Daily Papers of 2023-06-27

  1. Kosmos-2: Grounding Multimodal Large Language Models to the World 36 upvotes, #1 of 2023-06-27
  2. MotionGPT: Human Motion as a Foreign Language 28 upvotes, #2 of 2023-06-27
  3. DragDiffusion: Harnessing Diffusion Models for Interactive Point-based Image Editing 21 upvotes, #3 of 2023-06-27
  4. Faster Segment Anything: Towards Lightweight SAM for Mobile Applications 16 upvotes, #4 of 2023-06-27
  5. H_2O: Heavy-Hitter Oracle for Efficient Generative Inference of Large Language Models 14 upvotes, #5 of 2023-06-27
  6. Beyond Scale: the Diversity Coefficient as a Data Quality Metric Demonstrates LLMs are Pre-trained on Formally Diverse Data 12 upvotes, #6 of 2023-06-27
  7. Language models are weak learners 11 upvotes, #7 of 2023-06-27
  8. Thinking Like an Annotator: Generation of Dataset Labeling Instructions 10 upvotes, #8 of 2023-06-27
  9. Supervised Pretraining Can Learn In-Context Reinforcement Learning 9 upvotes, #9 of 2023-06-27
  10. ViNT: A Foundation Model for Visual Navigation 7 upvotes, #10 of 2023-06-27
  11. Zero-shot spatial layout conditioning for text-to-image diffusion models 6 upvotes, #11 of 2023-06-27
  12. DomainStudio: Fine-Tuning Diffusion Models for Domain-Driven Image Generation using Limited Data 6 upvotes, #11 of 2023-06-27
  13. RoboCook: Long-Horizon Elasto-Plastic Object Manipulation with Diverse Tools 6 upvotes, #11 of 2023-06-27
  14. Aligning Large Multi-Modal Model with Robust Instruction Tuning 6 upvotes, #11 of 2023-06-27
  15. Swin-Free: Achieving Better Cross-Window Attention and Efficiency with Size-varying Window 5 upvotes, #15 of 2023-06-27
  16. Restart Sampling for Improving Generative Processes 5 upvotes, #15 of 2023-06-27
  17. RVT: Robotic View Transformer for 3D Object Manipulation 2 upvotes, #17 of 2023-06-27
  18. SEEDS: Emulation of Weather Forecast Ensembles with Diffusion Models 1 upvotes, #18 of 2023-06-27

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.