Daily Papers of 2026-07-27

  1. DataPrep-Bench: Benchmarking LLMs as Training Data Preparators 55 upvotes, #1 of 2026-07-27
  2. Skill Self-Play: Pushing the Frontier of LLM Capability with Co-Evolving Skills 47 upvotes, #2 of 2026-07-27
  3. Molt: A Scalable PyTorch-Native Training Framework for Agentic Reinforcement Learning 32 upvotes, #3 of 2026-07-27
  4. Scaling Native Multimodal Pre-Training From Scratch 28 upvotes, #4 of 2026-07-27
  5. Agentic Context Management: Solving Agent Memory and Cost by Treating Them as Lifecycle and Architecture Problems 25 upvotes, #5 of 2026-07-27
  6. Interactive Training 2: Auditable Control Plane for Live Model Training 20 upvotes, #6 of 2026-07-27
  7. Three-Body Scattering for Generative Modeling 16 upvotes, #7 of 2026-07-27
  8. LAMAR: An Open Language-Aware Multilingual Alignment Reranker 16 upvotes, #7 of 2026-07-27
  9. O-VAD: Industrial Video Anomaly Detection through Object-Centric Tracking and Reasoning 11 upvotes, #9 of 2026-07-27
  10. Multi-Head Latent Control: A Unified Interface for LLM Agent Decision Making 10 upvotes, #10 of 2026-07-27
  11. ID-V2V: Identity-Preserving Video Restylization 10 upvotes, #10 of 2026-07-27
  12. Closing the Loop: Training-Free Revisit Consistency for Autoregressive Generative Rendering 9 upvotes, #12 of 2026-07-27
  13. IDEAgent: Agentic Quality-Diversity Search for Research Idea Generation 9 upvotes, #12 of 2026-07-27
  14. SceneActBench: Can Agents Act on the 3D Scenes They See? 7 upvotes, #14 of 2026-07-27
  15. VisCo: Leveraging Large Language Models as Intrinsic Encoders for Visual Token Compression 6 upvotes, #15 of 2026-07-27
  16. Spectral Prior for Reducing Exposure Bias in Diffusion Models 6 upvotes, #15 of 2026-07-27
  17. Multimodal Speaker Verification as a Threat to Speaker Anonymization 2 upvotes, #17 of 2026-07-27

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.