14B Video Model Runs Real-Time on a Single GPU

Featured

• $V_1$: Unifying Generation and Self-Verification for Parallel Reasoners score 12
  机构: Berkeley; 入选 HF Daily Papers; HF 热度: 12 upvotes (+3); 有代码实现; 关键词(3): scaling, code generation, reasoning
• CubeComposer: Spatio-Temporal Autoregressive 4K 360° Video Generation from Perspective Video score 11
  入选 HF Daily Papers; HF 热度: 11 upvotes (+3); 有代码实现; 顶会接收: CVPR
• Helios: Real Real-Time Long Video Generation Model score 10
  入选 HF Daily Papers; HF 热度: 132 upvotes (+4); 有代码实现; 关键词(2): quantization, real-time
• T2S-Bench & Structure-of-Thought: Benchmarking and Prompting Comprehensive Text-to-Structure Reasoning score 10
  入选 HF Daily Papers; HF 热度: 105 upvotes (+4); 有代码实现; 关键词(2): fine-tuning, reasoning
• MOOSE-Star: Unlocking Tractable Training for Scientific Discovery by Breaking the Complexity Barrier score 13
  机构: Cornell; 入选 HF Daily Papers; HF 热度: 75 upvotes (+4); 有代码实现; 关键词(2): scaling, reasoning
• ArtHOI: Articulated Human-Object Interaction Synthesis by 4D Reconstruction from Video Priors score 9
  入选 HF Daily Papers; HF 热度: 19 upvotes (+3); 有代码实现; 关键词(1): reasoning
• Phi-4-reasoning-vision-15B Technical Report score 9
  入选 HF Daily Papers; HF 热度: 15 upvotes (+3); 有代码实现; 关键词(2): reasoning, data curation
• InfinityStory: Unlimited Video Generation with World Consistency and Character-Aware Shot Transitions score 8
  机构: Allen Institute; 入选 HF Daily Papers; HF 热度: 7 upvotes (+2)
• Memex(RL): Scaling Long-Horizon LLM Agents via Indexed Experience Memory score 7
  入选 HF Daily Papers; HF 热度: 11 upvotes (+3); 关键词(2): scaling, reasoning
• RIVER: A Real-Time Interaction Benchmark for Video LLMs score 7
  入选 HF Daily Papers; HF 热度: 4 upvotes (+1); 有代码实现; 关键词(1): real-time

Also Worth Noting

• MEM: Multi-Scale Embodied Memory for Vision Language Action Models score 4
  机构: MIT; 关键词(1): embodied
• NuMuon: Nuclear-Norm-Constrained Muon for Compressible LLM Training score 4
  机构: Amazon; 关键词(3): compression, deployment, pretraining
• Parallax to Align Them All: An OmniParallax Attention Mechanism for Distributed Multi-View Image Compression score 4
  关键词(2): compression, coding; 顶会接收: CVPR
• A Rubric-Supervised Critic from Sparse Real-World Outcomes score 4
  机构: Carnegie Mellon; 关键词(3): scaling, coding, data curation
• UniRain: Unified Image Deraining with RAG-based Dataset Distillation and Multi-objective Reweighted Optimization score 4
  关键词(3): distillation, MoE, RAG; 顶会接收: CVPR
• Discriminative Perception via Anchored Description for Reasoning Segmentation score 4
  关键词(1): reasoning; 顶会接收: CVPR
• EgoPoseFormer v2: Accurate Egocentric Human Motion Estimation for AR/VR score 4
  关键词(2): distillation, latency; 顶会接收: CVPR
• When Sensors Fail: Temporal Sequence Models for Robust PPO under Sensor Drift score 4
  关键词(3): PPO, state space, reasoning; 顶会接收: ICLR
• MAGE: Meta-Reinforcement Learning for Language Agents toward Strategic Exploration and Exploitation score 3
  机构: Tsinghua
• QD-PCQA: Quality-Aware Domain Adaptation for Point Cloud Quality Assessment score 3
  顶会接收: CVPR
• TAP: A Token-Adaptive Predictor Framework for Training-Free Diffusion Acceleration score 3
  顶会接收: CVPR
• Fine-grained Image Aesthetic Assessment: Learning Discriminative Scores from Relative Ranks score 3
  顶会接收: CVPR
• BD-Merging: Bias-Aware Dynamic Model Merging with Evidence-Guided Contrastive Learning score 3
  顶会接收: CVPR
• STEM Faculty Perspectives on Generative AI in Higher Education score 3
  顶会接收: AAAI
• NOVA3R: Non-pixel-aligned Visual Transformer for Amodal 3D Reconstruction score 3
  顶会接收: ICLR