Daily Papers of 2026-10-05

  1. Does Learning Protein Folding Generalize to Broader Reasoning? 93 upvotes, #1 of 2026-10-05
  2. MotorMind: Scaffolding General Vision Language Models for Zero-Shot Robot Manipulation 78 upvotes, #2 of 2026-10-05
  3. FrameMorrow: Future-guided Frame Selection with Prospective Tokens for Long-Horizon Video Generation 76 upvotes, #3 of 2026-10-05
  4. Scaling Trajectories for Complex Tasks through Recursive Self-Rewrite 73 upvotes, #4 of 2026-10-05
  5. World Action Modeling with Progressive Visual Planning 57 upvotes, #5 of 2026-10-05
  6. Native Action-Prior Learning from Videos for World Action Models 46 upvotes, #6 of 2026-10-05
  7. On-Policy Parameter Update Direction Underlies Generalization in LLM Post-Training 42 upvotes, #7 of 2026-10-05
  8. Pivot-SD: Efficient Self-Distillation for Masked Diffusion Language Models 24 upvotes, #8 of 2026-10-05
  9. SimuVerity: Benchmarking Agents for Engineering-Grade Simulink Model Generation 23 upvotes, #9 of 2026-10-05
  10. PDE-JEPA: Predictive Representation Learning of Latent Dynamics Modeling for Parametric PDEs 22 upvotes, #10 of 2026-10-05
  11. HyperBrowseComp: A Multilingual and Multimodal Stress Test for Web-Browsing Agents 18 upvotes, #11 of 2026-10-05
  12. Science or Slop?: Benchmarking and Mitigating Scientific Slop in AI-Generated Papers 16 upvotes, #12 of 2026-10-05
  13. World Embedding Benchmark 16 upvotes, #12 of 2026-10-05
  14. Source Preference in the Wild: How LLM Agents Favor Items by Source, and How to Reduce It 15 upvotes, #14 of 2026-10-05
  15. LexReward: A Taxonomy-Driven Reward Framework for Legal Language Models 14 upvotes, #15 of 2026-10-05
  16. Science Utopia? Closed-Loop LLM Simulation of Academic Research Ecosystems 11 upvotes, #16 of 2026-10-05
  17. Spatial Memory Intelligence: Endowing World Models with Understanding-Driven Long-Term Memory 11 upvotes, #16 of 2026-10-05
  18. Multilingual GSM-Symbolic: What determines capability transfer across languages? 11 upvotes, #16 of 2026-10-05
  19. DEFINE: Exemplar-Guided Accent Control for Zero-Shot TTS 9 upvotes, #19 of 2026-10-05
  20. VeriHarness: Scaling Agentic Verification for Long-Horizon Tasks 9 upvotes, #19 of 2026-10-05
  21. EditHero: A Benchmark for Long-Horizon Part-Level 3D Editing and Vibe Modeling 9 upvotes, #19 of 2026-10-05
  22. Diptych: Scoped, AI-Interpreted Comparison for Reference Listening in Music Production 8 upvotes, #22 of 2026-10-05
  23. HelixWorld: A Real-time Interactive Audio-Visual World Model 7 upvotes, #23 of 2026-10-05
  24. Looping Beyond Twice: A Scalable Recipe for Looped Mixture-of-Experts 7 upvotes, #23 of 2026-10-05
  25. Efficient Reasoning Training Does Not Always Harm CoT Faithfulness and Monitorability 7 upvotes, #23 of 2026-10-05
  26. ProAR: Learning Prospective Reasoning with Autoregressive Video Models 7 upvotes, #23 of 2026-10-05
  27. MetaRubric: Learning to Reward for Rubric-Based Reinforcement Learning 6 upvotes, #27 of 2026-10-05
  28. Triadic Linear Attention: Three-Dimensional Recurrent States for Long-Context Sequence Modeling 5 upvotes, #28 of 2026-10-05
  29. Language Models that Play Chess and Explain Their Moves 5 upvotes, #28 of 2026-10-05
  30. GTR: Gated Token Recurrence for Efficient Dense Prediction 4 upvotes, #30 of 2026-10-05
  31. Covert Assistance: Helpful LLM Agents Evade Oversight in Multi-Agent Systems 4 upvotes, #30 of 2026-10-05
  32. FrugalEvo: Towards Cost-Aware LLM-Guided Program Evolution 4 upvotes, #30 of 2026-10-05
  33. LVMT: Video Mask Transformer for Long-term Video Segmentation 3 upvotes, #33 of 2026-10-05
  34. Rollout-Marginal Distillation for Long-Horizon Autoregressive Video Generation 3 upvotes, #33 of 2026-10-05
  35. Local Support Learning 3 upvotes, #33 of 2026-10-05
  36. Octrees as an Explicit 3D Language 3 upvotes, #33 of 2026-10-05
  37. Skill2Real: Agentic Skill Learning for Zero-Shot Sim-to-Real Robot Manipulation 3 upvotes, #33 of 2026-10-05
  38. Equal Ranking Quality, Different Decisions: Measuring and Reducing Order Dependence in LLM Scorers 2 upvotes, #38 of 2026-10-05
  39. From Retrieval to Typed Decisions: Calibrated System One Models from Biomedical Sentence Encoders 2 upvotes, #38 of 2026-10-05
  40. Collective Bias Mitigation via Model Routing and Collaboration 2 upvotes, #38 of 2026-10-05
  41. Rethinking Token Reweighting for SFT: Suppress, Reverse, and Extrapolate Learned Features 1 upvotes, #41 of 2026-10-05
  42. WEFT: Scaling Tool-Use Post-Training for General-Purpose Agents 1 upvotes, #41 of 2026-10-05
  43. Can Computation from Earlier Problems Help LLMs Solve New Ones? 1 upvotes, #41 of 2026-10-05
  44. Learning from Runtime Feedback through Failure-Bank Self-Evolution for Vision-Language-Action Models 1 upvotes, #41 of 2026-10-05
  45. Dream4ACT: A Shared Visual Action Interface for Multi-Embodiment Video-Action Modeling 1 upvotes, #41 of 2026-10-05
  46. QuantWM: Temporally Consistent 2-Bit KV Cache Quantization for Video World Models 0 upvotes, #46 of 2026-10-05
  47. Beyond Future Prediction: Denoising as Generative Adaptation for Robot Control 0 upvotes, #46 of 2026-10-05
  48. Receiver-Conditioned Latent Communication gives 94% CacheBack 0 upvotes, #46 of 2026-10-05
  49. Strike a Chord! Modal Kinetic Typography 0 upvotes, #46 of 2026-10-05
  50. From Gradients to Capabilities: Understanding Multi-Teacher On-Policy Distillation 0 upvotes, #46 of 2026-10-05
  51. Latent-MOPD: Latent Multi-Teacher On-Policy Distillation 0 upvotes, #46 of 2026-10-05

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.