Fu +11 authors
World Action Models unify predictive state modeling with action generation for embodied policy learning, forming a cohesive framework for understanding environment dynamics and action prediction.
Joowon Kim, Seungho Shin, Joonhyung Park +1 authors
A closed-loop framework collaboratively integrates vision-language models with video generation models at step level to improve visual reasoning by enabling real-time failure detection and correction during video generation.
Tong Zheng, Haolin Liu, Chengsong Huang +10 authors
AutoTTS automates test-time scaling strategy discovery by formulating it as controller synthesis over reasoning trajectories and probe signals, achieving improved accuracy-cost tradeoffs with minimal computational overhead.
Taebong Kim, Youngsik Hong, Minsik Kim +4 authors
The Darwin Family framework enables training-free evolutionary merging of large language models through gradient-free weight-space recombination, achieving superior reasoning performance without additional training.
Yun Qu, Qi Wang, Yixiu Mao +11 authors
Group-based policy gradient methods in reinforcement learning with verifiable rewards share a common geometric structure, leading to the development of Listwise Policy Optimization that explicitly handles target projection through divergence minimization for improved training performance and stability.
Jishnu Sethumadhavan Nair, Patrice Bechard, Rishabh Maheshwary +14 authors
Enterprise discovery agents that read system configuration at runtime outperform traditional world models in configurable environments where dynamics change over time.
Minghao Guo, Qingyue Jiao, Zeru Shi +14 authors
MemEye framework evaluates multimodal agent memory by measuring visual evidence granularity and retrieval usage complexity across 8 life-scenario tasks.
Zekai Zhang, Deqing Li, Kuan Cao +27 authors
Qwen-Image-VAE-2.0 is a high-compression Variational Autoencoder suite that improves reconstruction fidelity and diffusability through enhanced architecture, large-scale training, and semantic alignment strategies.
Guankai Li, Jiabin Chen, Yi Xu +2 authors
HyperEyes is a parallel multimodal search agent that enables concurrent entity searches while optimizing inference efficiency through dual-grained reinforcement learning and a specialized benchmark for evaluating both accuracy and efficiency.
Giridhar Ganapavarapu, Dhaval Patel
MCP-Cosmos integrates generative World Models into the Model Context Protocol ecosystem to enhance agent planning and execution through predictive simulation in latent space.
Yufan Deng, Daquan Zhou
HumanNet presents a large-scale human-centric video dataset with rich annotations for embodied intelligence, demonstrating that egocentric human video can effectively replace robot data for training vision-language-action models.
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