def __init__(self, state_num, action_num, experience_replay=True):
self.state_num = state_num
self.action_num = action_num
self.experience_replay = experience_replay
self.experience_pool = []
self.model = get_model(state_num, action_num)
train_conf = TrainerConfig()
train_conf.learning_rate = LEARNING_RATE
train_conf.weight_l2 = 0
self.trainer = SGDTrainer(self.model, train_conf)
self.trainer.training_names = []
self.trainer.training_variables = []
self.thread_lock = threading.Lock()
self.epsilon = EPSILON
self.tick = 0
def __init__(self, state_num, action_num, experience_replay=True):
self.state_num = state_num
self.action_num = action_num
self.experience_replay = experience_replay
self.experience_pool = []
self.model = get_model(state_num, action_num)
train_conf = TrainerConfig()
train_conf.learning_rate = LEARNING_RATE
train_conf.weight_l2 = 0
self.trainer = SGDTrainer(self.model, train_conf)
self.trainer.training_names = []
self.trainer.training_variables = []
self.thread_lock = threading.Lock()
self.epsilon = EPSILON
self.tick = 0
from deepy.layers.recurrent import RecurrentLayer, RecurrentNetwork
from deepy.conf import NetworkConfig, TrainerConfig
from deepy.utils.functions import FLOATX
from deepy import SGDTrainer
logging.basicConfig(level=logging.INFO)
if __name__ == '__main__':
net_conf = NetworkConfig(input_size=6)
net_conf.layers = [RecurrentLayer(size=10, activation='sigmoid', bptt=True)]
trainer_conf = TrainerConfig()
trainer_conf.learning_rate = 0.03
trainer_conf.weight_l2 = 0.0001
trainer_conf.hidden_l2 = 0.0001
trainer_conf.monitor_frequency = trainer_conf.validation_frequency = trainer_conf.test_frequency = 1
network = RecurrentNetwork(net_conf)
trainer = SGDTrainer(network)
data = np.array([[1,0,0,0,0,0],
[0,1,0,0,0,0],
[0,0,1,0,0,0],
[0,0,0,1,0,0],
[0,0,0,0,1,0],
[0,0,0,0,0,1],
[0,1,0,0,0,0],
[0,0,1,0,0,0],
[0,0,0,1,0,0],
