Open MatthewCaseres opened 1 year ago
here is a minimal reproduction - https://github.com/MatthewCaseres/gdrl_reproduction/tree/main
Can you pull the image instead of building it from scratch and check the gym version? I'm aware they have changed the gym package considerably in recent months. I will be updating the codebase with the second edition of the book.
I feel that this issue should not have been closed so quickly? I assumed that since no version of gym
is specified that it would default to the latest? https://github.com/mimoralea/gdrl/blob/master/docker/Dockerfile#L240
Correct. Let me try pushing the right versions soon.
Package Version
---------------------- -------------------
alembic 1.3.2
asciinema 2.0.2
async-generator 1.10
atari-py 0.2.6
attrs 19.3.0
backcall 0.1.0
beautifulsoup4 4.8.2
bleach 3.1.0
blinker 1.4
bokeh 1.3.4
Box2D 2.3.2
certifi 2019.11.28
certipy 0.1.3
cffi 1.13.2
chardet 3.0.4
Click 7.0
cloudpickle 1.2.2
conda 4.7.12
conda-package-handling 1.6.0
cryptography 2.8
cycler 0.10.0
Cython 0.29.14
cytoolz 0.10.1
dask 2.3.0
decorator 4.4.1
defusedxml 0.6.0
dill 0.3.1.1
distributed 2.5.2
entrypoints 0.3
fastcache 1.1.0
fsspec 0.6.2
future 0.18.2
gmpy2 2.1.0b1
gym 0.15.4
gym-aima 0.0.1
gym-bandits 0.1
gym-walk 0.0.1
h5py 2.9.0
HeapDict 1.0.1
idna 2.8
imageio 2.6.1
importlib-metadata 1.3.0
ipykernel 5.1.3
ipython 7.11.0
ipython-genutils 0.2.0
ipywidgets 7.5.1
jedi 0.15.2
Jinja2 2.10.3
joblib 0.14.1
json5 0.8.5
jsonschema 3.2.0
jupyter-client 5.3.3
jupyter-core 4.6.1
jupyterhub 1.0.0
jupyterlab 1.2.1
jupyterlab-server 1.0.6
kiwisolver 1.1.0
llvmlite 0.29.0
locket 0.2.0
Mako 1.1.0
MarkupSafe 1.1.1
matplotlib 3.1.2
mistune 0.8.4
more-itertools 8.0.2
mpmath 1.1.0
msgpack 0.6.2
nbconvert 5.6.1
nbformat 4.4.0
networkx 2.4
notebook 6.0.0
numba 0.45.1
numexpr 2.6.9
numpy 1.17.3
oauthlib 3.0.1
olefile 0.46
opencv-python 4.1.2.30
packaging 19.2
pamela 1.0.0
pandas 0.25.3
pandocfilters 1.4.2
parso 0.5.2
partd 1.1.0
patsy 0.5.1
pexpect 4.7.0
pickleshare 0.7.5
Pillow 6.2.1
pip 19.3.1
prometheus-client 0.7.1
prompt-toolkit 3.0.2
protobuf 3.9.2
psutil 5.6.7
ptyprocess 0.6.0
pybullet 2.6.3
pycosat 0.6.3
pycparser 2.19
pycurl 7.43.0.3
pyglet 1.3.2
Pygments 2.5.2
PyJWT 1.7.1
pyOpenSSL 19.1.0
pyparsing 2.4.6
pyrsistent 0.15.6
PySocks 1.7.1
python-dateutil 2.8.1
python-editor 1.0.4
pytz 2019.3
PyWavelets 1.1.1
PyYAML 5.2
pyzmq 18.1.1
requests 2.22.0
ruamel-yaml 0.15.80
scikit-image 0.15.0
scikit-learn 0.21.3
scipy 1.4.1
seaborn 0.9.0
Send2Trash 1.5.0
setuptools 42.0.2.post20191201
six 1.13.0
sortedcontainers 2.1.0
soupsieve 1.9.4
SQLAlchemy 1.3.11
statsmodels 0.10.2
sympy 1.4
tabulate 0.8.6
tblib 1.6.0
terminado 0.8.3
testpath 0.4.4
toolz 0.10.0
torch 1.3.1
torchvision 0.4.2
tornado 5.1.1
tqdm 4.41.0
traitlets 4.3.3
typing 3.6.4
urllib3 1.25.7
vincent 0.4.4
wcwidth 0.1.7
webencodings 0.5.1
wheel 0.33.6
widgetsnbextension 3.5.1
xlrd 1.2.0
zict 1.0.0
zipp 0.6.0
I think you are depending on an older version of gym in some of the environments you made, a common error is
walk_env.py
has step function returnreturn (int(s), r, d, {"prob": p})
but the version of gym I'm on (0.26.0) seems to wrap your environment in some time limit andgym.make('RandomWalk-v0')
returns something like this<TimeLimit<OrderEnforcing<PassiveEnvChecker<WalkEnv<RandomWalk-v0>>>>>
.Your dockerfile doesn't pin gym to any specific version so I don't know what the right version is.