Daily Papers of 2026-09-07

  1. Dr. Claw: An AI Scientist Workspace for Vibe Research 180 upvotes, #1 of 2026-09-07
  2. Bilevel Coordinated Reflection: A Game-Theoretic Approach to Multi-Agent LLM Systems 146 upvotes, #2 of 2026-09-07
  3. Iris: Climbing to the Search Frontier 69 upvotes, #3 of 2026-09-07
  4. The Attention Triangle in Audio-Video Models 42 upvotes, #4 of 2026-09-07
  5. Motion-Omni: End-to-End Joint Speech and Full-Body Motion for Spoken Dialogue 42 upvotes, #4 of 2026-09-07
  6. Enoki: Efficient Multi-Level Hallucination Detection 28 upvotes, #6 of 2026-09-07
  7. WorldSculpt: Generating Compositional Worlds from Grounded Videos 27 upvotes, #7 of 2026-09-07
  8. Don't Drop Dropout: Optimizing Layer Sparsity for Efficient LLM Training and Inference 23 upvotes, #8 of 2026-09-07
  9. Ask Before You Optimize: Dynamic Pre-Formulation Clarification for Interactive Optimization 21 upvotes, #9 of 2026-09-07
  10. MaxKernel: Agentic Kernel Generation for TPUs 17 upvotes, #10 of 2026-09-07
  11. Beneath the Surface of Chains-of-Thought: A Mechanistic Interpretation of Reasoning Operations in LLMs 17 upvotes, #10 of 2026-09-07
  12. UniMate: One Unified Model to Animate Diverse Skeletons 17 upvotes, #10 of 2026-09-07
  13. RISE: Recursive Improvement via Self-Extrapolating Policy Distillation 16 upvotes, #13 of 2026-09-07
  14. τ^τ-Bench: An Environment for End-To-End, Realistic Agent Construction 15 upvotes, #14 of 2026-09-07
  15. Unfold The World: Factorize 4D Properties in Reinforcing Spatial Reasoning 11 upvotes, #15 of 2026-09-07
  16. Group Adaptive Clipping Policy Optimization 10 upvotes, #16 of 2026-09-07
  17. When Models Edit Too Much: On the Fidelity of Minimal Code Edits 10 upvotes, #16 of 2026-09-07
  18. AdaptVPR: Route-Aware Hard Positive Generation for Robust Visual Place Recognition 10 upvotes, #16 of 2026-09-07
  19. To See a World in a Living Context: Unified Indoor-Outdoor Urban World Generation 9 upvotes, #19 of 2026-09-07
  20. One Editor, Many Edits: A Unified Training-Free Framework for Diverse Video Editing 8 upvotes, #20 of 2026-09-07
  21. When Quantization Breaks Memory: Recurrent-State Write-Back in Low-Precision Temporal Inference 8 upvotes, #20 of 2026-09-07
  22. Knowing What Not to Answer: Selective Non-Compliance in Vision-Language Models 8 upvotes, #20 of 2026-09-07
  23. Training-Free Speech-Centric Omni Understanding with Frozen VLMs 7 upvotes, #23 of 2026-09-07
  24. Refuse without Refusal: A Structural Analysis of Safety-Tuning Responses for Reducing False Refusals in Language Models 7 upvotes, #23 of 2026-09-07
  25. ShallowStream: Index Shallow then Answer Deep for Streaming Video Understanding 6 upvotes, #25 of 2026-09-07
  26. The 2026 PNPL Competition: Word Classification and Efficient Cross-Subject Generalisation in LibriBrain100 5 upvotes, #26 of 2026-09-07
  27. HarvestBench: Measuring Whether LLM Agents Will Pay to Avoid Killing Animals 5 upvotes, #26 of 2026-09-07
  28. Real-World Knowledge-Guided Change Data Synthesis for Remote Sensing 4 upvotes, #28 of 2026-09-07

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.