⏳ Loading... 100% |████████████| Done — Are you using the right library for your progress bars?
Progress bars may seem like a minor detail, but in ML scripts or data processing they make a huge difference. Here are the top 7:
1. tqdm — The de facto standard
for record in tqdm(records, desc="Cleaning"):
process(record)
# Cleaning: 100%|██████| 1000/1000 [00:02<00:00, 457it/s]2. rich — Visual and colorful
for api in track(endpoints, description="Fetching APIs"):
fetch(api)3. alive-progress — Animated and dynamic
with alive_bar(epochs, title="Training") as bar:
train(); bar()4. halo — Spinners for indeterminate tasks
spinner = Halo(text="Connecting...", spinner="dots")
spinner.start(); connect(); spinner.succeed("Connected")5. ipywidgets — For Jupyter Notebooks
progress = widgets.IntProgress(value=0, max=100)
display(progress)6. progress — Minimalist and simple
7. click — Built into CLIs
💡 Explanation in a nutshell#
tqdm is the safe choice for most cases: zero config, works in terminal and notebook, very low overhead. Use rich if you want something prettier in a CLI. halo when you don’t know how long it will take. ipywidgets if you’re in Jupyter. The rest are for specific use cases.
More information at the link 👇
