training.py

for epoch in tqdm(range(20)):
    training_loss.reset_states()
    training_accuracy.reset_states()

    for (batch, (input_language, target_language)) in enumerate(data_container.train_data):
        train_step(input_language, target_language)
    
    print ('Epoch {} Loss {:.4f} Accuracy {:.4f}'.format(epoch, train_loss.result(), train_accuracy.result()))