[Bug Report] AttributeError: module 'gym.envs.atari' has no attribute 'AtariEnv'

[htang6]

After installing gym and gym[atari], the breakout doesn't work ` import gym env = gym.make("ALE/Breakout-v5") observation = env.reset() for _ in range(1000): env.render() action = env.action_space.sample() # your agent here (this takes random actions) observation, reward, done, info = env.step(action)

if done: observation = env.reset() env.close()

Traceback (most recent call last): File "/Users/haocheng/playground/PolicyGradient/main.py", line 2, in env = gym.make("ALE/Breakout-v5") File "/Users/haocheng/opt/anaconda3/envs/ML/lib/python3.7/site-packages/gym/envs/registration.py", line 676, in make return registry.make(id, kwargs) File "/Users/haocheng/opt/anaconda3/envs/ML/lib/python3.7/site-packages/gym/envs/registration.py", line 520, in make return spec.make(kwargs) File "/Users/haocheng/opt/anaconda3/envs/ML/lib/python3.7/site-packages/gym/envs/registration.py", line 139, in make cls = load(self.entry_point) File "/Users/haocheng/opt/anaconda3/envs/ML/lib/python3.7/site-packages/gym/envs/registration.py", line 56, in load fn = getattr(mod, attr_name) AttributeError: module 'gym.envs.atari' has no attribute 'AtariEnv'

[pseudo-rnd-thoughts]

Try installing ale-py as v5 uses a different backend module (ale-py) compared to v0 and v4 with atari-lib

pip install ale-py

[pseudo-rnd-thoughts]

@JesseFarebro Could you explain how atari is added to gym.envs.atari as isn't in the git repo https://github.com/openai/gym/tree/master/gym/envs

Strangely, I have just installed gym to a new repository and am running into the same issue however last week as I was able to submit issue #2624 which had a different issue pip install gym[all]

import gym
gym.make('ALE/Breakout-v5')

>>> ModuleNotFoundError: No module named 'gym.envs.atari'

[JesseFarebro]

We are using a somewhat obscure packaging feature in Python called namespace packages. The gym.envs.atari module is in the ALE repository: https://github.com/mgbellemare/Arcade-Learning-Environment/blob/master/src/gym/envs/atari/environment.py

On install PyPi will place gym.envs.atari in the Gym module. When you uninstall ale-py it will remove that module from Gym. One of the issues people face with namespace packages is that they will install Gym in editable mode, i.e., pip install -e. Namespace packages just don't work when you install the root package in editable mode. The Python devs have been discussing standardizing editable packages which will hopefully fix these issues with namespace packages. You can read more here: https://discuss.python.org/t/standardising-editable-mode-installs-runtime-layout-not-hooks/4098

If you are experiencing any issues outside of editable mode feel free to post more info about your environment, e.g., Python version, distribution (pyenv, conda, etc.), OS, etc. You should also always try a fresh pip uninstall gym ale-py && pip install gym[atari]. If this still doesn't work can you post the result of:

ls -l $(python -c 'import site; print(site.getsitepackages()[0])')/gym/envs/{atari,}

[pseudo-rnd-thoughts]

Thanks, that explains the issue, I was using gym in editable mode

[sparisi]

I was having the same issue, and reinstalling ale-py worked.

pip install gym[atari]
pip uninstall ale-py
pip install ale-py

Hope this helps others.

[ZhihaoXu]

I have the same issue and also tried pip uninstall gym ale-py && pip install gym[atari] but does not work. I am running the code on Colab, the output of ls -l $(python -c 'import site; print(site.getsitepackages()[0])')/gym/envs/{atari,} is

/usr/local/lib/python3.7/dist-packages/gym/envs/:
total 80
drwxr-sr-x 3 root staff  4096 Apr 17 06:31 atari
drwxr-sr-x 3 root staff  4096 Apr 17 06:26 box2d
drwxr-sr-x 4 root staff  4096 Apr 17 06:26 classic_control
-rw-r--r-- 1 root staff  5130 Apr 17 06:26 __init__.py
drwxr-sr-x 4 root staff  4096 Apr 17 06:26 mujoco
drwxr-sr-x 2 root staff  4096 Apr 17 06:26 __pycache__
-rw-r--r-- 1 root staff 27364 Apr 17 06:26 registration.py
-rw-r--r-- 1 root staff 20429 Apr 17 06:26 __relocated__.py
drwxr-sr-x 5 root staff  4096 Apr 17 06:26 toy_text

/usr/local/lib/python3.7/dist-packages/gym/envs/atari:
total 24
-rw-r--r-- 1 root staff 14850 Apr 17 06:31 environment.py
-rw-r--r-- 1 root staff    48 Apr 17 06:31 __init__.py
drwxr-sr-x 2 root staff  4096 Apr 17 06:31 __pycache__

[JesseFarebro]

@ZhihaoXu what does your local directory look like? Do you have agym module in your current working directory?

[ZhihaoXu]

@JesseFarebro Thanks for your reply. I think my problem is related to the setup of the baselines package. After reinstalling the baselines, the problem is solved.

[lebrice]

Still somehow doesn't work on my end.

[JesseFarebro]

@lebrice can you give me more information, e.g., what's the output of:

ls -l $(python -c 'import site; print(site.getsitepackages()[0])')/gym/envs/{atari,}

• What's the directory tree look like in your current working directory? Feel free to post it if you can.
• How did you install Gym? make sure it's not in editable mode.
• Make sure there's no file / folder called Gym in your project.
• Make sure Gym isn't a git submodule.
• Make sure you're not working out of the Gym project directory, i.e., don't git clone https://github.com/openao/gym && cd gym and work out of that directory.

[lebrice]

Nevermind @JesseFarebro , At first I believed there was an older version of gym and/or ale-py in my pip cache. However it turns out I was inside the gym repository folder (not installed in editable mode).

Here is my "test" of sorts to check if the installation process works.

$ conda create -n temp-atari python=3.9
...
$ conda activate temp-atari
$ pip install gym[atari,accept-rom-license]
...
$ python -c "import gym; print(gym.make('Breakout-v0').reset().shape)"
/<some warning>
A.L.E: Arcade Learning Environment (version 0.7.5+db37282)
[Powered by Stella]
(210, 160, 3)

All is good! Still, I wish it was possible to use the atari envs inside the gym repo! :thinking:

[2291924061]

runfile('C:/Users/liao/Desktop/CNNPong/untitled2.py', wdir='C:/Users/liao/Desktop/CNNPong') Traceback (most recent call last):

File "D:\Anaconda3\envs\tfgpu\lib\site-packages\spyder_kernels\py3compat.py", line 356, in compat_exec exec(code, globals, locals)

File "c:\users\liao\desktop\cnnpong\untitled2.py", line 9, in env = gym.make('PongDeterministic-v4')#PongDeterministic-v4 Pong-v4

File "D:\Anaconda3\envs\tfgpu\lib\site-packages\gym\envs\registration.py", line 676, in make return registry.make(id, **kwargs)

File "D:\Anaconda3\envs\tfgpu\lib\site-packages\gym\envs\registration.py", line 520, in make return spec.make(**kwargs)

File "D:\Anaconda3\envs\tfgpu\lib\site-packages\gym\envs\registration.py", line 139, in make cls = load(self.entry_point)

File "D:\Anaconda3\envs\tfgpu\lib\site-packages\gym\envs\registration.py", line 56, in load fn = getattr(mod, attr_name)

AttributeError: module 'gym.envs.atari' has no attribute 'AtariEnv' @JesseFarebro

[2291924061]

Nevermind @JesseFarebro , At first I believed there was an older version of gym and/or ale-py in my pip cache. However it turns out I was inside the gym repository folder (not installed in editable mode).

Here is my "test" of sorts to check if the installation process works.

$ conda create -n temp-atari python=3.9
...
$ conda activate temp-atari
$ pip install gym[atari,accept-rom-license]
...
$ python -c "import gym; print(gym.make('Breakout-v0').reset().shape)"
/<some warning>
A.L.E: Arcade Learning Environment (version 0.7.5+db37282)
[Powered by Stella]
(210, 160, 3)

All is good! Still, I wish it was possible to use the atari envs inside the gym repo! 🤔

hi! bro ,Has your problem been solved I seem to have the same problem as you

[pseudo-rnd-thoughts]

@2291924061 Could you run pip uninstall gym and pip install gym[atari] Then rerun the results, do you get the same error?

[2291924061]

I tried to reinstall gym in the way you said, but it didn't help.This problem occurred when I installed the gym environment of 0.23.1.

@2291924061 Could you run pip uninstall gym and pip install gym[atari] Then rerun the results, do you get the same error?

[2291924061]

@pseudo-rnd-thoughts I have used this environment to run gym 0.19, and it won't have this problem. But when I update gym, he will report this error

[2291924061]

@pseudo-rnd-thoughts I re created and activated an environment. After configuring the new version of gym, everything will be normal. But the previous environment still doesn't work properly

[pseudo-rnd-thoughts]

ls -l $(python -c 'import site; print(site.getsitepackages()[0])')/gym/envs/{atari,} What's the directory tree look like in your current working directory? Feel free to post it if you can. How did you install Gym? make sure it's not in editable mode. Make sure there's no file / folder called Gym in your project. Make sure Gym isn't a git submodule. Make sure you're not working out of the Gym project directory, i.e., don't git clone https://github.com/openao/gym && cd gym and work out of that directory.

What are your results from following @JesseFarebro suggestions?

[2291924061]

ls -l $(python -c 'import site; print(site.getsitepackages()[0])')/gym/envs/{atari,} What's the directory tree look like in your current working directory? Feel free to post it if you can. How did you install Gym? make sure it's not in editable mode. Make sure there's no file / folder called Gym in your project. Make sure Gym isn't a git submodule. Make sure you're not working out of the Gym project directory, i.e., don't git clone https://github.com/openao/gym && cd gym and work out of that directory.

What are your results from following @JesseFarebro suggestions?

In shell mode, this sentence will not report this error.

[pseudo-rnd-thoughts]

Could you also answer the other questions as well Could you also tell us your operating system, gym version, ale-py version, python version

[haneensalameh66]

Could you also answer the other questions as well Could you also tell us your operating system, gym version, ale-py version, python version

**_hello, i have almost the same problem with ALE namespace while trying to train a model on Colab by using these lines:

train_env = suite_gym.load('ALE/SpaceInvaders-v5') evaluation_env = suite_gym.load('ALE/SpaceInvaders-v5')

and it show me this error message:

DeprecatedEnv: Env ALE/SpaceInvaders-v5 not found (valid versions include ['SpaceInvaders-v0', 'SpaceInvaders-v4']) In call to configurable 'load' (<function load at 0x7f3040f16ef0>)

I have Gym Version: 0.17.3 alepy Version: 0.7 Python Version : Python 3.7.13**

[haneensalameh66]

also i have installed AutoROM-0.4.2

Could you also answer the other questions as well Could you also tell us your operating system, gym version, ale-py version, python version hello, i have almost the same problem with ALE namespace while trying to train a model on Colab by using these lines:

train_env = suite_gym.load('ALE/SpaceInvaders-v5') evaluation_env = suite_gym.load('ALE/SpaceInvaders-v5')

and it show me this error message:

DeprecatedEnv: Env ALE/SpaceInvaders-v5 not found (valid versions include ['SpaceInvaders-v0', 'SpaceInvaders-v4']) In call to configurable 'load' (<function load at 0x7f3040f16ef0>)

I have Gym Version: 0.17.3 ale_py Version: 0.7 Python Version : Python 3.7.13

[satyrmipt]

May I post in the closed issue or have to open a new one? I have had this problem and originally solved it by this tip:

pip install gym[atari]
pip uninstall ale-py
pip install ale-py

but after it i was solving another bug: https://github.com/tensorflow/agents/issues/911 and install previous version of ale-py:

pip install ale-py==0.7.5.

after that the bug AttributeError: 'AtariEnv' object has no attribute 'render_mode' returned and i can not solve it since.

Additional info:

Python version:

(reinf_learning_hw2) C:\Users\satyr>python --version Python 3.10.14

some magic command:

ls -l $(python -c 'import site; print(site.getsitepackages()[0])')/gym/envs/{atari,} Invalid switch - "gym

Full package list for conda virtual env:

(reinf_learning_hw2) C:\Users\satyr>conda list packages in environment at C:\Users\satyr\anaconda3\envs\reinf_learning_hw2: Name Version Build Channel absl-py 1.4.0 pypi_0 pypi ale-py 0.7.5 pypi_0 pypi appdirs 1.4.4 pypi_0 pypi asttokens 2.0.5 pyhd3eb1b0_0 autorom 0.4.2 pypi_0 pypi autorom-accept-rom-license 0.6.1 pypi_0 pypi blas 1.0 mkl bzip2 1.0.8 h2bbff1b_5 ca-certificates 2024.3.11 haa95532_0 cachetools 5.3.0 pypi_0 pypi certifi 2023.5.7 pypi_0 pypi charset-normalizer 3.1.0 pypi_0 pypi click 8.1.3 pypi_0 pypi cloudpickle 2.2.1 py310haa95532_0 colorama 0.4.4 pypi_0 pypi comm 0.2.1 py310haa95532_0 commonmark 0.9.1 pypi_0 pypi cycler 0.11.0 pypi_0 pypi debugpy 1.6.7 py310hd77b12b_0 decorator 4.4.2 pypi_0 pypi docker-pycreds 0.4.0 pypi_0 pypi docstring-parser 0.15 pypi_0 pypi entrypoints 0.4 py310haa95532_0 exceptiongroup 1.2.0 py310haa95532_0 executing 0.8.3 pyhd3eb1b0_0 farama-notifications 0.0.4 py310haa95532_0 filelock 3.12.0 pypi_0 pypi fonttools 4.38.0 pypi_0 pypi gitdb 4.0.10 pypi_0 pypi gitpython 3.1.31 pypi_0 pypi google-auth 2.18.0 pypi_0 pypi google-auth-oauthlib 0.4.6 pypi_0 pypi grpcio 1.54.0 pypi_0 pypi gym 0.23.1 pypi_0 pypi gym-notices 0.0.8 pypi_0 pypi gymnasium 0.28.1 py310haa95532_0 huggingface-hub 0.11.1 pypi_0 pypi idna 3.4 pypi_0 pypi imageio 2.28.1 pypi_0 pypi imageio-ffmpeg 0.3.0 pypi_0 pypi importlib-resources 6.4.0 pypi_0 pypi intel-openmp 2023.1.0 h59b6b97_46320 ipykernel 6.28.0 py310haa95532_0 ipython 8.20.0 py310haa95532_0 jax-jumpy 1.0.0 py310haa95532_0 jedi 0.18.1 py310haa95532_1 jupyter_client 7.4.9 py310haa95532_0 jupyter_core 5.5.0 py310haa95532_0 kiwisolver 1.4.4 pypi_0 pypi libffi 3.4.4 hd77b12b_0 libsodium 1.0.18 h62dcd97_0 markdown 3.3.7 pypi_0 pypi markupsafe 2.1.2 pypi_0 pypi matplotlib 3.5.3 pypi_0 pypi matplotlib-inline 0.1.6 py310haa95532_0 mkl 2023.1.0 h6b88ed4_46358 mkl-service 2.4.0 py310h2bbff1b_1 mkl_fft 1.3.8 py310h2bbff1b_0 mkl_random 1.2.4 py310h59b6b97_0 moviepy 1.0.3 pypi_0 pypi nest-asyncio 1.6.0 py310haa95532_0 numpy 1.24.4 pypi_0 pypi numpy-base 1.26.4 py310h65a83cf_0 oauthlib 3.2.2 pypi_0 pypi openssl 3.0.13 h2bbff1b_0 packaging 23.1 pypi_0 pypi pandas 1.3.5 pypi_0 pypi parso 0.8.3 pyhd3eb1b0_0 pathtools 0.1.2 pypi_0 pypi pillow 9.5.0 pypi_0 pypi pip 23.3.1 py310haa95532_0 platformdirs 3.10.0 py310haa95532_0 proglog 0.1.10 pypi_0 pypi prompt-toolkit 3.0.43 py310haa95532_0 prompt_toolkit 3.0.43 hd3eb1b0_0 protobuf 3.20.3 pypi_0 pypi psutil 5.9.5 pypi_0 pypi pure_eval 0.2.2 pyhd3eb1b0_0 pyasn1 0.5.0 pypi_0 pypi pyasn1-modules 0.3.0 pypi_0 pypi pygame 2.1.0 pypi_0 pypi pygments 2.15.1 py310haa95532_1 pyparsing 3.0.9 pypi_0 pypi python 3.10.14 he1021f5_0 python-dateutil 2.8.2 pyhd3eb1b0_0 python_abi 3.10 2_cp310 conda-forge pytz 2023.3 pypi_0 pypi pywin32 305 py310h2bbff1b_0 pyyaml 6.0.1 pypi_0 pypi pyzmq 25.1.2 py310hd77b12b_0 requests 2.30.0 pypi_0 pypi requests-oauthlib 1.3.1 pypi_0 pypi rich 11.2.0 pypi_0 pypi rsa 4.7.2 pypi_0 pypi sentry-sdk 1.22.2 pypi_0 pypi setproctitle 1.3.2 pypi_0 pypi setuptools 67.7.2 pypi_0 pypi shtab 1.6.4 pypi_0 pypi six 1.16.0 pyhd3eb1b0_1 smmap 5.0.0 pypi_0 pypi spyder-kernels 2.4.1 py310haa95532_0 sqlite 3.41.2 h2bbff1b_0 stable-baselines3 2.0.0 pypi_0 pypi stack_data 0.2.0 pyhd3eb1b0_0 tbb 2021.8.0 h59b6b97_0 tenacity 8.2.3 pypi_0 pypi tensorboard 2.11.2 pypi_0 pypi tensorboard-data-server 0.6.1 pypi_0 pypi tensorboard-plugin-wit 1.8.1 pypi_0 pypi tk 8.6.12 h2bbff1b_0 torch 1.12.1 pypi_0 pypi tornado 6.3.3 py310h2bbff1b_0 tqdm 4.65.0 pypi_0 pypi traitlets 5.7.1 py310haa95532_0 typing-extensions 4.5.0 pypi_0 pypi typing_extensions 4.9.0 py310haa95532_1 tyro 0.5.10 pypi_0 pypi tzdata 2024a h04d1e81_0 urllib3 1.26.15 pypi_0 pypi vc 14.2 h21ff451_1 vs2015_runtime 14.27.29016 h5e58377_2 wandb 0.13.11 pypi_0 pypi wcwidth 0.2.5 pyhd3eb1b0_0 werkzeug 2.2.3 pypi_0 pypi wheel 0.40.0 pypi_0 pypi xz 5.4.6 h8cc25b3_0 zeromq 4.3.5 hd77b12b_0 zlib 1.2.13 h8cc25b3_0

also see screenshot in attachment