Best AI papers explained

203 Episoder

1. Causality-Aware Alignment for Large Language Model Debiasing

   Publisert: 29.4.2025
2. Reward Models Evaluate Consistency, Not Causality

   Publisert: 28.4.2025
3. Causal Rewards for Large Language Model Alignment

   Publisert: 28.4.2025
4. Sycophancy to subterfuge: Investigating reward-tampering in large language models

   Publisert: 28.4.2025
5. Bidirectional AI Alignment

   Publisert: 28.4.2025
6. Why Do Multi-Agent LLM Systems Fail?

   Publisert: 27.4.2025
7. LLMs as Greedy Agents: RL Fine-tuning for Decision-Making

   Publisert: 27.4.2025
8. LLM Feedback Loops and the Lock-in Hypothesis

   Publisert: 27.4.2025
9. Representational Alignment Drives Effective Teaching and Learning

   Publisert: 27.4.2025
10. Adaptive Parallel Reasoning with Language Models

    Publisert: 27.4.2025
11. AI: Rewiring the Flow of Ideas and Human Knowledge

    Publisert: 27.4.2025
12. Learning and Equilibrium with Ranking Feedback

    Publisert: 27.4.2025
13. Designing Human-AI Collaboration: A Sufficient-Statistic Approach

    Publisert: 27.4.2025
14. GOAT: Generative Adversarial Training for Human-AI Coordination

    Publisert: 27.4.2025
15. π0.5: Generalization in Robotic Manipulation via Diverse Data

    Publisert: 27.4.2025
16. NoWag: Unified Compression for Large Language Models

    Publisert: 26.4.2025
17. Optimal Tool Calls in Language Model Reasoning

    Publisert: 26.4.2025
18. Data Selection for Empirical Risk Minimization

    Publisert: 26.4.2025
19. LoRe: Low-Rank Reward Modeling for Personalized LLMs

    Publisert: 26.4.2025
20. ParaPO: Reducing Language Model Verbatim Reproduction

    Publisert: 26.4.2025