utils.summarize_data

  1. utils.summarize_data(
  2.     data,
  3.     summary_keys
  4. )

Processes log data into a per-iteration summary.

Args:

  • data: Dictionary loaded by load_statistics describing the data.
    This dictionary has keys iteration_0, iteration_1, … describing
    per-iteration data.
  • summary_keys: List of per-iteration data to be summarized.

Example: data = load_statistics(…) get_iteration_summmary(data,
[‘train_episode_returns’, ‘eval_episode_returns’])

Returns:

A dictionary mapping each key in returns_keys to a per-iteration summary.