550 Epizód

  1. Front-Loading Reasoning: The Synergy between Pretraining and Post-Training Data

    Közzétéve: 2025. 10. 18.
  2. Representation-Based Exploration for Language Models: From Test-Time to Post-Training

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

    Közzétéve: 2025. 10. 18.
  4. When can in-context learning generalize out of task distribution?

    Közzétéve: 2025. 10. 16.
  5. The Art of Scaling Reinforcement Learning Compute for LLMs

    Közzétéve: 2025. 10. 16.
  6. A small number of samples can poison LLMs of any size

    Közzétéve: 2025. 10. 16.
  7. Dual Goal Representations

    Közzétéve: 2025. 10. 14.
  8. Welcome to the Era of Experience

    Közzétéve: 2025. 10. 14.
  9. Value Flows: Flow-Based Distributional Reinforcement Learning

    Közzétéve: 2025. 10. 14.
  10. Self-Adapting Language Models

    Közzétéve: 2025. 10. 12.
  11. The Markovian Thinker

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

    Közzétéve: 2025. 10. 12.
  13. Transformer Predictor Dynamics and Task Diversity

    Közzétéve: 2025. 10. 11.
  14. Base models know how to reason, thinking models learn when

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

    Közzétéve: 2025. 10. 11.
  16. Understanding Prompt Tuning and In-Context Learning via Meta-Learning

    Közzétéve: 2025. 10. 11.
  17. MLPs Learn In-Context on Regression and Classification tasks

    Közzétéve: 2025. 10. 11.
  18. Is Pre-Training Truly Better than Meta-Learning?

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

    Közzétéve: 2025. 10. 11.
  20. Do LLMs Recognize Your Preferences? Evaluating Personalized Preference Following in LLMs

    Közzétéve: 2025. 10. 09.

4 / 28

Cut through the noise. We curate and break down the most important AI papers so you don’t have to.

Visit the podcast's native language site