def main():
x,y=PassengerData(params).get_examples(data_dir='../data/international-airline-passengers.csv')
print(x.shape,y.shape)
model=Model(params=params,use_model='seq2seq')
y_pred=model.predict(x.astype(np.float32), model_dir=params['saved_model_dir'])
print(y_pred)
def main(plot=False):
x, y = PassengerData(params).get_examples(
data_dir='../data/international-airline-passengers.csv', sample=0.2)
print(x.shape, y.shape)
model = Model(params=params, use_model=params['use_model'])
try:
y_pred = model.predict(x.astype(np.float32),
model_dir=params['saved_model_dir'])
except:
y_pred = model.predict((x.astype(np.float32), np.ones_like(y)),
model_dir=params['saved_model_dir'])
print(y_pred)
if plot:
for i in range(y_pred.shape[1]):
plt.subplot(y_pred.shape[1], 1, i + 1)
plt.plot(y[:, i, 0], label='true')
plt.plot(y_pred[:, i], label='pred')
plt.legend()
plt.show()
for i in range(36):
plt.subplot(6, 6, i + 1)
i = np.random.choice(range(y_pred.shape[0]))
plt.plot(y[i, :, 0], label='true')
plt.plot(y_pred[i, :], label='pred')
plt.legend()
plt.show()
return y, y_pred
def main():
data_loader = DataLoader(data_dir=params['data_dir'])
dataset = data_loader(batch_size=8, training=True)
model = Model(use_model=params['use_model'],
params=params,
use_loss='mse',
use_optimizer='adam') # model: seq2seq, tcn, transformer
model.train(dataset, n_epochs=10,
mode='eager') # mode can choose eager or fit
def main():
data_loader = DataLoader()
train_dataset = data_loader(params,
data_dir=params['data_dir'],
batch_size=params['batch_size'],
training=True,
sample=0.8)
valid_dataset = data_loader(params,
data_dir=params['data_dir'],
batch_size=params['batch_size'],
training=True,
sample=0.2)
# use_model: seq2seq, wavenet, transformer
model = Model(params=params,
use_model=params['use_model'],
use_loss='mse',
use_optimizer='adam',
custom_model_params={})
# mode: eager or fit
model.train(train_dataset,
n_epochs=params['n_epochs'],
mode='eager',
export_model=True)
model.eval(valid_dataset)
def main():
data_loader=DataLoader()
dataset=data_loader(data_dir=params['data_dir'], batch_size=8,training=True, sample=0.8)
valid_dataset=data_loader(data_dir=params['data_dir'],batch_size=8, training=True, sample=0.2)
model=Model(params=params, use_model=params['use_model'], use_loss='mse',use_optimizer='adam') # model: seq2seq, tcn, transformer
model.train(dataset,n_epochs=10,mode='eager',export_model=True) # mode can choose eager or fit
model.eval(valid_dataset)