Daily Papers of 2025-12-24

  1. SemanticGen: Video Generation in Semantic Space 88 upvotes, #1 of 2025-12-24
  2. Step-DeepResearch Technical Report 76 upvotes, #2 of 2025-12-24
  3. Bottom-up Policy Optimization: Your Language Model Policy Secretly Contains Internal Policies 60 upvotes, #3 of 2025-12-24
  4. LongVideoAgent: Multi-Agent Reasoning with Long Videos 52 upvotes, #4 of 2025-12-24
  5. SpatialTree: How Spatial Abilities Branch Out in MLLMs 42 upvotes, #5 of 2025-12-24
  6. Reinforcement Learning for Self-Improving Agent with Skill Library 29 upvotes, #6 of 2025-12-24
  7. MemEvolve: Meta-Evolution of Agent Memory Systems 27 upvotes, #7 of 2025-12-24
  8. SAM Audio: Segment Anything in Audio 21 upvotes, #8 of 2025-12-24
  9. INTELLECT-3: Technical Report 16 upvotes, #9 of 2025-12-24
  10. C2LLM Technical Report: A New Frontier in Code Retrieval via Adaptive Cross-Attention Pooling 14 upvotes, #10 of 2025-12-24
  11. QuantiPhy: A Quantitative Benchmark Evaluating Physical Reasoning Abilities of Vision-Language Models 10 upvotes, #11 of 2025-12-24
  12. Scaling Laws for Code: Every Programming Language Matters 8 upvotes, #12 of 2025-12-24
  13. Memory-T1: Reinforcement Learning for Temporal Reasoning in Multi-session Agents 8 upvotes, #12 of 2025-12-24
  14. FaithLens: Detecting and Explaining Faithfulness Hallucination 8 upvotes, #12 of 2025-12-24
  15. Active Intelligence in Video Avatars via Closed-loop World Modeling 8 upvotes, #12 of 2025-12-24
  16. Simulstream: Open-Source Toolkit for Evaluation and Demonstration of Streaming Speech-to-Text Translation Systems 3 upvotes, #16 of 2025-12-24
  17. Multi-LLM Thematic Analysis with Dual Reliability Metrics: Combining Cohen's Kappa and Semantic Similarity for Qualitative Research Validation 2 upvotes, #17 of 2025-12-24
  18. Toxicity Ahead: Forecasting Conversational Derailment on GitHub 1 upvotes, #18 of 2025-12-24
  19. Learning to Refocus with Video Diffusion Models 1 upvotes, #19 of 2025-12-24

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.