Daily Papers of 2026-02-20

  1. Unified Latents (UL): How to train your latents 53 upvotes, #1 of 2026-02-20
  2. Mobile-Agent-v3.5: Multi-platform Fundamental GUI Agents 46 upvotes, #2 of 2026-02-20
  3. SpargeAttention2: Trainable Sparse Attention via Hybrid Top-k+Top-p Masking and Distillation Fine-Tuning 43 upvotes, #3 of 2026-02-20
  4. Frontier AI Risk Management Framework in Practice: A Risk Analysis Technical Report v1.5 28 upvotes, #4 of 2026-02-20
  5. Arcee Trinity Large Technical Report 17 upvotes, #5 of 2026-02-20
  6. Discovering Multiagent Learning Algorithms with Large Language Models 16 upvotes, #6 of 2026-02-20
  7. Calibrate-Then-Act: Cost-Aware Exploration in LLM Agents 14 upvotes, #7 of 2026-02-20
  8. Computer-Using World Model 14 upvotes, #7 of 2026-02-20
  9. "What Are You Doing?": Effects of Intermediate Feedback from Agentic LLM In-Car Assistants During Multi-Step Processing 13 upvotes, #9 of 2026-02-20
  10. DDiT: Dynamic Patch Scheduling for Efficient Diffusion Transformers 11 upvotes, #10 of 2026-02-20
  11. TactAlign: Human-to-Robot Policy Transfer via Tactile Alignment 10 upvotes, #11 of 2026-02-20
  12. ArXiv-to-Model: A Practical Study of Scientific LM Training 7 upvotes, #12 of 2026-02-20
  13. 2Mamba2Furious: Linear in Complexity, Competitive in Accuracy 7 upvotes, #12 of 2026-02-20
  14. World Models for Policy Refinement in StarCraft II 6 upvotes, #14 of 2026-02-20
  15. On the Mechanism and Dynamics of Modular Addition: Fourier Features, Lottery Ticket, and Grokking 6 upvotes, #14 of 2026-02-20
  16. FRAPPE: Infusing World Modeling into Generalist Policies via Multiple Future Representation Alignment 5 upvotes, #16 of 2026-02-20
  17. Hardware Co-Design Scaling Laws via Roofline Modelling for On-Device LLMs 3 upvotes, #17 of 2026-02-20
  18. CrispEdit: Low-Curvature Projections for Scalable Non-Destructive LLM Editing 3 upvotes, #17 of 2026-02-20
  19. NESSiE: The Necessary Safety Benchmark -- Identifying Errors that should not Exist 3 upvotes, #17 of 2026-02-20
  20. Modeling Distinct Human Interaction in Web Agents 3 upvotes, #17 of 2026-02-20
  21. References Improve LLM Alignment in Non-Verifiable Domains 2 upvotes, #21 of 2026-02-20
  22. NeST: Neuron Selective Tuning for LLM Safety 1 upvotes, #22 of 2026-02-20
  23. StereoAdapter-2: Globally Structure-Consistent Underwater Stereo Depth Estimation 0 upvotes, #23 of 2026-02-20

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.