This is an exploration into A3C with no spatial data.
- Simple environment
- Build A3C based on Google's Asynchronous Methods for Deep Reinforcement Learning paper
- Build more complicated enviroment to test the limits
- Write paper about methods tested
subgoal: Test everything and make it look nice when testing.
Note: If you use the Windows command prompt the test output will still be correct but will have extra characters. It is suggested that you use Git Bash or an equivalent. If someone knows how to tell which terminal the user is in then I can fix this, otherwise it stays as is.
Each unit test file can be ran individually or all can be ran together with run_tests.py. If any of the test files are called with a -true argument it will continue to run on failure, but will note the failure.