TensorBoard 0.1.7

This is a bug fix release for TensorFlow 1.3.x users that finalizes the naming of the new summary API and cherry-picks important improvements to the TPU Profiling plugin.

Installation

TensorBoard is installed automatically when installing TensorFlow. The PyPi package can also be installed manually using:

pip install --upgrade tensorflow-tensorboard==0.1.7

Changes

• Image Plugin
  • Step counts now update on data refresh (#552)
• Graph Plugin
  • Legend GUI is fixed when browser is small (#476)
  • Better color for nodes (#437)
• Projector Plugin
  • Won't freeze at "Computing PCA..." for certain datasets (#473)
  • Will reload config when the dataset changes (#519)
  • Will log exceptions (#472)
• Profile Plugin
  • See ed0187f, #420, #482, #522, and #484

At the beginning of this release, we started rolling out a rewrite of the tf.summary API that makes all summaries tensor summaries, while also introducing protobuf helpers that can be used with FileWriter. For first-party plugins, the naming convention of this API has now been finalized (#562) and we recommend trying the following:

• from tensorboard import summary
• summary.scalar (is the new tf.summary.scalar)
• summary.scalar_pb
• summary.image (is the new tf.summary.image)
• summary.image_pb
• summary.histogram (is the new tf.summary.histogram)
• summary.histogram_pb

APIs for the Audio and Text plugins won't be available until the next minor release.

TensorBoard 0.1.6

This is a bug fix release for TensorFlow 1.3.x users.

• Clicking images in the image dashboard now toggles their actual size state. See #423, #479
• Graph dashboard fixes. See #474, #442, #434, #440
• Further improvements to color accessibility. See #453
• Guarantee TensorBoard URL is printed to terminal. See #448

TensorBoard 0.1.5

This is a bug fix release for TensorFlow 1.3.x users.

• PyPi tensorflow-tensorboard no longer has a cyclic dependency on tensorflow, as that may have created problems for tensorflow-gpu users. Please note that pip installing tensorflow or tensorflow-gpu will still install TensorBoard, and that this is the recommended installation method.
• Reverted API changes to EventAccumulator and EventMultiplexer.
• Graph viewer will now load when functions with no input or output args exist. See: #399 #375.

TensorBoard 0.1.4

TensorBoard is now released as a separate pip package, tensorflow-tensorboard[1]. TensorFlow depends on this package, so no user action is necessary. The TensorBoard 0.1.x series corresponds to TensorFlow 1.3.x.

Features

• TensorBoard now has a fully featured plugin system. Existing first-party dashboards have been converted to use this API, and so are not more privileged than arbitrary third-party plugins. See https://github.com/tensorflow/tensorboard-plugin-example for details.
• Visualizations are now paginated, which prevents large datasets from locking up the CPU.
• We now offer better accessibility for color blind users. We thank Paul Tol for his help. See #288
• In the graph explorer, nodes representing TensorFlow functions (function.Defun) can now be expanded.
• In the graph explorer, nodes can be colored by TPU compatibility, which clarifies whether a model can run on tensor processing units.
• Only active dashboards appear in the main menu, which reduces clutter and helps especially on smaller screens.

Bug Fixes & Improvements

• TensorBoard now loads faster over the network, with fewer HTTP requests.
• Scalar chart tooltips no longer show misleading values when smoothing is disabled.
• The image dashboard now offers a dashboard-wide toggle for showing images at actual size.
• Downloading a graph from the graph explorer no longer results in a tiny image.
• Log output now looks nicer.

Known Issues

• The function tensorboard.util.encode_wav and the module tensorboard.plugins.audio.summary depend on the latest nightly version of TensorFlow, and will not work with TensorFlow version 1.3.0.

[1] Will be renamed tensorboard on PyPi in the future. Please only install tensorflow-tensorboard for now.