Daily Papers of 2026-09-01

  1. Does On-Policy Distillation Really Distill? From Noisy Teacher to Self-Improvement 144 upvotes, #1 of 2026-09-01
  2. Lucida: Parse, Generate, and Place for Composable Real-to-Sim Scene Modeling 116 upvotes, #2 of 2026-09-01
  3. DreamX-Creator: Democratizing Native Audio-Video Generation at 2K Resolution 99 upvotes, #3 of 2026-09-01
  4. GenFirst: Generation Before Reconstruction for Stable End-to-End Latent Generative Modeling 67 upvotes, #4 of 2026-09-01
  5. On the Design of Qwen3.8-Next Architecture: Evaluation, Efficiency, and Training Stability 54 upvotes, #5 of 2026-09-01
  6. Normalized Low-Rank Adaptation 51 upvotes, #6 of 2026-09-01
  7. PaperGym: Rubric-Centered Evolution for Research-Plan Generation 39 upvotes, #7 of 2026-09-01
  8. LightNav-0: Eliciting VLM Spatial Intelligence for Generalist Embodied Navigation 32 upvotes, #8 of 2026-09-01
  9. SHAPE of Chain-of-Thought in Math Reasoning 31 upvotes, #9 of 2026-09-01
  10. CogEvol: Towards Efficient and Reliable Learning Environment Generation 30 upvotes, #10 of 2026-09-01
  11. Scaling Large Reasoning Models beyond Human Supervision: A Path toward Superintelligence 29 upvotes, #11 of 2026-09-01
  12. Evaluating the Hidden Costs of Personalization in Large Language Models 28 upvotes, #12 of 2026-09-01
  13. Super Library Agent: Joint Generation and Maintenance of Multiple Applications Beyond the Single Codebase 27 upvotes, #13 of 2026-09-01
  14. Learning to Evaluate Before Improving: Automatic Rubric Induction for Automatic Research Agents 22 upvotes, #14 of 2026-09-01
  15. Matrix-Game 3.5: Enhancing Real-Time Streaming Interactive World Models with Patch Memory 18 upvotes, #15 of 2026-09-01
  16. Lies We Can See: Joint Verbal and Non-Verbal Deception by VLM Agents in Embodied Social Interactions 14 upvotes, #16 of 2026-09-01
  17. Keep-or-Drop? Adaptive Tokenizer for Compact Video Representation 13 upvotes, #17 of 2026-09-01
  18. Chain-of-Thought Faithfulness of Reasoning Models Varies with Where and How Preference Cues Are Delivered 13 upvotes, #17 of 2026-09-01
  19. PaperBanana-Interact: Scientific Diagram Refinement with Multi-Turn Human Feedback 12 upvotes, #19 of 2026-09-01
  20. Verification-Aware Training for Speculative Decoding 10 upvotes, #20 of 2026-09-01
  21. Scaffolding Foundation Models into Physical-World Agents Pushes the Frontier of Long-Horizon Navigation 10 upvotes, #20 of 2026-09-01
  22. Weaving Visual Narratives: Agentic Image Bundle Composition Beyond Atomic Visual Matching 8 upvotes, #22 of 2026-09-01
  23. CAST: Critique-Aware Supervision for Training Reliable Long-Horizon Tool-Calling Agents 8 upvotes, #22 of 2026-09-01
  24. WebWorld: The Browser as a World Model for Self-Improving Web Code 8 upvotes, #22 of 2026-09-01
  25. MNIST-PRO: MNIST is Back as a Partially Observable World for AI Agents 8 upvotes, #22 of 2026-09-01
  26. SafeAtlas-VL: Beyond Binary Multimodal Safety with Large-Scale Data and Guard Models 7 upvotes, #26 of 2026-09-01
  27. DICS: Exploring Data Intrinsic Consistency for Visual Instruction Selection 7 upvotes, #26 of 2026-09-01
  28. CoVA-SFT: A Large-Scale Dataset for Chain of Visual Abstractions 6 upvotes, #28 of 2026-09-01
  29. Cross-lingual Functional Vectors for Emotion Detection in Large Language Models 6 upvotes, #28 of 2026-09-01
  30. RECAP-Forcing: Retaining Content Appearances for Long Video Generation 5 upvotes, #30 of 2026-09-01
  31. EvoGenUI-Bench: Evaluating LLMs as Multi-Turn Generative UI Assistants 5 upvotes, #30 of 2026-09-01
  32. ContextBias: Controlled Evaluation of Bias Persistence Under Context Shift in Text-to-Image Models 5 upvotes, #30 of 2026-09-01
  33. SpanCalib-VLM: Calibrated Hallucination Span Detection in Vision-Language Models 5 upvotes, #30 of 2026-09-01
  34. Chat-Edit-3D++: Interactive 3D and 4D Scene Editing via Large Language Models 4 upvotes, #34 of 2026-09-01
  35. Dynamic Important Example Mining for Reinforcement Finetuning 4 upvotes, #34 of 2026-09-01
  36. MMMMM: A Unified Taxonomy for Investigating the Mechanisms of Multilingual MultiModal Misinformation 4 upvotes, #34 of 2026-09-01
  37. BLARM: Animating 3D Objects from Video via Blending Latent Rigid Motion Primitives 4 upvotes, #34 of 2026-09-01
  38. The Safeguard Worked. Is the LLM System Safer? 4 upvotes, #34 of 2026-09-01
  39. Uncertainty-Aware End-to-End AI Weather Forecasting: Disentangling Observation and Model Contributions 2 upvotes, #39 of 2026-09-01

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.