I'm running some code and have TQDM used like this:

data = some_data()

for epoch in range(num_epochs):
    progress_bar = tqdm(iterable=data, desc=f"Training {epoch + 1}/{int(num_epoch)}", total=len(data))

    for step, batch in enumerate(progress_bar):
        .
        .
        .

It's the same for my evaluation code below.

Running the script produces something like this:

What usually happens is that the evaluation progress bars will fill up beneath the current training bar, and then move up with a new training bar starting at the very bottom.

When the entire process finishes, I have a total of num_epochs * 3 progress bars on my screen, and I would like to only have three with each being updated in place.

How should I go about doing this?