Nanjing University · 25 records

dateidtitleauthorsvenue↑
2026-11-202611.26511Direct3D-V3: Latent-Free Diffusion Transformers in the Native Voxel DomainShuang Wu et al.arXiv4
2026-11-092611.12886Reframe: Observation-Aligned Multi-Teacher On-Policy Distillation for Agentic Visual ReasoningHaozhi Yuan et al.arXiv39
2026-11-042611.02568SeMo3D: Unifying 3D Generation, Editing, and Completion via Selective ModelingXiaokun Sun et al.arXiv5
2026-10-262610.27103Omni-Woodpecker: Hallucination Correction via Evidence-Prior Rebalancing for Omni LLMsZijian Liu et al.arXiv33
2026-10-222610.33106VibeGame: Prompt-to-Game Development with AI-Native Engine and Self-Evolving Adversarial Agent TeamWenbo Hu et al.arXiv22
2026-10-202610.26453Remembering Has a Price: Learning Memory Worth under Resource ConstraintsGuannan Lai et al.arXiv6
2026-10-192610.14157OpenMobile-2: Building Versatile Mobile Agents with Scalable Environments and App-Native ToolsChenyang Yan et al.arXiv54
2026-10-132610.12933Learning More by Training Less: Uncertainty-Guided Progressive Contraction for Efficient Reasoning Post-TrainingSuorong Yang et al.arXiv14
2026-10-022610.01762OneStreamer: Unifying Perception, Memory, and Proactive Response in Streaming Video InteractionXiangyu Zeng et al.arXiv77
2026-10-022609.37362Pretrain Once, Route Anywhere: Towards a Foundation Model for LLM RoutingGuannan Lai et al.arXiv9
2026-09-212609.21619Calibrating Teacher--Student Discrepancy for On-Policy DistillationQiangqiang He et al.arXiv10
2026-08-242608.21360OmniAssistBench: Assistant-style Interaction Benchmark for Omni-LLMsXianyun Sun et al.arXiv27
2026-08-052608.02602AURORA-LM: Autoencoding Unified Representation for Continuous-Latent Diffusion Language ModelingJiajun Liang et al.arXiv71
2026-08-032607.27888Not All Tokens Deserve Equal Credit: Counterfactual Sensitivity Credit Reallocation for Long-CoT ReasoningQiangqiang He et al.arXiv7
2026-07-172607.14660VIABench: A Comprehensive Video Benchmark Collected from Blind Individuals for Visual Impairment AssistanceYunfeng Liu et al.arXiv5
2026-07-082607.05511Light-Omni: Reflex over Reasoning in Agentic Video Understanding with Long-Term MemoryChang Nie et al.arXiv21
2026-07-072607.03751Look Before You Leap: Distilling Tree Search into Action Evaluation for Frozen VLA ModelsXinyi Xie et al.arXiv17
2026-06-232606.21649EvoEmbedding: Evolvable Representations for Long-Context Retrieval and Agentic MemoryChang Nie et al.arXiv34
2026-06-162606.16255UniDDT: Unifying Multimodal Understanding and Generation with Decoupled Diffusion TransformerShuai Wang et al.arXiv14
2026-06-152606.14702OmniVideo-100K: A Dataset for Audio-Visual Reasoning through Structured Scripts and Evidence ChainsXinyue Cai et al.arXiv32
2026-06-152606.13657Dense Supervision, Sparse Updates: On the Sparsity and Geometry of On-Policy DistillationGuo Yu et al.arXiv3
2026-06-122606.13289HYDRA-X: Native Unified Multimodal Models with Holistic Visual TokenizersGuozhen Zhang et al.arXiv31
2026-06-052606.03645The Shape of Addition: Geometric Structures of Arithmetic in Large Language ModelsLiuyuan Wen et al.arXiv6
2026-05-192605.16893NGM: A Plug-and-Play Training-Free Memory Module for LLMsYuwen Qu et al.arXiv8
2026-05-132605.06658Relit-LiVE: Relight Video by Jointly Learning Environment VideoWeiqing Xiao et al.arXiv3