Append Time Step Index as Node Input

Time node is the time information of data. By configuring the information of time step node, the time step index can be used as input for specified nodes.

Here’s an example of a .yaml file to demonstrate the configuration of time step:

metadata:

   graph:
     action:
     - observation
     next_observation:
     - action
     - observation

   columns:
   ...

The above is the original .yaml file. In the modified .yaml file below, we add the keyword step_input as a node property to configure the time step information as input for next_observation. During the model training process, time_step will be automatically added as an input for the action node.

metadata:

   graph:
     action:
     - observation
     next_observation:
     - action
     - observation

   columns:
   ...

   nodes:
     action:
       step_input: True

Important

There is no need to add a time_step node as input in the graph. If the step_input property of a node is configured, the time step node information will be automatically added as input for the node.