A Pathmind Professional subscription is required to access advanced settings.
Pathmind's experiment settings enable you to adjust the duration and the depth of policy training.
Upon creating a new experiment, you will be given the option to adjust several settings. These settings only apply to the experiment in question and must be set again whenever you create a new experiment.
Enable Longer Training - This setting enables you to train a policy beyond the 500 training iteration limit. You may train your policy for up to 2,000 training iterations with this setting.
We recommend that you read this article before adjusting the settings below. The default number of nodes and layers are usually sufficient for most use cases. If your policy is not learning well, it's most likely an issue with your action space, observations, or rewards.
Number of Hidden Nodes - Adjust the number of nodes (i.e. width) in the neural network.
Number of Hidden Layers - Adjust the number of layers (i.e. depth) of the neural network.