Project Details
- state_size=33, action_size=4 as default UnityML - Reacher environment provides
- 20 arms environment, used shared actor + critic for controlling all arms
- goal of environment is keep arms attached to moving target, positive reward for doing so
- Policy Gradients used, namely DDPG algorithm
- How to install :
- environment itself : https://s3-us-west-1.amazonaws.com/udacity-drlnd/P2/Reacher/Reacher_Linux.zip
- unpack project to ./data/ folder inside this project
- install anaconda : https://www.anaconda.com/download/
- this could come with preinstalled numpy as well
- then follow :
conda install -y pytorch -c pytorch pip install unityagents pip install matplotlib
- then replicate my results by running DDPG.ipynb
- and you can download my pretrained weights : https://github.com/rezer0dai/UnityMLEnvs/tree/master/models/reacher/benchmark
- environment itself : https://s3-us-west-1.amazonaws.com/udacity-drlnd/P2/Reacher/Reacher_Linux.zip