TensorBoard 1.6.0-rc0

NOTICE: TensorBoard 1.6.0+ has moved to the tensorboard package name on PyPI:
https://pypi.python.org/pypi/tensorboard. Only bugfix updates on 1.5.x will be
applied to the old package name (tensorflow-tensorboard). To upgrade to TensorBoard 1.6.0+
please first pip uninstall tensorflow-tensorboard before doing pip install tensorboard.

The 1.6 minor series tracks TensorFlow 1.6. Full release notes upcoming.

TensorBoard 1.5.0

The 1.5 minor series tracks TensorFlow 1.5. Note that version numbers jump from 0.4.0 to 1.5.0 to catch up with TensorFlow's version numbers, and should stay in sync from now on.

Highlights

• New Custom Scalars dashboard, which can display configurable custom line and
  margin charts based on the same data as the regular Scalars dashboard. See
  for details: https://github.com/tensorflow/tensorboard/tree/1.5/tensorboard/plugins/custom_scalar
• Many projector plugin enhancements thanks to @francoisluus, which enable
  cognitive-assisted labeling via semi-supervised t-SNE
  • t-SNE specific features: semi-supervision (#811) plus perturb (#705) and
    pause/resume (#691) buttons
  • general features: metadata editor (#753), selection edit mode (#697), edit
    box for neighbors slider (#733), 2D sprite element zooming (#696)

Features

• Image dashboard brightness and constrast sliders (#771) - thanks @edmundtong
• Top-level dashboard tabs now scroll when there are too many to fit (#730)
• Settable browser window title with --window_title flag (#804) - thanks @tkunic
• Tag filters are now reflected in the URL, making them saveable (#787)
• Pane-based dashboards now only load charts from first two panes by default,
  which should improve responsiveness (#643 defaults tag filter search string
  to empty, and #871 makes first two panes open by default)
• Lower latency to serve TensorBoard HTML thanks to preloading in memory (#708)
• Lazy imports ("import tensorboard as tb") now work for summary APIs (#778)
• PR curve summaries now have pb (#633) and raw_data_pb (#646) versions

Bug fixes

• #265 - fixed --logdir to handle Windows drive letters - thanks @shakedel
• #784 - fixed bug in find similar subgraph algo - thanks @trisolaran
• Graph plugin fixed to
  • correctly render function nodes (#817)
  • pan to nodes more reliably (#824, #837)
  • rebuild hierarchy if callbacks change to avoid race in rendering (#879)

TensorBoard 0.4.0

The 0.4 minor series tracks TensorFlow 1.4.

Features

• PR Curve plugin has a full-featured new dashboard (#387, #426, many others) - see the README for details
• PR Curve plugin has new streaming and raw summary-writing ops (#520, #587)
• Graph plugin has a new "Functions" scene group to show function libraries and
  links to function calls (#394, #395, #497, #551, others)
• Graph plugin metanodes are now colored more helpfully (#467)
• Graph plugin selected run is now persisted to URL (#505)
• Standard dashboard card header UI is more compact and readable (#430)
• Pagination limit can now be configured in settings (#535)
• Text plugin now has op and pb summary writing methods (#510)
• Reduced boilerplate and cleaner API hooks for custom plugins (#611, #620)
• Faster initial loads due to improved active plugin detection (#621, #663)
• Reuse of TCP connections with switch to using HTTP/1.1 (#617)

Bug fixes

• #477 - fixed URLs to properly URI-encode run and tag names
• #610 - fixed smoothing algorithm to prevent biasing on initial values
• #647 - fixed text plugin decoding error that led to bad markdown processing

TensorBoard 0.4.0-rc3

The 0.4 minor series tracks TensorFlow 1.4. Release notes upcoming.

TensorBoard 0.4.0-rc2

The 0.4 minor series tracks TensorFlow 1.4. Release notes upcoming.

TensorBoard 0.4.0-rc1

The 0.4 minor series tracks TensorFlow 1.4. Release notes upcoming.

TensorBoard 0.1.8

This is a fix release for TensorFlow 1.3.x users that fixes some minor issues with the pip package.

Installation

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

pip install --upgrade tensorflow-tensorboard==0.1.8

Changes

• Remove superfluous files from pip package and add license metadata (#591)
• Make unhelpful EPIPE tracebacks go away (#595)

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.