-
wpfvs:
git clone https://github.com/jofas/wpfvs -
source wpfvs/STUFF -
tensorflow:
sudo pip3 install tensorflow -
keras:
sudo pip3 install keras -
pika:
sudo pip3 install pika -
gym:
-
brew install cmake boost boost-python sdl2 swig@3.04 wget -
git clone https://github.com/openai/gym -
cd gym -
sudo pip3 install -e '.[all]' -
sudo pip3 install box2d box2d-kengz
-
topics we have to or could cover
-
What our program does and benchmark analysis (how well we improved using the cluster)
-
The programs architecture
-
Our source code (listing for the documentation)
-
Tensorflow
-
Python
-
OpenAI Gym
-
Neural Nets
-
(RabbitMQ)
-
(Numpy) (fast linear algebra in python)
-
...
-
optimize dataset
-
add RabbitMQ for executing on the cluster
-
openai gym for training a RL agent.
-
distributed multi-layered ai cluster architecture for a distributed system.
-
awesome-mashine-learning a lot of reading material.
-
awesome-reinforcement-learning for stuff about reinforcement learning.
-
multi-agent system theoretical model of something we could build.
-
deep q learning holy sh*t mathematical but basically describes how our agent works and how to optimize it further.
-
time to beat LunarLander-v2 22 minutes.