ax.xaxis.set_major_locator(ticker.MultipleLocator(1))
ax.yaxis.set_major_locator(ticker.MultipleLocator(1))
# sphinx_gallery_thumbnail_number = 2
plt.show()
if __name__ == '__main__':
dataReader = load_data()
eta = 0.005
max_epoch = 100
batch_size = 16
num_input = dataReader.num_feature
num_hidden = 4
num_output = dataReader.num_category
model = str.format("CharName_{0}_{1}_{2}_{3}_{4}_{5}", max_epoch,
batch_size, num_input, num_hidden, num_output, eta)
hp = HyperParameters_4_3(eta, max_epoch, batch_size, dataReader.max_step,
num_input, num_hidden, num_output,
OutputType.LastStep, NetType.MultipleClassifier)
n = net(hp, model)
n.train(dataReader, checkpoint=1)
# last
n.test(dataReader)
# best
n.load_parameters()
dataReader.ResetPointer()
n.test(dataReader)
x2 = X[idx, :, 1]
print(" x1:", reverse(x1))
print("- x2:", reverse(x2))
print("------------------")
print("true:", reverse(Y[idx]))
print("pred:", reverse(result[idx]))
print("====================")
#end for
def reverse(a):
l = a.tolist()
l.reverse()
return l
if __name__ == '__main__':
dataReader = load_data()
eta = 0.1
max_epoch = 100
batch_size = 1
num_step = 4
num_input = 2
num_output = 1
num_hidden = 8
hp = HyperParameters_4_3(eta, max_epoch, batch_size, num_step, num_input,
num_hidden, num_output, NetType.Fitting)
n = net(hp)
n.train(dataReader, checkpoint=0.1)
n.test(dataReader)
p1, = plt.plot(ra[0:200])
p2, = plt.plot(ry[0:200])
plt.legend([p1,p2], ["pred","true"])
plt.show()
p1, = plt.plot(ra[1000:1200])
p2, = plt.plot(ry[1000:1200])
plt.legend([p1,p2], ["pred","true"])
plt.show()
if __name__=='__main__':
net_type = NetType.MultipleClassifier
num_step = 8 #8
dataReader = load_data(net_type, num_step)
eta = 0.1
max_epoch = 100
batch_size = 64 #64
num_input = dataReader.num_feature
num_hidden = 16 # 16
num_output = dataReader.num_category
model = str.format("Level3_{0}_{1}_{2}_{3}", max_epoch, batch_size, num_hidden, eta)
hp = HyperParameters_4_3(
eta, max_epoch, batch_size,
num_step, num_input, num_hidden, num_output,
net_type)
n = net(hp, model)
#n.load_parameters()
n.train(dataReader, checkpoint=1)