Fix warning in quickstart.py (#823)

* Fix warning in quickstart.py caused by not properly setting the render mode in the evaluation environment. * Add workaround for pygame under MacOS * Fix typo in .circleci/config.yml Co-authored-by: Adam Gleave <[email protected]> --------- Co-authored-by: Adam Gleave <[email protected]>
HumanCompatibleAI · Dec 12, 2023 · ab45b47 · ab45b47
1 parent 24a6384
commit ab45b47
Show file tree

Hide file tree

Showing 2 changed files with 14 additions and 2 deletions.
diff --git a/.circleci/config.yml b/.circleci/config.yml
@@ -365,6 +365,10 @@ jobs:
  TESTFILES=$(circleci tests glob "tests/**/test*.py" | circleci tests split --split-by=timings)
  echo "This shard testing: ${TESTFILES}"
  pytest -n auto --junitxml=/tmp/test-reports/junit.xml -vv $TESTFILES
+ environment:
+ # This is needed in newer versions of pygame. See
+ # https://github.com/pygame/pygame/issues/3835#issuecomment-1703717368
+ SDL_VIDEODRIVER: dummy
 
  - save-pytest-cache
  - store-test-output-unix

diff --git a/examples/quickstart.py b/examples/quickstart.py
@@ -71,9 +71,16 @@ def sample_expert_transitions():
  rng=rng,
 )
 
+evaluation_env = make_vec_env(
+ "seals:seals/CartPole-v0",
+ rng=rng,
+ env_make_kwargs={"render_mode": "human"}, # for rendering
+)
+
+print("Evaluating the untrained policy.")
 reward, _ = evaluate_policy(
  bc_trainer.policy, # type: ignore[arg-type]
- env,
+ evaluation_env,
  n_eval_episodes=3,
  render=True, # comment out to speed up
 )
@@ -82,9 +89,10 @@ def sample_expert_transitions():
 print("Training a policy using Behavior Cloning")
 bc_trainer.train(n_epochs=1)
 
+print("Evaluating the trained policy.")
 reward, _ = evaluate_policy(
  bc_trainer.policy, # type: ignore[arg-type]
- env,
+ evaluation_env,
  n_eval_episodes=3,
  render=True, # comment out to speed up
 )