Daily Papers of 2025-02-03

  1. s1: Simple test-time scaling 97 upvotes, #1 of 2025-02-03
  2. Reward-Guided Speculative Decoding for Efficient LLM Reasoning 34 upvotes, #2 of 2025-02-03
  3. MatAnyone: Stable Video Matting with Consistent Memory Propagation 28 upvotes, #3 of 2025-02-03
  4. Self-supervised Quantized Representation for Seamlessly Integrating Knowledge Graphs with Large Language Models 23 upvotes, #4 of 2025-02-03
  5. Scalable-Softmax Is Superior for Attention 20 upvotes, #5 of 2025-02-03
  6. PixelWorld: Towards Perceiving Everything as Pixels 14 upvotes, #6 of 2025-02-03
  7. DINO-WM: World Models on Pre-trained Visual Features enable Zero-shot Planning 11 upvotes, #7 of 2025-02-03
  8. Constitutional Classifiers: Defending against Universal Jailbreaks across Thousands of Hours of Red Teaming 8 upvotes, #8 of 2025-02-03
  9. SAeUron: Interpretable Concept Unlearning in Diffusion Models with Sparse Autoencoders 6 upvotes, #9 of 2025-02-03
  10. Zero-Shot Novel View and Depth Synthesis with Multi-View Geometric Diffusion 5 upvotes, #10 of 2025-02-03
  11. The Surprising Agreement Between Convex Optimization Theory and Learning-Rate Scheduling for Large Model Training 5 upvotes, #10 of 2025-02-03
  12. Fast Encoder-Based 3D from Casual Videos via Point Track Processing 4 upvotes, #12 of 2025-02-03
  13. Unraveling the Capabilities of Language Models in News Summarization 3 upvotes, #13 of 2025-02-03
  14. Trading Inference-Time Compute for Adversarial Robustness 3 upvotes, #13 of 2025-02-03
  15. INT: Instance-Specific Negative Mining for Task-Generic Promptable Segmentation 2 upvotes, #15 of 2025-02-03
  16. ChunkKV: Semantic-Preserving KV Cache Compression for Efficient Long-Context LLM Inference 2 upvotes, #15 of 2025-02-03

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.