Daily Papers of 2026-10-20

  1. Scaling Parameter-Efficient Fine-Tuning for Brain Encoding 247 upvotes, #1 of 2026-10-20
  2. EgoXperience: Learning World-Grounded Egocentric Representations from Human Experience 189 upvotes, #2 of 2026-10-20
  3. Holi-World: Building Holistic World Intelligence from Human-Ego Experience 172 upvotes, #3 of 2026-10-20
  4. MemT-1.0: Towards Memory-Native Evolution 146 upvotes, #4 of 2026-10-20
  5. Learning from Fleet Experience: A Collective Learning Paradigm for Robots 138 upvotes, #5 of 2026-10-20
  6. OR-License: Evaluating LLM Agents as Real-World Operations Research Practitioners 128 upvotes, #6 of 2026-10-20
  7. OpenVLHarness: Real World Visual Agentic Tool-Use with Multimodal Memory 121 upvotes, #7 of 2026-10-20
  8. ScaleNav: Scalable Internet Video Pretraining for Language-Conditioned Local Navigation 91 upvotes, #8 of 2026-10-20
  9. SpecMRL: Redesign Matryoshka Representation Learning within Parametric Spectrum 87 upvotes, #9 of 2026-10-20
  10. Self-Evolving Coding Agents via Harness-Policy Optimization 56 upvotes, #10 of 2026-10-20
  11. CoworkEnv: Synthesizing Training Environments for Long-Horizon Cowork Agents 55 upvotes, #11 of 2026-10-20
  12. Interlock: From Local Correctness to Task Completion in Long-Horizon Agents 49 upvotes, #12 of 2026-10-20
  13. MT Forcing: Towards Stable and Consistent Multi-Turn Image Editing with Autoregressive Diffusion 47 upvotes, #13 of 2026-10-20
  14. AnyGoal: On World Model Generalization Beyond Expert Rollouts 46 upvotes, #14 of 2026-10-20
  15. MemPilot: Orchestrating On-Demand Multimodal Memory Curation for LLM Agents 41 upvotes, #15 of 2026-10-20
  16. RoAR-VAE: Improving Latent Diffusability through Robust Prediction 41 upvotes, #15 of 2026-10-20
  17. Kronecker Delta Rule for Long Context Linear Attention 41 upvotes, #15 of 2026-10-20
  18. BlenderWorld: Benchmarking Visual Agents for 3D Creation from Real-World References 40 upvotes, #18 of 2026-10-20
  19. AdaExploreBench: Benchmarking LLM Agents for Proactive Exploration and Test-time Adaptation in Evolving Environments 38 upvotes, #19 of 2026-10-20
  20. InfiniDream: Open-Ended Photorealistic 3D Worlds Through Procedural Generation 35 upvotes, #20 of 2026-10-20
  21. FineSplat: Fine Patches, Local Groups, Detailed Splats 34 upvotes, #21 of 2026-10-20
  22. HarnessPro: Recursive Harness Self-Improvement with Pareto Optimization for Performance-Cost Trade-offs 33 upvotes, #22 of 2026-10-20
  23. Manga2Animate: An Agentic Framework for Faithful Manga Animation with Script and Storyboard Planning 33 upvotes, #22 of 2026-10-20
  24. LIBERO-Industry: Benchmarking Robotic Manipulation under Industrial Process Constraints 30 upvotes, #24 of 2026-10-20
  25. Harnessing Textual Refusal Directions for Multimodal Safety 26 upvotes, #25 of 2026-10-20
  26. AffordanceWAM: Affordance-Aware Joint World-Action Modeling for Robot Manipulation 24 upvotes, #26 of 2026-10-20
  27. Evidence-First Agentic Verification 24 upvotes, #26 of 2026-10-20
  28. Think Visually Across Space and Time: Grounded Reasoning for Video Segmentation 24 upvotes, #26 of 2026-10-20
  29. Where is Physical RSI capability from? An Observational Scaling Law of LLM-as-policy 24 upvotes, #26 of 2026-10-20
  30. Frechet Forcing: Training Long-Horizon Streaming World Models from Short Clips 22 upvotes, #30 of 2026-10-20
  31. RoboValue: A Fine-Grained Sim-and-Real Benchmark for Unified Evaluation of Robotic Value Models 21 upvotes, #31 of 2026-10-20
  32. ECHO: Efficient Causal Self-Recursion Beyond Auxiliary Multi-Token Prediction 19 upvotes, #32 of 2026-10-20
  33. DeOverlay: A Benchmark and Source-Anchored Flow for Video Overlay Removal 19 upvotes, #32 of 2026-10-20
  34. Policy Is Not Capability: Some Safety Benchmarks Measure Which Definition of "Unsafe" You Trained On 18 upvotes, #34 of 2026-10-20
  35. State World Models: Execute the Rules, Generate the Pixels 17 upvotes, #35 of 2026-10-20
  36. Self-Evaluating Recursive Agents 16 upvotes, #36 of 2026-10-20
  37. Empowering Edge Agents: A Systematic Analysis of Edge-Cloud Collaboration in LLM Agent Execution 16 upvotes, #36 of 2026-10-20
  38. Co-Evolving Acting and Forecasting for Language Agents 16 upvotes, #36 of 2026-10-20
  39. ACEdit: Stable Multi-Turn Image Editing via Asymmetric Conditioning 16 upvotes, #36 of 2026-10-20
  40. Factorization-Guided Training and Inference for Diffusion-Drafted Speculative Decoding 16 upvotes, #36 of 2026-10-20
  41. Progress-Aware Distillation for Mitigating Stagnation in Small Language Model Agents 15 upvotes, #41 of 2026-10-20
  42. StopLoss: Teaching Agents to Predict Failure and Stop Early 15 upvotes, #41 of 2026-10-20
  43. UniAVC: Unified Audio-Video Creation via Progressive Training and Asymmetric Scheduling 14 upvotes, #43 of 2026-10-20
  44. Learning to Discriminate Scene Structures Makes Self-Supervised Depth Learning Scalable 13 upvotes, #44 of 2026-10-20
  45. Remembering Has a Price: Learning Memory Worth under Resource Constraints 13 upvotes, #44 of 2026-10-20
  46. Competing Learning in Hybrid Architectures 12 upvotes, #46 of 2026-10-20
  47. TokenGate: Adapting LLMs for Fast and Memory-Efficient Prefill with Learned Token Pruning 11 upvotes, #47 of 2026-10-20
  48. Predicting Privileged Information Effectiveness for On-Policy Self-Distillation 10 upvotes, #48 of 2026-10-20
  49. HandsOn: A Large-Scale Simulation Benchmark for Generalizable Dexterous Manipulation 10 upvotes, #48 of 2026-10-20
  50. A Conditional No-Hair Theorem for Transformers: How Language Models Forget Their Prompt 9 upvotes, #50 of 2026-10-20
  51. World Model Interpretability: From Decoding Representations to Understanding Latent Worlds 9 upvotes, #50 of 2026-10-20
  52. RSIFlow: Adaptive Self-Improvement Workflows with an Evolving Meta-Agent 8 upvotes, #52 of 2026-10-20
  53. N-Gram Memories at Two Bits: Quantizing Lookup Tables in Large Language Models 8 upvotes, #52 of 2026-10-20
  54. HSPR: Recovering Software-Agent Benchmark Scores via Hierarchical Posterior Inference 8 upvotes, #52 of 2026-10-20
  55. Text-to-Image Generative Modeling Without Language Pretraining 8 upvotes, #52 of 2026-10-20
  56. Sparse Feature Policy Unlearning Mitigates State Hallucination in Vision-Language-Action Models 7 upvotes, #56 of 2026-10-20
  57. Heterogeneous Agentic Coevolution 7 upvotes, #56 of 2026-10-20
  58. HERA: Harness-Environment Co-Evolution for Reliable Agentic Abstention 6 upvotes, #58 of 2026-10-20
  59. Training Reasoning Models to Generate Diverse Scientific Ideas 6 upvotes, #58 of 2026-10-20
  60. QD-Rec: Open-Ended Evolution of Recommendations via Quality-Diversity 6 upvotes, #58 of 2026-10-20
  61. ActionRoPE: action as coordinates for spatially consistent control in planar-camera game world models 6 upvotes, #58 of 2026-10-20
  62. Any 3D Scene is Worth 1K Tokens: 3D-Grounded Representation for Scene Generation at Scale 5 upvotes, #62 of 2026-10-20
  63. CyberArena: Benchmarking AI Agents in Live Symmetric Attack-and-Defense Combat 5 upvotes, #62 of 2026-10-20
  64. Harness-Delta Attribution: Dissecting Gains From Harness Evolution 5 upvotes, #62 of 2026-10-20
  65. SPARC: Spoken Profile Authorship Recognition Corpus 5 upvotes, #62 of 2026-10-20
  66. Vision-Language-Delta-Action: Learning Compact Deltas as Actionable Transition Interfaces for Robotic Manipulation 5 upvotes, #62 of 2026-10-20
  67. Competence-Based Evaluation of Logical Reasoning in Large Language Models 4 upvotes, #67 of 2026-10-20
  68. Token-Set Choice Confounds POPE: A Systematic Audit of Yes/No Extraction in VLM Hallucination Evaluation 4 upvotes, #67 of 2026-10-20
  69. Beyond Maliciousness: Static and Runtime Attack-Surface Profiling of Agent Skills 3 upvotes, #69 of 2026-10-20
  70. Measuring Mechanistic Uncertainty in Large Language Models via Causal Robustness 3 upvotes, #69 of 2026-10-20
  71. A Language-Informed Loss for Visual Generation 3 upvotes, #69 of 2026-10-20
  72. Legal Guidance in the Wild: How Users Seek Legal Help in Real-World LLM Conversations 3 upvotes, #69 of 2026-10-20
  73. Vision as Proprioception: Estimating Humanoid Robot Pose from Video 3 upvotes, #69 of 2026-10-20
  74. RealityBench: Simulated Environments to Evaluate a Vision Language Model's Ability to Document Human Physical Work 2 upvotes, #74 of 2026-10-20
  75. First Passage Attention 1 upvotes, #75 of 2026-10-20

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.