Daily Papers of 2025-09-19

  1. ScaleCUA: Scaling Open-Source Computer Use Agents with Cross-Platform Data 101 upvotes, #1 of 2025-09-19
  2. FlowRL: Matching Reward Distributions for LLM Reasoning 100 upvotes, #2 of 2025-09-19
  3. Reasoning over Boundaries: Enhancing Specification Alignment via Test-time Delibration 50 upvotes, #3 of 2025-09-19
  4. Evolving Language Models without Labels: Majority Drives Selection, Novelty Promotes Variation 32 upvotes, #4 of 2025-09-19
  5. AToken: A Unified Tokenizer for Vision 30 upvotes, #5 of 2025-09-19
  6. WorldForge: Unlocking Emergent 3D/4D Generation in Video Diffusion Model via Training-Free Guidance 30 upvotes, #5 of 2025-09-19
  7. FinSearchComp: Towards a Realistic, Expert-Level Evaluation of Financial Search and Reasoning 29 upvotes, #7 of 2025-09-19
  8. Understand Before You Generate: Self-Guided Training for Autoregressive Image Generation 27 upvotes, #8 of 2025-09-19
  9. RynnVLA-001: Using Human Demonstrations to Improve Robot Manipulation 20 upvotes, #9 of 2025-09-19
  10. MultiEdit: Advancing Instruction-based Image Editing on Diverse and Challenging Tasks 11 upvotes, #10 of 2025-09-19
  11. Apertus: Democratizing Open and Compliant LLMs for Global Language Environments 9 upvotes, #11 of 2025-09-19
  12. Agentic Software Engineering: Foundational Pillars and a Research Roadmap 7 upvotes, #12 of 2025-09-19
  13. RecoWorld: Building Simulated Environments for Agentic Recommender Systems 6 upvotes, #13 of 2025-09-19
  14. Can Multimodal LLMs See Materials Clearly? A Multimodal Benchmark on Materials Characterization 5 upvotes, #14 of 2025-09-19
  15. Unleashing the Potential of Multimodal LLMs for Zero-Shot Spatio-Temporal Video Grounding 5 upvotes, #14 of 2025-09-19
  16. Developer-LLM Conversations: An Empirical Study of Interactions and Generated Code Quality 4 upvotes, #16 of 2025-09-19
  17. Mind the Gap: A Closer Look at Tokenization for Multiple-Choice Question Answering with LLMs 4 upvotes, #16 of 2025-09-19
  18. EdiVal-Agent: An Object-Centric Framework for Automated, Scalable, Fine-Grained Evaluation of Multi-Turn Editing 3 upvotes, #18 of 2025-09-19
  19. EchoVLM: Dynamic Mixture-of-Experts Vision-Language Model for Universal Ultrasound Intelligence 3 upvotes, #18 of 2025-09-19
  20. FSG-Net: Frequency-Spatial Synergistic Gated Network for High-Resolution Remote Sensing Change Detection 0 upvotes, #20 of 2025-09-19

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.