run_experiment.TrainRunner

Class TrainRunner

Inherits From: Runner

Object that handles running Atari 2600 experiments.

The TrainRunner differs from the base Runner class in that it does not the
evaluation phase. Checkpointing and logging for the train phase are preserved as
before.

Methods

init

  1. __init__(
  2. *args,
  3. **kwargs
  4. )

Initialize the TrainRunner object in charge of running a full experiment.

Args:

  • base_dir: str, the base directory to host all required
    sub-directories.
  • create_agent_fn: A function that takes as args a Tensorflow session
    and an Atari 2600 Gym environment, and returns an agent.

run_experiment

  1. run_experiment()

Runs a full experiment, spread over multiple iterations.