Best AI papers explained

442 Episodes

1. Metastable Dynamics of Chain-of-Thought Reasoning: Provable Benefits of Search, RL and Distillation

   Published: 5/28/2025
2. Selective induction heads: how transformers select causal structures in context

   Published: 5/28/2025
3. The Evolution of Statistical Induction Heads: In-Context Learning Markov Chains

   Published: 5/28/2025
4. How Transformers Learn Causal Structure with Gradient Descent

   Published: 5/28/2025
5. Planning anything with rigor: general-purpose zero-shot planning with llm-based formalized programming

   Published: 5/28/2025
6. Automated Design of Agentic Systems

   Published: 5/28/2025
7. What’s the Magic Word? A Control Theory of LLM Prompting

   Published: 5/28/2025
8. BoNBoN Alignment for Large Language Models and the Sweetness of Best-of-n Sampling

   Published: 5/27/2025
9. RL with KL penalties is better viewed as Bayesian inference

   Published: 5/27/2025
10. Asymptotics of Language Model Alignment

    Published: 5/27/2025
11. Qwen 2.5, RL, and Random Rewards

    Published: 5/27/2025
12. Theoretical guarantees on the best-of-n alignment policy

    Published: 5/27/2025
13. Score Matching Enables Causal Discovery of Nonlinear Additive Noise Models

    Published: 5/27/2025
14. Improved Techniques for Training Score-Based Generative Models

    Published: 5/27/2025
15. Your Pre-trained LLM is Secretly an Unsupervised Confidence Calibrator

    Published: 5/27/2025
16. AlphaEvolve: A coding agent for scientific and algorithmic discovery

    Published: 5/27/2025
17. Harnessing the Universal Geometry of Embeddings

    Published: 5/27/2025
18. Goal Inference using Reward-Producing Programs in a Novel Physics Environment

    Published: 5/27/2025
19. Trial-Error-Explain In-Context Learning for Personalized Text Generation

    Published: 5/27/2025
20. Reinforcement Learning for Reasoning in Large Language Models with One Training Example

    Published: 5/27/2025