Daily Papers of 2024-09-27

  1. MaskLLM: Learnable Semi-Structured Sparsity for Large Language Models 43 upvotes, #1 of 2024-09-27
  2. EMOVA: Empowering Language Models to See, Hear and Speak with Vivid Emotions 33 upvotes, #2 of 2024-09-27
  3. LLaVA-3D: A Simple yet Effective Pathway to Empowering LMMs with 3D-awareness 32 upvotes, #3 of 2024-09-27
  4. Instruction Following without Instruction Tuning 25 upvotes, #4 of 2024-09-27
  5. Lotus: Diffusion-based Visual Foundation Model for High-quality Dense Prediction 23 upvotes, #5 of 2024-09-27
  6. Discovering the Gems in Early Layers: Accelerating Long-Context LLMs with 1000x Input Token Reduction 22 upvotes, #6 of 2024-09-27
  7. Pixel-Space Post-Training of Latent Diffusion Models 18 upvotes, #7 of 2024-09-27
  8. The Imperative of Conversation Analysis in the Era of LLMs: A Survey of Tasks, Techniques, and Trends 10 upvotes, #8 of 2024-09-27
  9. Reducing the Footprint of Multi-Vector Retrieval with Minimal Performance Impact via Token Pooling 8 upvotes, #9 of 2024-09-27
  10. Disco4D: Disentangled 4D Human Generation and Animation from a Single Image 8 upvotes, #9 of 2024-09-27
  11. Robot See Robot Do: Imitating Articulated Object Manipulation with Monocular 4D Reconstruction 7 upvotes, #11 of 2024-09-27
  12. Enhancing Structured-Data Retrieval with GraphRAG: Soccer Data Case Study 6 upvotes, #12 of 2024-09-27

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.