duo_ai.utils.evaluation
=======================

.. py:module:: duo_ai.utils.evaluation


Functions
---------

.. autoapisummary::

   duo_ai.utils.evaluation.get_test_eval_info


Module Contents
---------------

.. py:function:: get_test_eval_info(env_suite, env_name, test_env, evaluator)

   Retrieve or compute evaluation statistics for a test environment and expert agent.

   If statistics for the given environment are missing, this function evaluates the expert
   agent on the test environment, saves the results, and returns the summary.

   :param env_suite: Name of the environment suite (e.g., 'procgen').
   :type env_suite: str
   :param env_name: Name of the specific environment (e.g., 'coinrun').
   :type env_name: str
   :param test_env: Test environment instance, expected to have 'expert', 'base_env', and 'num_envs' attributes.
   :type test_env: object
   :param evaluator: Evaluator instance with an 'eval' method for running evaluation.
   :type evaluator: object

   :returns: **info** -- Dictionary of evaluation statistics for the test environment and expert agent.
   :rtype: dict

   .. rubric:: Examples

   >>> info = get_test_eval_info('procgen', 'coinrun', test_env, evaluator)
   >>> print(info['reward_mean'])