Daily Papers of 2026-03-05

  1. Heterogeneous Agent Collaborative Reinforcement Learning 170 upvotes, #1 of 2026-03-05
  2. Helios: Real Real-Time Long Video Generation Model 159 upvotes, #2 of 2026-03-05
  3. T2S-Bench & Structure-of-Thought: Benchmarking and Prompting Comprehensive Text-to-Structure Reasoning 113 upvotes, #3 of 2026-03-05
  4. Proact-VL: A Proactive VideoLLM for Real-Time AI Companions 31 upvotes, #4 of 2026-03-05
  5. MemSifter: Offloading LLM Memory Retrieval via Outcome-Driven Proxy Reasoning 28 upvotes, #5 of 2026-03-05
  6. ArtHOI: Articulated Human-Object Interaction Synthesis by 4D Reconstruction from Video Priors 21 upvotes, #6 of 2026-03-05
  7. Phi-4-reasoning-vision-15B Technical Report 19 upvotes, #7 of 2026-03-05
  8. Memex(RL): Scaling Long-Horizon LLM Agents via Indexed Experience Memory 19 upvotes, #7 of 2026-03-05
  9. V_1: Unifying Generation and Self-Verification for Parallel Reasoners 14 upvotes, #9 of 2026-03-05
  10. Specificity-aware reinforcement learning for fine-grained open-world classification 13 upvotes, #10 of 2026-03-05
  11. CubeComposer: Spatio-Temporal Autoregressive 4K 360° Video Generation from Perspective Video 13 upvotes, #10 of 2026-03-05
  12. AgilePruner: An Empirical Study of Attention and Diversity for Adaptive Visual Token Pruning in Large Vision-Language Models 11 upvotes, #12 of 2026-03-05
  13. InfinityStory: Unlimited Video Generation with World Consistency and Character-Aware Shot Transitions 8 upvotes, #13 of 2026-03-05
  14. RIVER: A Real-Time Interaction Benchmark for Video LLMs 5 upvotes, #14 of 2026-03-05
  15. MIBURI: Towards Expressive Interactive Gesture Synthesis 4 upvotes, #15 of 2026-03-05
  16. SWE-CI: Evaluating Agent Capabilities in Maintaining Codebases via Continuous Integration 4 upvotes, #15 of 2026-03-05
  17. MUSE: A Run-Centric Platform for Multimodal Unified Safety Evaluation of Large Language Models 3 upvotes, #17 of 2026-03-05
  18. GroupEnsemble: Efficient Uncertainty Estimation for DETR-based Object Detection 1 upvotes, #18 of 2026-03-05
  19. HDINO: A Concise and Efficient Open-Vocabulary Detector 1 upvotes, #18 of 2026-03-05
  20. BeamPERL: Parameter-Efficient RL with Verifiable Rewards Specializes Compact LLMs for Structured Beam Mechanics Reasoning 1 upvotes, #18 of 2026-03-05
  21. EmbodiedSplat: Online Feed-Forward Semantic 3DGS for Open-Vocabulary 3D Scene Understanding 1 upvotes, #18 of 2026-03-05

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.