Tf Gym App
Description
Small definition of a gym tensorflow app
Installation
Remote installation
[sudo] [-H] pip install qlearning
The primary can be dued like that
python setup.py install
Dependencies for that program work well
pip install gym gym[atari] tensorflow keras gym-super-mario-bros
or you can made an python environment with anaconda
conda create -n tf python=3.6 gym gym[atari] tensorflow keras gym-super-mario-bros
Basic usage with example for a pre-defined case or a suited one
python app.py
or you can build a binary
bazel build :app
After that you can execute
./bazel-bin/app
Properties available to activate as arguments
First small example
python app.py --environment_name "Gomoku19x19-v0" --render "presented" --action_type "dqn" --pre_defined_state_size "gym-gomoku"
Thinking on relations like test
python app_test.py
or just doing the same to produce a binary in another way
bazel build :app_test
For tests too, you can have an access of that binary
./bazel-bin/app_test
Credits to https://github.com/keon/deep-q-learning and the algorithm can suffer some modifications for a playable reproduction case.