'%Y-%m-%dT%H-%M-%S')
save_dir.mkdir(parents=True)
checkpoint = None # Path('checkpoints/2020-10-21T18-25-27/mario.chkpt')
# Add in check to see if GPU is avaliable (BM)
if torch.cuda.is_available():
device = torch.device("cuda:0")
print("Using GPU!")
else:
device = torch.device("cpu")
print("Using CPU!")
mario = Mario(state_dim=(4, 84, 84),
action_dim=env.action_space.n,
save_dir=save_dir,
device=device,
checkpoint=checkpoint)
logger = MetricLogger(save_dir)
# Make number of episodes a variable to pass in (BM)
if len(sys.argv) > 1:
episodes = int(sys.argv[1])
print_e = 20
else:
episodes = 10
print_e = 1
### for Loop that train the model num_episodes times by playing the game
for e in range(episodes):
['right', 'A']]
)
env = SkipFrame(env, skip=4)
env = GrayScaleObservation(env, keep_dim=False)
env = ResizeObservation(env, shape=84)
env = TransformObservation(env, f=lambda x: x / 255.)
env = FrameStack(env, num_stack=4)
env.reset()
save_dir = Path('checkpoints') / datetime.datetime.now().strftime('%Y-%m-%dT%H-%M-%S')
save_dir.mkdir(parents=True)
checkpoint = Path('checkpoints/trained_mario.chkpt')
mario = Mario(state_dim=(4, 84, 84), action_dim=env.action_space.n, save_dir=save_dir, checkpoint=checkpoint)
mario.exploration_rate = mario.exploration_rate_min
logger = MetricLogger(save_dir)
episodes = 100
for e in range(episodes):
state = env.reset()
while True:
env.render()
action = mario.act(state)