reinforcement

appears in 12 paper titles

Definition

Names the reinforcement learning setting: an agent takes actions, receives scalar rewards, and improves its policy from that feedback. Two things separate it from supervised learning — the signal is an evaluation rather than a correct answer, and the agent's own actions change the distribution of what it sees next. In LLM work it appears mainly as post-training from human or model-generated preference signals.

Explainers using this term

Paper Walkthrough: It Takes Two to Match — Co-Evolving Both Sides of Retrieval with RLIt Takes Two to Match: Co-Evolving Generative Retriever with Reinforcement Learning
Self-Improving AI — Self-Play, Co-Evolution, and Generated CurriculaMastering Chess and Shogi by Self-Play with a General Reinforcement Learning Algorithm
Designing Distillation Data — Deciding What to Ask the TeacherDeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning