训练智能体#

• Training using REINFORCE for Mujoco
  • Policy Network
  • Building an agent
  • Plot learning curve
  • References
• Frozenlake benchmark
  • Dependencies
  • Parameters we’ll use
  • The FrozenLake environment
    • Creating the Q-table
    • Running the environment
    • Visualization
      • Map size: $4\times 4$
      • Map size: $7\times 7$
      • Map size: $9\times 9$
      • Map size: $11\times 11$
  • References