Daily Papers of 2026-10-27

  1. Trajectory2Gym: Decision-Grounded Compositional Interaction Synthesis for Tool-Use Agents 367 upvotes, #1 of 2026-10-27
  2. Trial by Trace: Benchmarking AI Agent Harnesses for Reliability and Failure Attribution 298 upvotes, #2 of 2026-10-27
  3. StreamVibe: Proactive Video Agents for Evolving User Intent 258 upvotes, #3 of 2026-10-27
  4. DynaFlow: Learning Free-Form Multi-Agent Orchestration 205 upvotes, #4 of 2026-10-27
  5. InteractiveWorld: A Verifiable Human-Centric Environment for Building Better Interactive MLLM Assistants 126 upvotes, #5 of 2026-10-27
  6. Work Agents: Systems, Capabilities, and Optimization 118 upvotes, #6 of 2026-10-27
  7. Can Test-Time Memory Evolution Outperform Test-Time Training for Scientific Discovery? 110 upvotes, #7 of 2026-10-27
  8. Continual Learning Exam: Benchmarking Long-Horizon RSI and Experiential Learning in LLM Agents 60 upvotes, #8 of 2026-10-27
  9. Representation Redistribution in Sparse Autoencoders 59 upvotes, #9 of 2026-10-27
  10. Spatial Audio Generation and Editing with Executable Plans and Self-Distilling Transfusion 57 upvotes, #10 of 2026-10-27
  11. WorldCast: World-Consistent Character Video Generation with Large-Angle Camera Control 55 upvotes, #11 of 2026-10-27
  12. Eat Your Precision: Precision Objective Mitigates Reward Hacking in Rubric-based RL 50 upvotes, #12 of 2026-10-27
  13. AV-FABLE: Audio-Visual Manipulation Forensics in Long-Form Videos 48 upvotes, #13 of 2026-10-27
  14. HiP3R: Hierarchical Pose Graphs for Long-Horizon 3D Reconstruction 43 upvotes, #14 of 2026-10-27
  15. Principia: Test-Time Hypotheses to Principle Construction for Scientific Discovery 42 upvotes, #15 of 2026-10-27
  16. Qwen-Image-PE: Learning Visual Design for Text-to-Image Prompt Enhancement 41 upvotes, #16 of 2026-10-27
  17. World-in-Context: Interactive Video World Modeling with In-Context Conditioning 35 upvotes, #17 of 2026-10-27
  18. S3AC: Stabilized, Selective, and Scalable Actor-Critic for LLM Reasoning 33 upvotes, #18 of 2026-10-27
  19. Evolving to Match Capabilities: Improving Harness Self-Evolution via Skill Adaptation 31 upvotes, #19 of 2026-10-27
  20. RynnVLA-Latent: Scaling Robotic Manipulation Foundation Models with 3D-Grounded Latent Actions 30 upvotes, #20 of 2026-10-27
  21. Δτ-Bench: Pushing Agents Toward a Moving Capability 28 upvotes, #21 of 2026-10-27
  22. Build, Don't Play: Self-Improving LLM Research Agents Develop a Hearthstone Battlegrounds AI That Rivals Top Players 27 upvotes, #22 of 2026-10-27
  23. ProAS: A Progress-Aware Advantage Shaping Framework for Multi-Turn Agentic Reinforcement Learning 26 upvotes, #23 of 2026-10-27
  24. Mycelium: A Meta-Study of Hallucination Detection Across Provenance, Typology, and Granularities 26 upvotes, #23 of 2026-10-27
  25. miniTE: Diagnosing and Bridging the Execution Gap in Native NVFP4 Reinforcement Learning 24 upvotes, #25 of 2026-10-27
  26. More Memory, Better Videos? A Controlled Study of Visual Memory in Long-Video Generation 24 upvotes, #25 of 2026-10-27
  27. Trustworthy Recursive Self-Improvement in the Era of Large Models: Foundation, Risks, Evaluation and Open Challenges 24 upvotes, #25 of 2026-10-27
  28. RoboHarness: A Simple Harness Outperforms VLA and World Action Models 23 upvotes, #28 of 2026-10-27
  29. CausalAlign: Causal Alignment of Video Generation Models for Consistent Ego-Motion 22 upvotes, #29 of 2026-10-27
  30. iPrefBench: A User-Centric Benchmark for Personalized Mobile Use Agents 21 upvotes, #30 of 2026-10-27
  31. EmbodiedLab: Benchmarking Embodied Agents That Act to Know 21 upvotes, #30 of 2026-10-27
  32. How Far Can Coding Agents Go in Visual Generation? 20 upvotes, #32 of 2026-10-27
  33. Extreme Compression of Per-Layer Embeddings via Cross-Layer Latent Vector Quantization 20 upvotes, #32 of 2026-10-27
  34. Thinking-in-Bets at Test Time Reduces Sycophantic Revisions in LLM Agents 18 upvotes, #34 of 2026-10-27
  35. Steering Black-Box Large Language Model Agents with Pitfall Anticipation World Model 16 upvotes, #35 of 2026-10-27
  36. ReGE: Stronger Agents through Experience Synthesis 16 upvotes, #35 of 2026-10-27
  37. Rethinking Video Storage for AI Training: Breaking Cross-Frame Dependencies for Efficient Random Access 15 upvotes, #37 of 2026-10-27
  38. Novel or Not? PatEx-Gym: Agentic Prior-Art Search for Novelty Assessment in Patent Examination 15 upvotes, #37 of 2026-10-27
  39. ParetoRSI-Gym: Can AI Agents Improve Language Models Across Multiple Domains? 15 upvotes, #37 of 2026-10-27
  40. AgentProofBench: Benchmarking Reusable Verifiers for Web Agents 15 upvotes, #37 of 2026-10-27
  41. Nested Sparsity for Scalable Pretraining and Continual Learning 14 upvotes, #41 of 2026-10-27
  42. RoboJev: An Open Vision-Language Judge for Robot Success Detection 13 upvotes, #42 of 2026-10-27
  43. Beyond Benchmarks: Can Small Open-Weight Models Get Real Coding Work Done? 12 upvotes, #43 of 2026-10-27
  44. DialFork: Going Beyond Factuality to Evaluate Procedural Hallucination in Conversational Agents 11 upvotes, #44 of 2026-10-27
  45. Joint or Per-Option? Compact Encoders for Closed-Set Decisions from 17 Million to 1 Billion Parameters 11 upvotes, #44 of 2026-10-27
  46. CRM-Agent-Bench: Benchmarking Tool-Using Agents in Stateful Enterprise Environments 11 upvotes, #44 of 2026-10-27
  47. Skill^{infty}: Recursive Self-Improvement for Agent Skill Optimization 11 upvotes, #44 of 2026-10-27
  48. Rethinking Verification in Agent Harnesses 10 upvotes, #48 of 2026-10-27
  49. Constrained Decoding: Harmful or Beneficial? Revisiting Its Impact on LLM Performance 10 upvotes, #48 of 2026-10-27
  50. OBB-PBD: Extending Parallel Box Decoding to Oriented Bounding Boxes for Referring Expression Grounding in Aerial Imagery 10 upvotes, #48 of 2026-10-27
  51. Trust Mechanisms for Offline Generative AI at the Edge: A Connectivity-Regime Framework and an Empirical Study on a Raspberry Pi 5 9 upvotes, #51 of 2026-10-27
  52. Assessing the CAD Capabilities of General-Purpose AI Models: A Case Study of GPT-6 Astra 8 upvotes, #52 of 2026-10-27
  53. Semantic Concept Steering Breaks the Explanation Drift Loop in Continual Learning 7 upvotes, #53 of 2026-10-27
  54. Intrinsic Goals for Autonomous Agents: Hierarchical Exploration Captures Future Volitional Action in Drosophila 7 upvotes, #53 of 2026-10-27
  55. Gluing Local Contexts into Global Meaning: A Sheaf-Theoretic Decomposition of Transformer Representations 6 upvotes, #55 of 2026-10-27
  56. Ψ-0.6: Adaptive Streams of Thought in a Unified Vision-Language World Model 6 upvotes, #55 of 2026-10-27
  57. EnvLunch: The Environment as a Free Lunch for Learning to Interact 6 upvotes, #55 of 2026-10-27
  58. Humans in Space: Can MLLMs Construct Human-Centric Spatial Understanding via Active Exocentric Exploration? 6 upvotes, #55 of 2026-10-27
  59. Did the Model Use That? Counterfactual Training Boosts Self-Attribution in Language Models 6 upvotes, #55 of 2026-10-27
  60. Learning from Automatic 3D Gaussian Splatting Edits for Online Bayesian Scene Change Detection 5 upvotes, #60 of 2026-10-27
  61. ThinkBit: Search Codewords, Recalibrate Scales, Select by Reasoning 5 upvotes, #60 of 2026-10-27
  62. OpenESM: Energy-Steered Models Are Scalable Foundation Learners 4 upvotes, #62 of 2026-10-27
  63. SO-101 Bench: Accessible Spatial Manipulation 4 upvotes, #62 of 2026-10-27
  64. CellTrack-Bench: Microscopic Cell Tracking Benchmark for Vision-Language Models 2 upvotes, #64 of 2026-10-27

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.