Best AI papers explained

490 Episoder

1. DoubleGen - Debiased Generative Modeling of Counterfactuals

   Publisert: 27.9.2025
2. What Characterizes Effective Reasoning? Revisiting Length, Review, and Structure of CoT

   Publisert: 27.9.2025
3. Compute as Teacher: Turning Inference Compute Into Reference-Free Supervision

   Publisert: 27.9.2025
4. Learning without training: The implicit dynamics of in-context learning

   Publisert: 24.9.2025
5. Does Reinforcement Learning Really Incentivize Reasoning Capacity in LLMs Beyond the Base Model

   Publisert: 24.9.2025
6. Open Problems in Mechanistic Interpretability

   Publisert: 21.9.2025
7. Maestro: Joint Graph & Config Optimization for Reliable AI Agents

   Publisert: 21.9.2025
8. Thought Anchors: Which LLM Reasoning Steps Matter?

   Publisert: 21.9.2025
9. Sample Complexity and Representation Ability of Test-time Scaling Paradigms

   Publisert: 9.9.2025
10. RL's Razor: Why Online RL Forgets Less

    Publisert: 7.9.2025
11. Why Language Models Hallucinate

    Publisert: 6.9.2025
12. ALFA: Aligning LLMs to Ask Good Questions A Case Study in Clinical Reasoning

    Publisert: 6.9.2025
13. Sample Efficient Preference Alignment in LLMs via Active Exploration

    Publisert: 6.9.2025
14. Adventures in Demand Analysis Using AI

    Publisert: 4.9.2025
15. Memento: Fine-tuning LLM Agents without Fine-tuning LLMs

    Publisert: 1.9.2025
16. On the Theoretical Limitations of Embedding-Based Retrieval

    Publisert: 31.8.2025
17. Performance Prediction for Large Systems via Text-to-Text Regression

    Publisert: 30.8.2025
18. Demystifying the Visual Quality Paradox in Multimodal Large Language Models

    Publisert: 30.8.2025
19. Chain-of-Agents: End-to-End Agent Foundation Models via Multi-Agent Distillation and Agentic RL

    Publisert: 30.8.2025
20. Compute-Optimal Scaling for Value-Based Deep RL

    Publisert: 25.8.2025