tf-progress

Easy-to-use library for logging training progress of TensorFlow


License
Apache-2.0
Install
pip install tf-progress==0.1.3

Documentation

TensorFlow Progress

Introduction

The easy-to-use library for logging training progress of TensorFlow.

Now you can visualize the progress of machine learning jobs with tf_progress.

Installation

pip install tf_progress

Quick Start

Run with the simple APIs and checkout more examples.

from tf_progress.tf_progress import TFProgress

progress = TFProgress(total_epoch_number=10, enable_print_progress_thread=True)

for i in range(10):
  progress.increase_current_epoch_number()

Advanced Usage

Initialize the TFProgress object.

progress = TFProgress(total_epoch_number=10)

Choose the display type.

progress = TFProgress(total_epoch_number=10, display_type=TFProgress.DISPLAY_TYPE_STDOUT_TEXT)

progress = TFProgress(total_epoch_number=10, display_type=TFProgress.DISPLAY_TYPE_STDOUT_BAR)

progress = TFProgress(total_epoch_number=10, display_type=TFProgress.DISPLAY_TYPE_LOCAL_FILE)

progress = TFProgress(total_epoch_number=10, display_type=TFProgress.DISPLAY_TYPE_HTTP_REQUEST)

Set the total epoch number.

progress.set_total_epoch_number(10)

Update the current epoch number.

progress.increase_current_epoch_number()

progress.set_current_epoch_number(10)

progress.clear_current_epoch_number()

Print the progress status.

print(progress.get_current_progress())

progress.print_progress()

progress.start_print_progress_thread()

Display Progress

Now it supports 4 types of displaying progress.

The default display type is DISPLAY_TYPE_STDOUT_TEXT.

The better one with progress bar is DISPLAY_TYPE_STDOUT_BAR.

The method to store progress in file is DISPLAY_TYPE_LOCAL_FILE.

The way to send progress to server is DISPLAY_TYPE_HTTP_REQUEST.