Daily Papers of 2026-07-22

  1. ABot-World-0: Infinite Interactive World Rollout on a Single Desktop GPU 303 upvotes, #1 of 2026-07-22
  2. DataFlow-Harness: A Grounded Code-Agent Platform for Constructing Editable LLM Data Pipelines 137 upvotes, #2 of 2026-07-22
  3. Generative World Renderer at the Speed of Play 80 upvotes, #3 of 2026-07-22
  4. Text Template Tokens Are Implicit Semantic Registers in Diffusion Transformers 73 upvotes, #4 of 2026-07-22
  5. Mage-Flow: An Efficient Native-Resolution Foundation Model for Image Generation and Editing 72 upvotes, #5 of 2026-07-22
  6. AlayaWorld: Interactive Long-Horizon World Modeling -- Full Technical Report 57 upvotes, #6 of 2026-07-22
  7. Subliminal Clocks: Latent Time Modelling in Diffusion Language Models 38 upvotes, #7 of 2026-07-22
  8. Stale but Stable: Staleness-Adaptive Trust Regions for Stabilizing Asynchronous Reinforcement Learning 35 upvotes, #8 of 2026-07-22
  9. AgentDebugX: An Open-Source Toolkit for Failure Observability, Attribution, and Recovery in LLM Agents 24 upvotes, #9 of 2026-07-22
  10. SciForma: Structure-Faithful Generation of Scientific Diagrams 22 upvotes, #10 of 2026-07-22
  11. HPD-Parsing: Hierarchical Parallel Document Parsing 11 upvotes, #11 of 2026-07-22
  12. AutoIndex: Learning Representation Programs for Retrieval 10 upvotes, #12 of 2026-07-22
  13. Two-Level Meta-Rubrics for Evaluating Open-Ended Generation: GAMUT, a Benchmark for Factual Completeness 9 upvotes, #13 of 2026-07-22
  14. Masked Visual Actions for Unified World Modeling 9 upvotes, #13 of 2026-07-22
  15. Trajectory-aware Cross-view Geo-localization with Sequential Observations 8 upvotes, #15 of 2026-07-22
  16. ISO: An RLVR-Native Optimization Stack 8 upvotes, #15 of 2026-07-22
  17. H^2SD: Hybrid Hindsight Self-Distillation 7 upvotes, #17 of 2026-07-22
  18. Transcription Policy as a Latent Variable: Activating Controllable Verbatim ASR with Word-Level Timing 6 upvotes, #18 of 2026-07-22
  19. Where Should Optimizer State Live? Tiered State Allocation for Memory-Efficient Mixture-of-Experts Training 6 upvotes, #18 of 2026-07-22
  20. Delineate Anything v2: A Global Foundation Model for Field Delineation 5 upvotes, #20 of 2026-07-22
  21. Reading and Steering Representations of Materials-Science Mechanisms in an Open-Weight Language Model 5 upvotes, #20 of 2026-07-22
  22. EduPanel: A Three-Agent LLM Judge for Teaching Videos -- Reliability, Complementarity, and Human Trust Calibration 4 upvotes, #22 of 2026-07-22
  23. Appearance Pointers -- Multimodal Region Control of Diffusion Transformers 4 upvotes, #22 of 2026-07-22
  24. ConsiSpace: Learning Geometric Consistency Matters for Video Spatial Reasoning 2 upvotes, #24 of 2026-07-22
  25. Computational Humor with Multimodal LLMs: Methods, Datasets, Evaluation, and Challenges 2 upvotes, #24 of 2026-07-22

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.