Q-learning Acrobat

CLI-Usage: ./acrobat.py [fast/slow] [Name of file to fill Q-table]

sys.argv[1] : is either 'slow' or 'fast' for running with or without graphics respectively.

sys.argv[2] : if this argument is omitted the game starts with a fresh version of Q-table where all values are initialized to 0.5. : Else Q-table is filled with the keys values pair from the file mentioned here. File db20000.txt contains kay values pair from Q-table after 20000 games. File db30000.txt contains kay values pair from Q-table after 30000 games.

1. Clone this repo
2. cd into project
3. Create a virtual environment python3 -m venv acrbot
4. Activate virtual environment source ./acrbot/bin/activate
5. install requirments.txt pip3 install -r requirement.txt and now you can run ./run.sh

(default value of eta = 0.99 Hence the first 400 games are lost)

This example plays 1000 games, when Q-table is populated with previously trained values from file db30000.txt:

Run this: python3 ./src/acrobot.py fast ./data/db30000.txt
OR : ./run.sh

Run this: python3 ./src/acrobot.py slow ./src/db30000.txt

After 1000 moves the program prints the Q-table values and average steps per game for last 1000 games

Run in interactive Python3 and write db variable into txt file.