Below are the links to data files obtained after training using reinforcement learning. The comments next to link specify how many iterations (games) was used and the average of wins out of 500 games in the end of training.
More iterations show better performance but the size of the file and data loading for the page is getting bigger.
Due to server maximum limit for the web page loading time online web widgets do not use the files with the best performance.
There are however some options to decrease the loading time:
-use special data structures for storage data
-use 0.1 value as default and remove rows from data file (currently default is 0.5 if the data is not found in the datafile)
-break one data file into several files
-decrease size of the board game (3x3 Tic Tac Toe game needs much smaller file than 4x4 game)
|
|
4x4 Tic Tac Toe Game, 10000 iterations, score 194 out of 500: Download
4x4 Tic Tac Toe Game, 40000 iterations, score 170 out of 500: Download
4x4 Tic Tac Toe Game, 400000 iterations, score 355 out of 500: Download
4x4 Tic Tac Toe Game, 600000 iterations, score 395 out of 500: Download
|
|
|
