July 2018
Intermediate to advanced
254 pages
8h 12m
English
The following is a list of hyperparameters that our deep Q-learner uses, with a brief description of what they are and the types of values they take:
| Hyperparameter | Brief description | Value type |
| max_num_episodes | Maximum number of episodes to run the agent. | Integer (for example, 100,000) |
| replay_memory_capacity | Total capacity of the experience memory capacity. | Integer or exponential notation (for example, 1e6) |
| replay_batch_size | Number of transitions used in a (mini) batch to update the Q-function in each update iteration during experience replay. | Integer (for example, 2,000) |
| use_target_network | Whether a target Q-network is to be used or not. | Boolean (true/false) |
| target_network_update_freq | The number ... |
Read now
Unlock full access