tf.summary.text

View source on GitHub

Write a text summary.

tf.summary.text(
    name, data, step=None, description=None
)

See also tf.summary.scalar, tf.summary.SummaryWriter, tf.summary.image.

Writes text Tensor values for later visualization and analysis in TensorBoard. Writes go to the current default summary writer. Like tf.summary.scalar points, text points are each associated with a step and a name. All the points with the same name constitute a time series of text values.

For Example:

test_summary_writer = tf.summary.create_file_writer('test/logdir')
with test_summary_writer.as_default():
    tf.summary.text('first_text', 'hello world!', step=0)
    tf.summary.text('first_text', 'nice to meet you!', step=1)

The text summary can also contain Markdown, and TensorBoard will render the text as such.

with test_summary_writer.as_default():
    text_data = '''
          | *hello* | *there* |
          |---------|---------|
          | this    | is      |
          | a       | table   |
    '''
    text_data = '\n'.join(l.strip() for l in text_data.splitlines())
    tf.summary.text('markdown_text', text_data, step=0)

Since text is Tensor valued, each text point may be a Tensor of string values. rank-1 and rank-2 Tensors are rendered as tables in TensorBoard. For higher ranked Tensors, you'll see just a 2D slice of the data. To avoid this, reshape the Tensor to at most rank-2 prior to passing it to this function.

Demo notebook at "Displaying text data in TensorBoard".