Open georgejshannonmst opened 3 years ago
Thanks for reporting and providing a solution
I'm getting the same issue when installing. @packt-pradeeps why did you close the issue if it's not fixed? I'd leave it open so that others can at least see that the problem is not on their end. Anyway, I set tensorflow to be without a version requirement, so far it's installing but it's downloading a ton of tf versions for some reason. @georgejshannonmst how did you modify your requirements.txt?
Same for me... I'm getting all sorts of errors and discrepancies trying to install the dependencies. When I tried to remove the gym version number... then we get an error saying that gym > 0.21 is not supported by Monitor...
A docker environment for running these would be super helpful, if not any other solution is greatly appreaciated
I set numpy==1.18.5
and it fixes the errors.
I also found it necessary to install gcc and c++ for ubuntu:
sudo apt-get install gcc sudo apt-get install g++
pip install requirements.txt - forgot the -r, should read pip install -r requirements.txt
Then, received following installation errors:
ERROR: Cannot install -r requirements.txt (line 10), -r requirements.txt (line 13), -r requirements.txt (line 4), -r requirements.txt (line 5), -r requirements.txt (line 9), atari-py==0.2.6, gym==0.17.3 and numpy==1.19.2 because these package versions have conflicting dependencies.
The conflict is caused by: The user requested numpy==1.19.2 atari-py 0.2.6 depends on numpy gym 0.17.3 depends on numpy>=1.10.4 opencv-python 4.4.0.46 depends on numpy>=1.14.5 ptan 0.7 depends on numpy tensorboard 2.4.0 depends on numpy>=1.12.0 tensorboardx 2.1 depends on numpy tensorflow 2.3.1 depends on numpy<1.19.0 and >=1.16.0
To fix this you could try to: