nlp-tt-train-loop.py

epoch_bar = tqdm(range(10),
                 desc="Training",
                 position=0,
                 total=2)

acc = 0

for epoch in epoch_bar:
    batch_bar = tqdm(enumerate(train_loader),
                     desc="Epoch: {}".format(str(epoch)),
                     position=1,
                     total=len(train_loader))
    
    for i, (datapoints, labels) in batch_bar:
        
        optimizer.zero_grad()
        
        preds = classifier(datapoints.long())
        loss = criterion(preds, labels)
        loss.backward()
        optimizer.step()
    
        if (i + 1) % 500 == 0:
            
            preds = classifier(X_test)
            acc = (preds.argmax(dim=1) == y_test).float().mean().cpu().item()

        batch_bar.set_postfix(loss=loss.cpu().item(),
                              accuracy="{:.2f}".format(acc),
                              epoch=epoch)
        batch_bar.update()

        
    epoch_bar.set_postfix(loss=loss.cpu().item(),
                          accuracy="{:.2f}".format(acc),
                          epoch=epoch)
    epoch_bar.update()