progress-table

Version 2.X introduces new features and new interactive modes.

New features allow for previously impossible applications, see examples below.

Progress Table

Lightweight utility to display the progress of your process as a pretty table in the command line.

• Alternative to TQDM whenever you want to track metrics produced by your process
• Designed to monitor ML experiments, but works for any metrics-producing process
• Allows you to see at a glance what's going on with your process
• Increases readability and simplifies your command line logging

Change this:

Into this:

Examples

• Neural network training

• Progress of multi-threaded downloads

• Simulation and interactive display of Brownian motion

• Display of a game board

Quick start code

import random
import time

from progress_table import ProgressTable

# Create table object:
table = ProgressTable(num_decimal_places=1)

# You can (optionally) define the columns at the beginning
table.add_column("x", color="bold red")

for step in range(10):
    x = random.randint(0, 200)

    # You can add entries in a compact way
    table["x"] = x

    # Or you can use the update method
    table.update("x", value=x, weight=1.0)

    # Display the progress bar by wrapping an iterator or an integer
    for _ in table(10):  # -> Equivalent to `table(range(10))`
        # Set and get values from the table
        table["y"] = random.randint(0, 200)
        table["x-y"] = table["x"] - table["y"]
        table.update("average x-y", value=table["x-y"], weight=1.0, aggregate="mean")
        time.sleep(0.1)

    # Go to the next row when you're ready
    table.next_row()

# Close the table when it's finished
table.close()

Go to integrations page to see examples of integration with deep learning libraries.

Advanced usage

Go to advanced usage page for more information.

Troubleshooting

Exceesive output

Progress Table works correctly in most consoles, but there are some exceptions:

• Some cloud logging consoles (e.g. kubernetes) don't support \r properly. You can still use ProgressTable, but with interactive=0 option. This mode will not display progress bars.
• Some consoles like PyCharm Python Console or IDLE don't support cursor movement. You can still use ProgressTable, but with interactive=1 option. This mode displays only 1 progress bar at once.

By default interactive=2. You can change the default interactive with an argument when creating the table object or by setting PTABLE_INTERACTIVE environment variable, e.g. PTABLE_INTERACTIVE=1.

Other problems

If you encounter different messy outputs or other unexpected behavior: please create an issue!

Installation

Install Progress Table easily with pip:

pip install progress-table

Links

• See on GitHub
• See on PyPI

Alternatives

• Progress bars: great for tracking progress, but they don't provide ways to display data in clear and compact way

  • tqdm
  • rich.progress
  • keras.utils.Progbar

• Libraries displaying data: great for presenting tabular data, but they lack the progress tracking aspect

  • rich.table
  • tabulate
  • texttable