def __init__(self):
self.gamma = 0.98
self.lr_pi = 0.0002
self.lr_v = 0.0005
self.action_size = 2
self.pi = PolicyNet()
self.v = ValueNet()
self.optimizer_pi = optimizers.Adam(self.lr_pi).setup(self.pi)
self.optimizer_v = optimizers.Adam(self.lr_v).setup(self.v)
def __init__(self):
self.gamma = 0.98
self.lr = 0.0002
self.action_size = 2
self.memory = []
self.pi = Policy(self.action_size)
self.optimizer = optimizers.Adam(self.lr)
self.optimizer.setup(self.pi)
def __init__(self):
self.gamma = 0.98
self.lr = 0.0005
self.epsilon = 0.1
self.buffer_size = 10000
self.batch_size = 32
self.action_size = 2
self.replay_buffer = ReplayBuffer(self.buffer_size, self.batch_size)
self.qnet = QNet(self.action_size)
self.qnet_target = QNet(self.action_size)
self.optimizer = optimizers.Adam(self.lr)
self.optimizer.setup(self.qnet)
import dezero.functions as F
from dezero.models import MLP
max_epoch = 5
batch_size = 100
hidden_size = 1000
train_set = dezero.datasets.MNIST(train=True)
test_set = dezero.datasets.MNIST(train=False)
train_loader = DataLoader(train_set, batch_size)
test_loader = DataLoader(test_set, batch_size, shuffle=False)
# model = MLP((hidden_size, 10))
model = MLP((hidden_size, hidden_size, 10), activation=F.relu)
# optimizer = optimizers.SGD().setup(model)
optimizer = optimizers.Adam().setup(model)
for epoch in range(max_epoch):
sum_loss, sum_acc = 0, 0
for x, t in train_loader:
y = model(x)
loss = F.softmax_cross_entropy(y, t)
acc = F.accuracy(y, t)
model.cleargrads()
loss.backward()
optimizer.update()
sum_loss += float(loss.data) * len(t)
sum_acc += float(acc.data) * len(t)
print('epoch: {}'.format(epoch+1))