Windows: basic test failed

[yceny]

I am on Windows, and run the basic test python -um batch_rl.tests.atari_init_test. But it failed.

Here is the traceback:

Running tests under Python 3.9.12: C:\Users\cenyyang\Anaconda3\python.exe
[ RUN ] AtariInitTest.test_atari_init
INFO:tensorflow:Saving replay buffer data to C:\Users\cenyyang\OneDrive\ -\ City\ University\ of\ Hong Kong\batch_rl\replay_logs
I0131 22:32:19.065589 27504 train.py:75] Saving replay buffer data to C:\Users\cenyyang\OneDrive\ -\ City\ University\ of\ Hong Kong\batch_rl\replay_logs
W0131 22:32:19.066590 27504 run_experiment.py:267] DEPRECATION WARNING: Logger is being deprecated. Please switch to CollectorDispatcher!
INFO:tensorflow:time(main.AtariInitTest.test_atari_init): 0.02s
I0131 22:32:19.066590 27504 test_util.py:2462] time(main.AtariInitTest.test_atari_init): 0.02s
[ FAILED ] AtariInitTest.test_atari_init
[ RUN ] AtariInitTest.test_session
[ SKIPPED ] AtariInitTest.test_session

ERROR: test_atari_init (main.AtariInitTest)
AtariInitTest.test_atari_init
Tests that a DQN agent is initialized.

Traceback (most recent call last):
File "C:\Users\cenyyang\OneDrive - City University of Hong Kong\batch_rl\batch_rl\tests\atari_init_test.py", line 49, in test_atari_init
train.main([])
File "C:\Users\cenyyang\OneDrive - City University of Hong Kong\batch_rl\batch_rl\baselines\train.py", line 78, in main
runner = LoggedRunner(FLAGS.base_dir, create_agent_fn)
File "C:\Users\cenyyang\Anaconda3\lib\site-packages\gin\config.py", line 1605, in gin_wrapper
utils.augment_exception_message_and_reraise(e, err_str)
File "C:\Users\cenyyang\Anaconda3\lib\site-packages\gin\utils.py", line 41, in augment_exception_message_and_reraise
raise proxy.with_traceback(exception.traceback) from None
File "C:\Users\cenyyang\Anaconda3\lib\site-packages\gin\config.py", line 1582, in gin_wrapper
return fn(*new_args, **new_kwargs)
File "C:\Users\cenyyang\Anaconda3\lib\site-packages\gin\config.py", line 1605, in gin_wrapper
utils.augment_exception_message_and_reraise(e, err_str)
File "C:\Users\cenyyang\Anaconda3\lib\site-packages\gin\utils.py", line 41, in augment_exception_message_and_reraise
raise proxy.with_traceback(exception.traceback) from None
File "C:\Users\cenyyang\Anaconda3\lib\site-packages\gin\config.py", line 1582, in gin_wrapper
return fn(*new_args, **new_kwargs)
File "C:\Users\cenyyang\Anaconda3\lib\site-packages\dopamine\discrete_domains\run_experiment.py", line 222, in init
self._environment = create_environment_fn()
File "C:\Users\cenyyang\Anaconda3\lib\site-packages\gin\config.py", line 1605, in gin_wrapper
utils.augment_exception_message_and_reraise(e, err_str)
File "C:\Users\cenyyang\Anaconda3\lib\site-packages\gin\utils.py", line 41, in augment_exception_message_and_reraise
raise proxy.with_traceback(exception.traceback) from None
File "C:\Users\cenyyang\Anaconda3\lib\site-packages\gin\config.py", line 1582, in gin_wrapper
return fn(*new_args, **new_kwargs)
File "C:\Users\cenyyang\Anaconda3\lib\site-packages\dopamine\discrete_domains\atari_lib.py", line 96, in create_atari_environment
env = gym.make(full_game_name)
File "C:\Users\cenyyang\Anaconda3\lib\site-packages\gym\envs\registration.py", line 607, in make
_check_version_exists(ns, name, version)
File "C:\Users\cenyyang\Anaconda3\lib\site-packages\gym\envs\registration.py", line 234, in _check_version_exists
_check_name_exists(ns, name)
File "C:\Users\cenyyang\Anaconda3\lib\site-packages\gym\envs\registration.py", line 212, in _check_name_exists
raise error.NameNotFound(
gym.error.NameNotFound: Environment PongNoFrameskip doesn't exist.
In call to configurable 'create_atari_environment' (<function create_atari_environment at 0x000001E26C2BB280>)
In call to configurable 'Runner' (<class 'dopamine.discrete_domains.run_experiment.Runner'>)
In call to configurable 'LoggedRunner' (<class 'batch_rl.baselines.run_experiment.LoggedRunner'>)

Ran 2 tests in 0.027s

FAILED (errors=1, skipped=1)

The error seems to be related to the base_dir?

[agarwl]

Hi @yceny, you can look at the more up-to-date working code in dopamine/labs/offline_rl. Also, since this is offline RL, you'd only need the offline dataset and can use your favorite framework.