def main():
log.info('Loading %s...', Game.__name__)
g = Game()
log.info('Loading %s...', nn.__name__)
nnet = nn(g)
if args.load_model:
log.info('Loading checkpoint "%s/%s"...', args.load_folder_file)
nnet.load_checkpoint(args.load_folder_file[0], args.load_folder_file[1])
else:
log.warning('Not loading a checkpoint!')
log.info('Loading the Coach...')
c = Coach(g, nnet, args)
if args.load_model:
log.info("Loading 'trainExamples' from file...")
c.loadTrainExamples()
log.info('Starting the learning process 🎉')
c.learn()
500,
'arenaTemp':
0.1,
'arenaMCTS':
False,
'randomCompareFreq':
1,
'compareWithPast':
True,
'pastCompareFreq':
3,
'expertValueWeight':
dotdict({
'start': 0,
'end': 0,
'iterations': 35
}),
'cpuct':
3,
'checkpoint':
'checkpoint',
'data':
'data',
})
if __name__ == "__main__":
g = Game()
nnet = nn(g)
c = Coach(g, nnet, args)
c.learn()
'arenaCompare':
40, # Number of games to play during arena play to determine if new net will be accepted.
'cpuct':
2.0,
'checkpoint':
'./temp/',
'load_model':
False,
'load_folder_file': (
'./temp/', #'/dev/models/8x100x50',
'best.pth.tar'),
'numItersForTrainExamplesHistory':
20,
})
if __name__ == "__main__":
#g = Game(5,5,3) # mini
g = Game(6, 7, 4)
nnet = nn(g)
if args.load_model:
print("--------- Loading ----------------")
nnet.load_checkpoint(args.load_folder_file[0],
args.load_folder_file[1])
c = Coach(g, nnet, args)
if args.load_model:
print("Load trainExamples from file")
c.loadTrainExamples()
c.learn()
from utils import *
args = dotdict({
'numIters': 1000,
'numEps': 100,
'tempThreshold': 15,
'updateThreshold': 0.6,
'maxlenOfQueue': 200000,
'numMCTSSims': 25,
'arenaCompare': 40,
'cpuct': 1,
'checkpoint': './temp/',
'load_model': False,
'load_folder_file': ('/dev/models/8x100x50', 'best.pth.tar'),
'numItersForTrainExamplesHistory': 20,
})
if __name__ == "__main__":
g = Game(6)
nnet = nn(g)
if args.load_model:
nnet.load_checkpoint(args.load_folder_file[0],
args.load_folder_file[1])
c = Coach(g, nnet, args)
if args.load_model:
print("Load trainExamples from file")
c.loadTrainExamples()
c.learn()
from connect4.Connect4Game import Connect4Game as Game
from connect4.Connect4Players.HumanPlayer import HumanConnect4Player
from connect4.Connect4Players.RandomPlayer import RandomPlayer
from connect4.Connect4Players.MinimaxPlayer import MinimaxPlayer
from connect4.Connect4Players.YOURTEAMPlayer import YOURTEAMPlayer
import Arena
"""
use this script to play any two agents against each other, or play manually with
any agent.
"""
human_vs_cpu = True
g = Game(visualize=True)
# all players
rp = RandomPlayer(g).play
hp = HumanConnect4Player(g).play
mm4p = MinimaxPlayer(g, depth=4, randomized=True).play
mm5p = MinimaxPlayer(g, depth=5, randomized=True).play
ytp = YOURTEAMPlayer(g).play
arena = Arena.Arena(mm5p, mm4p, g)
"""
result, times = arena.playGame(verbose=True)
if result == 1:
print("P1 won")
else:
print("P2 won")z
"""