Definition
Reinforcement learning is the framework where an agent learns to act in an environment by maximizing cumulative reward, with no explicit supervision on individual actions. In the LLM era, it’s how models are shaped after pretraining — from preferences, from rubrics, from outcomes.
Episodes covering this
Worth reading next
Papers we haven't done a deep dive on yet, but would recommend on this topic.
- Is Reinforcement Learning (Not) the Solution to Robust Language Agent Tasks?
- GRPO: Group Relative Policy Optimization for Mathematical Reasoning
- Search-o1: Agentic Search-Enhanced Large Reasoning Models
- Scaling LLM Test-Time Compute Optimally Can Be More Effective than Scaling Model Parameters
- Neural Architecture Search with Reinforcement Learning
- RLVR Is Not RL: Rethinking Credit Assignment in Verifiable Reward Settings
- Reward is Enough
- Reinforcement Learning as Probabilistic Inference: A Survey
- Universal and Transferable Adversarial Attacks on Aligned Language Models
- Constitutional AI: Harmlessness from AI Feedback
- Weak-to-Strong Generalization: Eliciting Strong Capabilities With Weak Supervisors
- RLEF: Grounding Code LLMs in Execution Feedback with Reinforcement Learning
- Let's Verify Step by Step
- Reward Tampering Problems and Solutions in Reinforcement Learning: A Causal Influence Diagram Perspective
- Self-Refine: Iterative Refinement with Self-Feedback
- Reflexion: Language Agents with Verbal Reinforcement Learning
- ECHO: Environment-Conditioned Hierarchical Offline Reinforcement Learning