Fixed Q-targets

You are preparing to train your Lunar Lander with fixed Q-targets. As a prerequisite, you need to instantiate both the online network (which chooses the action) and the target network (used for TD-target calculation).

You also need to implement an update_target_network function which you can use at each training step. The target network is not updated by gradient descent; instead, update_target_network nudges its weights towards the Q-network by a small amount, ensuring that it remains quite stable over time.

Note that, for this exercise only, you use a very small network so we can easily print and inspect its state dictionary. It has only one hidden layer of size two; its action space and state space are also of dimension 2.

The function print_state_dict() is available in your environment to print the state dict.

This exercise is part of the course

Deep Reinforcement Learning in Python

Exercise instructions

Obtain the .state_dict() for both the target and online networks.
Update the state dict for the target network by taking the weighted average between the parameters of the online network and of the target network, using tau as weight for the online network.
Load the updated state dict back onto the target network.

Hands-on interactive exercise

Have a go at this exercise by completing this sample code.

def update_target_network(target_network, online_network, tau):
    # Obtain the state dicts for both networks
    target_net_state_dict = ____
    online_net_state_dict = ____
    for key in online_net_state_dict:
        # Calculate the updated state dict for the target network
        target_net_state_dict[key] = (online_net_state_dict[____] * ____ + target_net_state_dict[____] * ____)
        # Load the updated state dict into the target network
        target_network.____
    return None
  
print("online network weights:", print_state_dict(online_network))
print("target network weights (pre-update):", print_state_dict(target_network))
update_target_network(target_network, online_network, .001)
print("target network weights (post-update):", print_state_dict(target_network))

Edit and Run Code

This exercise is part of the course

Deep Reinforcement Learning in Python

AdvancedSkill Level

4.8+

Start Course for Free

Discover how deep reinforcement learning improves upon traditional Reinforcement Learning while studying and implementing your first Deep Q Learning algorithm.

Exercise 1: Introduction to deep reinforcement learning Exercise 2: Environment and neural network setup Exercise 3: DRL training loop Exercise 4: Introduction to deep Q learning Exercise 5: Deep learning and DQN Exercise 6: The Q-Network architecture Exercise 7: Instantiating the Q-Network Exercise 8: The barebone DQN algorithm Exercise 9: Barebone DQN action selection Exercise 10: Barebone DQN loss function Exercise 11: Training the barebone DQN

Dive into Deep Q-learning by implementing the original DQN algorithm, featuring Experience Replay, epsilon-greediness and fixed Q-targets. Beyond DQN, you will then explore two fascinating extensions that improve the performance and stability of Deep Q-learning: Double DQN and Prioritized Experience Replay.

Exercise 1: DQN with experience replay Exercise 2: The double-ended queue Exercise 3: Experience replay buffer Exercise 4: DQN with experience replay Exercise 5: The complete DQN algorithm Exercise 6: Epsilon-greediness Exercise 7: Fixed Q-targets

Current Exercise

Exercise 8: Implementing the complete DQN algorithm Exercise 9: Double DQN Exercise 10: Online network and target network in DDQN Exercise 11: Training the double DQN Exercise 12: Prioritized experience replay Exercise 13: Prioritized experience replay buffer Exercise 14: Sampling from the PER buffer Exercise 15: DQN with prioritized experience replay

Learn about the foundational concepts of policy gradient methods found in DRL. You will begin with the policy gradient theorem, which forms the basis for these methods. Then, you will implement the REINFORCE algorithm, a powerful approach to learning policies. The chapter will then guide you through Actor-Critic methods, focusing on the Advantage Actor-Critic (A2C) algorithm, which combines the strengths of both policy gradient and value-based methods to enhance learning efficiency and stability.

Exercise 1: Introduction to policy gradient Exercise 2: The policy network architecture Exercise 3: Working with discrete distributions Exercise 4: Policy gradient and REINFORCE Exercise 5: Action selection in REINFORCE Exercise 6: Training the REINFORCE algorithm Exercise 7: Advantage Actor Critic Exercise 8: Critic network Exercise 9: Actor Critic loss calculations Exercise 10: Training the A2C algorithm

Explore Proximal Policy Optimization (PPO) for robust DRL performance. Next, you will examine using an entropy bonus in PPO, which encourages exploration by preventing premature convergence to deterministic policies. You'll also learn about batch updates in policy gradient methods. Finally, you will learn about hyperparameter optimization with Optuna, a powerful tool for optimizing performance in your DRL models.

Exercise 1: Proximal policy optimization Exercise 2: The clipped probability ratio Exercise 3: The clipped surrogate objective function Exercise 4: Entropy bonus and PPO Exercise 5: Entropy playground Exercise 6: Training the PPO algorithm Exercise 7: Batch updates in policy gradient Exercise 8: Minibatch and DRL Exercise 9: A2C with batch updates Exercise 10: Hyperparameter optimization with Optuna Exercise 11: Hyperparameter or not?Exercise 12: Hands-on with Optuna Exercise 13: Congratulations!