Daily Papers of 2023-12-21

  1. StreamDiffusion: A Pipeline-level Solution for Real-time Interactive Generation 75 upvotes, #1 of 2023-12-21
  2. PowerInfer: Fast Large Language Model Serving with a Consumer-grade GPU 46 upvotes, #2 of 2023-12-21
  3. Generative Multimodal Models are In-Context Learners 36 upvotes, #3 of 2023-12-21
  4. Zero-Shot Metric Depth with a Field-of-View Conditioned Diffusion Model 27 upvotes, #4 of 2023-12-21
  5. InstructVideo: Instructing Video Diffusion Models with Human Feedback 18 upvotes, #5 of 2023-12-21
  6. Splatter Image: Ultra-Fast Single-View 3D Reconstruction 15 upvotes, #6 of 2023-12-21
  7. Cached Transformers: Improving Transformers with Differentiable Memory Cache 13 upvotes, #7 of 2023-12-21
  8. MaskINT: Video Editing via Interpolative Non-autoregressive Masked Transformers 10 upvotes, #8 of 2023-12-21
  9. Adaptive Guidance: Training-free Acceleration of Conditional Diffusion Models 9 upvotes, #9 of 2023-12-21
  10. Mini-GPTs: Efficient Large Language Models through Contextual Pruning 9 upvotes, #9 of 2023-12-21
  11. SpecNeRF: Gaussian Directional Encoding for Specular Reflections 7 upvotes, #11 of 2023-12-21
  12. Model-Based Control with Sparse Neural Dynamics 6 upvotes, #12 of 2023-12-21
  13. UniSDF: Unifying Neural Representations for High-Fidelity 3D Reconstruction of Complex Scenes with Reflections 6 upvotes, #12 of 2023-12-21
  14. RadEdit: stress-testing biomedical vision models via diffusion image editing 4 upvotes, #14 of 2023-12-21
  15. Repaint123: Fast and High-quality One Image to 3D Generation with Progressive Controllable 2D Repainting 4 upvotes, #14 of 2023-12-21

Data: hysts-bot-data/daily-papers-stats and the Daily Papers API. Open data: tardellirs/paper-pulse-data. Sister project: Model Pulse, the download history of every model on the Hub.