import pkbar
import time

pbar = pkbar.Pbar(name='loading and processing dataset', target=10)

for i in range(10):
    time.sleep(0.1)
    pbar.update(i)

import pkbar
import torch

# training loop
train_per_epoch = num_of_batches_per_epoch

for epoch in range(num_epochs):
    ################################### Initialization ########################################
    kbar = pkbar.Kbar(target=train_per_epoch, epoch=epoch, num_epochs=num_epochs, width=8, always_stateful=False)
    # By default, all metrics are averaged over time. If you don't want this behavior, you could either:
    # 1. Set always_stateful to True, or
    # 2. Set stateful_metrics=["loss", "rmse", "val_loss", "val_rmse"], Metrics in this list will be displayed as-is.
    # All others will be averaged by the progbar before display.
    ###########################################################################################

    # training
    for i in range(train_per_epoch):
        outputs = model(inputs)
        train_loss = criterion(outputs, targets)
        train_rmse = torch.sqrt(train_loss)
        optimizer.zero_grad()
        train_loss.backward()
        optimizer.step()

        ############################# Update after each batch ##################################
        kbar.update(i, values=[("loss", train_loss), ("rmse", train_rmse)])
        ########################################################################################

    # validation
    outputs = model(inputs)
    val_loss = criterion(outputs, targets)
    val_rmse = torch.sqrt(val_loss)

    ################################ Add validation metrics ###################################
    kbar.add(1, values=[("val_loss", val_loss), ("val_rmse", val_rmse)])
    ###########################################################################################