Create a logger that prefixes messages with the current run name.

Create a TensorFlow session with sensible default parameters. Returns: Session.

Reset environments, intermediate scores and durations for new episodes. Args: agent_indices: Tensor containing batch indices starting an ep

Notify the algorithm of ending episodes. Also updates the mean score and length counters used for summaries. Args: agent_indices: Tens

Request actions from the algorithm and apply them to the environments. Increments the lengths of all episodes and increases their scores by the

Combine operations of a phase. Keeps track of the mean score and when to report it. Args: name: Name of the phase used for the score s

Reset the average score and duration, and return them as summary. Returns: Summary string.

Runs the operations for updating training and target ops.

Determine the current phase based on the global step. This ensures continuing the correct phase after restoring checkoints. Args: glob

Returns randomly selected initial positions for agents and trees in relative coordinates. No agent or trees share the same cell in initial po

Yield run numbers in the order they should be considered. The user of this function must check whether a run is already finished or being wor