Daily Papers of 2025-04-18

  1. CLIMB: CLustering-based Iterative Data Mixture Bootstrapping for Language Model Pre-training 87 upvotes, #1 of 2025-04-18
  2. Antidistillation Sampling 59 upvotes, #2 of 2025-04-18
  3. Packing Input Frame Context in Next-Frame Prediction Models for Video Generation 47 upvotes, #3 of 2025-04-18
  4. Generate, but Verify: Reducing Hallucination in Vision-Language Models with Retrospective Resampling 39 upvotes, #4 of 2025-04-18
  5. Perception Encoder: The best visual embeddings are not at the output of the network 31 upvotes, #5 of 2025-04-18
  6. WORLDMEM: Long-term Consistent World Simulation with Memory 30 upvotes, #6 of 2025-04-18
  7. A Strategic Coordination Framework of Small LLMs Matches Large LLMs in Data Synthesis 27 upvotes, #7 of 2025-04-18
  8. ChartQAPro: A More Diverse and Challenging Benchmark for Chart Question Answering 21 upvotes, #8 of 2025-04-18
  9. VistaDPO: Video Hierarchical Spatial-Temporal Direct Preference Optimization for Large Video Models 21 upvotes, #8 of 2025-04-18
  10. DMM: Building a Versatile Image Generation Model via Distillation-Based Model Merging 19 upvotes, #10 of 2025-04-18
  11. NoisyRollout: Reinforcing Visual Reasoning with Data Augmentation 18 upvotes, #11 of 2025-04-18
  12. 70% Size, 100% Accuracy: Lossless LLM Compression for Efficient GPU Inference via Dynamic-Length Float 17 upvotes, #12 of 2025-04-18
  13. InstantCharacter: Personalize Any Characters with a Scalable Diffusion Transformer Framework 17 upvotes, #12 of 2025-04-18
  14. PerceptionLM: Open-Access Data and Models for Detailed Visual Understanding 17 upvotes, #12 of 2025-04-18
  15. Sleep-time Compute: Beyond Inference Scaling at Test-time 14 upvotes, #15 of 2025-04-18
  16. CCMNet: Leveraging Calibrated Color Correction Matrices for Cross-Camera Color Constancy 11 upvotes, #16 of 2025-04-18
  17. Exploring Expert Failures Improves LLM Agent Tuning 11 upvotes, #16 of 2025-04-18
  18. FocusedAD: Character-centric Movie Audio Description 9 upvotes, #18 of 2025-04-18
  19. Complex-Edit: CoT-Like Instruction Generation for Complexity-Controllable Image Editing Benchmark 8 upvotes, #19 of 2025-04-18
  20. Retrieval-Augmented Generation with Conflicting Evidence 7 upvotes, #20 of 2025-04-18
  21. Learning Occlusion-Robust Vision Transformers for Real-Time UAV Tracking 4 upvotes, #21 of 2025-04-18
  22. MetaSynth: Meta-Prompting-Driven Agentic Scaffolds for Diverse Synthetic Data Generation 4 upvotes, #21 of 2025-04-18
  23. Set You Straight: Auto-Steering Denoising Trajectories to Sidestep Unwanted Concepts 4 upvotes, #21 of 2025-04-18

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.