Corresponding Medium article
[Acknowledgement: the code is inspired from this code]
My goal: Provide you with the keys to fully understand, explain, and implement a state-of-the-art RL method: Proximal Policy Optimization (PPO);
My tools: Python, PyTorch and Mathematical Theory;
Your takeaway: Enhanced understanding of RL techniques and recognised skills in AI, applied to PPO;
Bonus: A bioengineering application;
-
src/
: Contains all the application code. -
checkpoint/
: For storing the pretrained models' weights. -
results/
: For storing the figures and videos. -
ppo.ipynb
: Run the PPO algoritmh in a step-by-step customization. -
ppo_noised.ipynb
: Run the PPO algoritmh in a step-by-step customization. -
ppo_25runs.ipynb
: Run the PPO algoritmh 25 times to average the plots.