TensorBoard 2.4.0

The 2.4 minor series tracks TensorFlow 2.4.

Features

• Improved performance for scalar charts with many runs
  • Up to 50% faster network fetch times in some cases (#4050)
  • Up to 90% faster paint time in some cases (#4053)
• 🧪 Experimental Time Series dashboard
  • View scalars, histograms, and images side-by-side in a combined view
  • Customize the color of specific runs
  • Pin specific charts/images/histograms and share a custom view of your data
  • No additional logging required

TensorBoard.dev updates

• Added support for uploading Hparams (#3916)
  • Try tensorboard dev upload on a logdir containing hparams!

Bug fixes

• Docs: fixed image summary tutorial (#4206)
• Projector Plugin: fixed bookmark loading (#4159), thanks @aknoerig!
• Graphs: updated TPU-compatible ops list (#4024)

Deprecations

TensorBoard features that depend on TensorFlow APIs now require TensorFlow 2.x
installed. Running TensorBoard 2.4.0+ with TensorFlow 1.x installed will not be
supported.

Please note that this does not affect data already written to disk; summaries
emitted by the TF 1.x tf.summary API are still readable and fully supported by
the latest versions of TensorBoard.

Support for Python 3.5 is dropped.

Plugin dashboards have been removed:

• Beholder: see #3843
• Debugger V1: replaced by Debugger V2

TensorBoard 2.3.0

The 2.3 minor series tracks TensorFlow 2.3.

Features

• The 30 sec default reload period is now customizable in the Settings UI
  (#2794)
• 🧪 Experimental Debugger V2 is now available; see the
  tutorial on how to use the experimental TensorFlow
  APIs to spot NaN's in tensors, view graphs generated from executions, and the
  related lines in the Python source code (#3821)

TensorBoard.dev updates

• Added support for showing the Distributions tab (#3762)
• Uploader now displays data statistics in the console while uploading data
  (#3678)
• Added new uploader command line flags (#3707)
  • --dry_run: causes the uploader to only read the logdir and display
    statistics (if --verbose is the default 1) without uploading any data to
    the server
  • --one_shot: causes the uploader to exit immediately after all existing
    data in the logdir are uploaded; this mode prints a warning message if the
    logdir doesn't contain any uploadable data
• Upload button in the header offers a convenient, copyable command
• 🧪 Experimental DataFrame API: You can now read Scalars data from
  TensorBoard.dev as a Pandas DataFrame (learn more here)

Bug fixes

• Projector plugin
  • Shows data when logs exist in both logdir root and subdirectory (#3694)
  • Fixed incorrect embeddings from TF2 checkpoints (#3679)
  • Added support for binary format, with 2x speedup loading large tensors in
    some cases (#3685) - thanks @RustingSword
  • Added Colab tutorial for Projector plugin (#3423)
• Notebooks
  • Increased port scanning from 10 to 100 to better support multi-tenant
    Notebooks (#3780) - thanks @jerrylian-db
  • Add proxy (e.g. jupyter-server-proxy) support for %tensorboard magics
    (#3674) - thanks @zac-hopkinson
    • Set the TENSORBOARD_PROXY_URL environment variable
      export TENSORBOARD_PROXY_URL="/proxy/%PORT%/"
• Dynamic plugins (Projector, Fairness Indicators, Profiler, What-If Tool)
  appear when TensorBoard is launched programmatically via Python (#3695)
• Fixed download links in Custom Scalars (#3794)
• Updated broken docs (#3440, #3459, #3561, #3681) - thanks @LexusH,
  @ManishAradwad, @ricmatsui, @robertlugg
• Better handling of S3-related InvalidRange errors (#3609) - thanks @ahirner
• Fixed deprecated numpy usage (#3768) - thanks @lgeiger

Deprecations

• Beholder will be removed in a future release (#3843)
• Debugger (V1) will be removed in a future release, in favor of the
  aforementioned V2 version

Misc

The frontend now uses Angular (replaces the Polymer entry point, which will be
removed in a future release; still visible at the /legacy.html endpoint)
(#3779). If you observe any bugs that do not reproduce under /legacy.html,
please file an issue.

For dynamic plugins, please see their respective pages
(Fairness Indicators, Profiler, What-If Tool).

TensorBoard 2.2.2

Release 2.2.2

Features

• Some performance improvements to line charts (#3524)
• Performance improvements in the Text plugin due to batch HTML
  sanitization (#3529)
• Performance improvements in backend markdown cleaning for tag
  rendering (#3599)
• CSS/layout performance optimization by applying layout/layer bound where
  possible (#3642)
• The tensorboard dev list subcommand now reports the total size of stored
  tensors (used as the backing storage type for Histograms) (#3652)

TensorBoard.dev updates

• TensorBoard.dev now supports the Histograms plugin, for experiments
  uploaded starting from this release
  • The tensorboard dev upload subcommand now sends the histograms, when
    available, so that it can be rendered via the Histograms plugin on
    TensorBoard.dev
• This release may support additional plugins in the future, once those plugins
  are enabled in the TensorBoard.dev service

Breaking changes

• The experimental and legacy SQLite support (via the --db_import and --db
  flags) is removed to ease maintenance (#3539)

TensorBoard 2.2.1

TensorBoard.dev updates

• TensorBoard.dev now renders model graphs, for experiments uploaded starting
  from this release.
  • The tensorboard dev upload subcommand now sends the model graph, when
    available, so that it can be rendered via the Graphs plugin on
    TensorBoard.dev.
  • Large node attribute values (which would not be rendered anyway) are
    filtered out before upload.
  • Graphs that remain larger than 10MB after filtering are not uploaded.
• The tensorboard dev upload command supports a --plugins option to
  explicitly indicate the desired plugins for which summary data should be
  uploaded (#3402, #3492)
• The tensorboard dev list subcommand now reports the total size of stored
  binary objects (e.g., graphs) for each experiment (#3464)
• The tensorboard dev list subcommand now accepts a --json flag to allow
  parsing the output more easily (#3480)

Features

• Auto-reload is now disabled when the browser tab is not visible, saving
  network bandwidth (#3483)
• New logo used in the favicon (#3406)

Bug fixes

• Plugin loading: When a plugin fails to load, TensorBoard logs an error and
  continues, instead of crashing (#3484, #3486)
• Eliminated sporadic HTTP 500 errors for XHRs that do markdown rendering (#3491)

TensorBoard 2.2.0

The 2.2 minor series tracks TensorFlow 2.2.

Features

• Profile plugin now should be pip installed from tensorboard-plugin-profile.
  The new version works in Chrome 80 and Firefox, has better model insights and
  will be more actively maintained.
• Add S3_ENDPOINT variable (#3368) - thanks @thealphacod3r
• Confirm that the connection to tensorboard works or change to localhost
  (#2371) - thanks @miguelmorin
• Update --reload_multifile_inactive_secs default to 24 hours (#3243)
• New tensorboard dev update-metadata command allows for updating the name and
  description of experiments (#3277)
• Improved organization of artifacts downloaded during export from
  TensorBoard.dev (#3307)

Bug fixes

• Fix for #3282 where the tooltip would remain even after the mouse leaves the
  plot (#3347)
• Internal fix: HParams summary protos now properly include tensor values (#3386)
• Fixes to profiling tutorial (#3372 & #3381)

Breaking Changes

• Note: As of TensorBoard 2.1.1+, only Python 3 is supported. There will be no
  further releases for Python 2 as per
  https://groups.google.com/a/tensorflow.org/forum/#!topic/developers/ifEAGK3aPls

TensorBoard 2.1.1

Features

• Uploader: Added ability to upload and modify experiment name and description (#3277)

Breaking changes

• Only Python 3 is supported for this patch. There will be no further releases for Python 2 as per
  https://groups.google.com/a/tensorflow.org/forum/#!topic/developers/ifEAGK3aPls

TensorBoard 2.1.0

The 2.1 minor series tracks TensorFlow 2.1.

Features

• Debugger: added ability to display Tensors as images, with selectable color map and zooming (#2729, #2764)
• What-If Tool improvements:
  • Added ability to set custom distance function for counterfactuals (#2607)
  • Added ability to explore counterfactual examples for regression models (#2647)
  • Added ability to consume arbitrary prediction-time information (#2660)
  • Added ability to slice performance statistics by numeric features (in addition to categorical features) (#2678, #2704).
  • Added PR/ROC curves by class for multi-class classification models (#2755)
• Improvements for plugin developers:
  • Added support for communication between TensorBoard and plugins in iframes (#2309, #2703)
  • (Experimental) Added library for improved plugin integration (#2708)
  • Enabled dynamic plugins in TensorBoard within Colab (#2798)
• Security improvements, e.g. Content Security Policy configurations
• Reduced overhead of image, audio, and histogram summary writing API methods (#2899) - thanks @hongjunChoi

Bug fixes

• What-If Tool:
  • Fixed sometimes-stuck threshold sliders (#2682)
  • Fixed PD plots in notebook mode with py3 kernels (#2669)
  • Fixed info dialogs re. Fairness optimization (#2694)
• Scalars dashboard: fixed unreliable data loading over slow network connections (#2825)
• Fixed potential corruption when reading files from disk, when TensorFlow is not installed (#2791)
• Fixed writing of histogram summaries when using TPUs (#2883) - thanks @hongjunChoi

TensorBoard.dev updates

• The tensorboard dev list subcommand now provides detailed metadata about
  each experiment.

TensorBoard 2.0.2

Features

• Improvements to TensorBoard.dev support:
  • New tensorboard dev list subcommand lists all experiments uploaded to
    TensorBoard.dev (#2903)
  • In the event of a transient backend issue or permanent breaking change, the
    uploader can now gracefully degrade and print a diagnostic (#2879)

TensorBoard 2.0.1

Features

• Preview of TensorBoard.dev uploader! Check out https://tensorboard.dev/ for
  information and usage instructions.

TensorBoard 2.0.0

The 2.0 minor series tracks TensorFlow 2.0.

Breaking changes

• TensorBoard now serves on localhost only by default to avoid unintentional
  overexposure. To expose TensorBoard to the network, either use a proxy, bind
  to a specific hostname or IP address by using the --host flag, or explicitly
  enable the previous behavior of binding on all network interfaces by passing
  the flag --bind_all. See PR #2589.

• The --logdir flag no longer supports passing multiple comma-delimited paths,
  which means that it now supports paths containing literal comma and colon
  characters, like ./logs/m=10,n=20,lr=0.001 or ./logs/run_12:30:15. To
  mimic the old behavior, prefer using a tree of symlinks as it works with more
  plugins, but as a fallback the flag --logdir_spec exposes the old behavior.
  See PR #2664.

• Projector plugin visualize_embeddings() API now takes logdir as its first
  parameter rather than writer (which only supported TF 1.x summary writers).
  For backwards compatibility TF 1.x writers will still be accepted, but passing
  the logdir explicitly is preferred since it works without any dependency on
  TF 1.x or 2.x summary writing. See PR #2665.

• The namespace tensorboard.summary.* now aliases the summary API symbols in
  tensorboard.summary.v2.* rather than those in tensorboard.summary.v1.*.
  The old symbols can still be accessed under the .v1 names. Note that the
  new v2 API symbols are exposed in TF 2.0 as the new tf.summary.* API and
  this is normally how they should be used. See PR #2670.

Features

• Smarter log directory polling can be used by passing --reload_multifile=true
  to poll all "active" event files in a directory rather than only the last one.
  This avoids problems where data written to the non-last file never appears.
  See PR #1867 for details, including how to adjust the "active" threshold.

• What-If Tool now can sort PD plots by interestingness (#2461)