Best AI papers explained

515 Episodes

1. Richard Sutton Declares LLMs a Dead End

   Published: 20/10/2025
2. Demystifying Reinforcement Learning in Agentic Reasoning

   Published: 19/10/2025
3. Emergent coordination in multi-agent language models

   Published: 19/10/2025
4. Learning-to-measure: in-context active feature acquisition

   Published: 19/10/2025
5. Andrej Karpathy's insights: AGI, Intelligence, and Evolution

   Published: 19/10/2025
6. Front-Loading Reasoning: The Synergy between Pretraining and Post-Training Data

   Published: 18/10/2025
7. Representation-Based Exploration for Language Models: From Test-Time to Post-Training

   Published: 18/10/2025
8. The attacker moves second: stronger adaptive attacks bypass defenses against LLM jail- Breaks and prompt injections

   Published: 18/10/2025
9. When can in-context learning generalize out of task distribution?

   Published: 16/10/2025
10. The Art of Scaling Reinforcement Learning Compute for LLMs

    Published: 16/10/2025
11. A small number of samples can poison LLMs of any size

    Published: 16/10/2025
12. Dual Goal Representations

    Published: 14/10/2025
13. Welcome to the Era of Experience

    Published: 14/10/2025
14. Value Flows: Flow-Based Distributional Reinforcement Learning

    Published: 14/10/2025
15. Self-Adapting Language Models

    Published: 12/10/2025
16. The Markovian Thinker

    Published: 12/10/2025
17. Moloch’s Bargain: emergent misalignment when LLMs compete for audiences

    Published: 12/10/2025
18. Transformer Predictor Dynamics and Task Diversity

    Published: 11/10/2025
19. Base models know how to reason, thinking models learn when

    Published: 11/10/2025
20. Spectrum tuning: Post-training for distributional coverage and in-context steerability

    Published: 11/10/2025