All Courses
6 lessons

Reinforcement Learning

Learn from reward signals — the algorithms behind AlphaGo and RLHF

Lessons

  1. 01

    Markov Decision Processes

    States, actions, rewards, transitions — the RL contract.

    Medium
  2. 02

    Q-Learning

    Learn a value function from experience, one update at a time.

    Medium
  3. 03

    Policy Gradients

    Optimize the policy directly via gradient ascent on expected reward.

    Hard
  4. 04

    REINFORCE

    The cleanest policy-gradient algorithm — and its variance problem.

    Hard
  5. 05

    Actor-Critic

    Combine policy learning with a value baseline.

    Hard
  6. 06

    Proximal Policy Optimization

    The stable RL algorithm behind RLHF.

    Hard