Daily Papers of 2025-08-29

  1. rStar2-Agent: Agentic Reasoning Technical Report 97 upvotes, #1 of 2025-08-29
  2. Pref-GRPO: Pairwise Preference Reward-based GRPO for Stable Text-to-Image Reinforcement Learning 85 upvotes, #2 of 2025-08-29
  3. MCP-Bench: Benchmarking Tool-Using LLM Agents with Complex Real-World Tasks via MCP Servers 56 upvotes, #3 of 2025-08-29
  4. USO: Unified Style and Subject-Driven Generation via Disentangled and Reward Learning 55 upvotes, #4 of 2025-08-29
  5. AWorld: Orchestrating the Training Recipe for Agentic AI 37 upvotes, #5 of 2025-08-29
  6. Mixture of Contexts for Long Video Generation 29 upvotes, #6 of 2025-08-29
  7. TCIA: A Task-Centric Instruction Augmentation Method for Instruction Finetuning 21 upvotes, #7 of 2025-08-29
  8. Multi-View 3D Point Tracking 20 upvotes, #8 of 2025-08-29
  9. Turning the Spell Around: Lightweight Alignment Amplification via Rank-One Safety Injection 14 upvotes, #9 of 2025-08-29
  10. OneReward: Unified Mask-Guided Image Generation via Multi-Task Human Preference Learning 11 upvotes, #10 of 2025-08-29
  11. Provable Benefits of In-Tool Learning for Large Language Models 9 upvotes, #11 of 2025-08-29
  12. Persuasion Dynamics in LLMs: Investigating Robustness and Adaptability in Knowledge and Safety with DuET-PD 8 upvotes, #12 of 2025-08-29
  13. CogVLA: Cognition-Aligned Vision-Language-Action Model via Instruction-Driven Routing & Sparsification 8 upvotes, #12 of 2025-08-29
  14. ROSE: Remove Objects with Side Effects in Videos 7 upvotes, #14 of 2025-08-29
  15. FakeParts: a New Family of AI-Generated DeepFakes 7 upvotes, #14 of 2025-08-29
  16. Dress&Dance: Dress up and Dance as You Like It - Technical Preview 5 upvotes, #16 of 2025-08-29
  17. Collaborative Multi-Modal Coding for High-Quality 3D Generation 4 upvotes, #17 of 2025-08-29
  18. OnGoal: Tracking and Visualizing Conversational Goals in Multi-Turn Dialogue with Large Language Models 2 upvotes, #18 of 2025-08-29
  19. Social-MAE: A Transformer-Based Multimodal Autoencoder for Face and Voice 1 upvotes, #19 of 2025-08-29

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.