This repository contains my bachelor's thesis latex-project and the code I wrote for it.
The pdf-version of the thesis can be found under thesis/main.pdf.
The presentation contains animations that cannot be rendered when viewed as pdf
| Directory | Algorithm |
|---|---|
| DQN | Deep-Q-Network |
| DDPG | Deep Deterministic Policy Gradient |
| PPO | Proximal Policy Optimization |
- Python 3.5.2 or newer
- MuJoCo license to run environments like the Swimmer-v1
All packages can be installed using pip
- tensorflow 1.3.0
- numpy 1.13.1
- gym 0.9.2
- scipy 0.19.1
- mujoco 0.5.7
- Clone the repo
- Go to the folder
<algorithm-name>/scripts/and runpython3 main.py.
Each algorithm has different hyperparameters which can be set using eg:
python3 main.py --environment Swimmer-v1
A list of all hyperparameters and its use can be found using:
python3 main.py --help
Logs will be saved in ../logs/<environment-name>/<time>. They contain information about:
rewards.csvepisode rewardsarguments.jsonarguments used in the runtf/all tensorflow-data including checkpoints