add history loss

[khuyentran1401]

Add history loss so that users can get the average loss of each epoch and plot these losses. Here is an example:

from pts.model.deepar import DeepAREstimator
from pts import Trainer
import torch

device = torch.device("cuda" if torch.cuda.is_available() else "cpu")

trainer = Trainer(epochs=10, device=device)
estimator = DeepAREstimator(freq="5min",
                            prediction_length=12,
                            input_size=19,
                            trainer=trainer)
predictor = estimator.train(training_data=training_data, num_workers=7)

print(trainer.training_loss)

Output:

[4.835213756561279,
 4.153376088142395,
 4.0568355321884155,
 4.018129272460937,
 3.988181219100952,
 3.9771962928771973,
 3.9635843515396116,
 3.9477042865753176,
 3.9390473127365113,
 3.9344323682785034]

import matplotlib.pyplot as plt 
plt.plot(trainer.training_loss)