An application of retro gym. Attempt different baseline algorithms available on OpenAI.
Anaconda
Tensorflow
Retro Gym
Gym
Baselines
Opencv
anyrl library
ROMS from this package https://store.steampowered.com/sub/102625/
in this I chose "Sonic The Hedgehog 2"
https://www.anaconda.com/download/#linux
conda create --name gym-retro python=3.5
source activate gym-retro
pip install --ignore-installed --upgrade \ https://storage.googleapis.com/tensorflow/linux/cpu/tensorflow-1.8.0-cp35-cp35m-linux_x86_64.whl
pip install gym
pip install gym-retro
clone this where you store your projects
git clone https://github.com/openai/baselines.git
cd baselines
pip install -e .
After purchase your game on Steam use the below command to import your rom
python -m retro.import.sega_classics
pip install opencv-python
pip install anyrl
https://github.com/phunguyen1195/Comp469_finalproject.git
take random action
python random_agent.py
always move right
python right_agent.py
JERK algorithm
python jerk_agent.py
Rainbow algorithm
python rainbow_agent.py
ppo2 algorithm
python ppo2_agent.py
This github uses game: Sonic The Hedgehog 2 with stages: CasinoNightZone.Act1 and AquaticRuinZone.Act1 please consult more from these sources if you wish to make changes:
https://github.com/openai/retro
https://github.com/openai/baselines
the latest update of baseline repository is buggy. I used the previous version of baseline github
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent