Best AI papers explained

547 Avsnitt

1. When can in-context learning generalize out of task distribution?

   Publicerades: 2025-10-16
2. The Art of Scaling Reinforcement Learning Compute for LLMs

   Publicerades: 2025-10-16
3. A small number of samples can poison LLMs of any size

   Publicerades: 2025-10-16
4. Dual Goal Representations

   Publicerades: 2025-10-14
5. Welcome to the Era of Experience

   Publicerades: 2025-10-14
6. Value Flows: Flow-Based Distributional Reinforcement Learning

   Publicerades: 2025-10-14
7. Self-Adapting Language Models

   Publicerades: 2025-10-12
8. The Markovian Thinker

   Publicerades: 2025-10-12
9. Moloch’s Bargain: emergent misalignment when LLMs compete for audiences

   Publicerades: 2025-10-12
10. Transformer Predictor Dynamics and Task Diversity

    Publicerades: 2025-10-11
11. Base models know how to reason, thinking models learn when

    Publicerades: 2025-10-11
12. Spectrum tuning: Post-training for distributional coverage and in-context steerability

    Publicerades: 2025-10-11
13. Understanding Prompt Tuning and In-Context Learning via Meta-Learning

    Publicerades: 2025-10-11
14. MLPs Learn In-Context on Regression and Classification tasks

    Publicerades: 2025-10-11
15. Is Pre-Training Truly Better than Meta-Learning?

    Publicerades: 2025-10-11
16. Agentic Context Engineering: Evolving Contexts for Self-Improving Language Models

    Publicerades: 2025-10-11
17. Do LLMs Recognize Your Preferences? Evaluating Personalized Preference Following in LLMs

    Publicerades: 2025-10-09
18. Learning dynamics of LLM finetuning

    Publicerades: 2025-10-09
19. Iterative Data Smoothing: Mitigating Reward Overfitting and Overoptimization in RLHF

    Publicerades: 2025-10-09
20. OpenAI Agent Builder and n8n: Orchestrating Reasoning Versus Automating Process

    Publicerades: 2025-10-08