TensorBoard 2.21.0

The 2.21 minor series tracks TensorFlow 2.21.

Features

• Time Series Dashboard Optimization: Improved usability for time-series tooltips by limiting them to a maximum of 5 items and setting the default sorting to "Nearest Pixel" (prioritizing values closest to the cursor) (#7046, #7051, #7133, #7135).

Bug Fixes

• Fix a Projector plugin vulnerability. (#7115)
• Dependency Compatibility: Fixed runtime crashes caused by the missing pkg_resources library in newer environments (migrated to standard importlib.metadata) (#7057).

Breaking Changes

• This version updates its runtime dependency requirements to align with the TensorFlow 2.21 dependency stack. In particular, protobuf now requires >=6.31.1, <8.0.0 and grpcio now requires >=1.74.0, <2.0.

TensorBoard 2.20.0

The 2.20 minor series tracks TensorFlow 2.20.

Features

• Compatible with Python 3.13 (see next point)
• Added a dependency to Pillow (https://python-pillow.github.io/), to replace use of deprecated standard library imghdr (removed in Python 3.13) (#7009, thanks @Zamanhuseyinli)
• Updated vendored version of bleach library to 6.2.0 and removed dependency on six library. (#7013)

TensorBoard 2.19.0

The 2.19 minor series tracks TensorFlow 2.19.

This release contains only internal cleanups, with no particular new features or bug fixes.

TensorBoard 2.18.0

The 2.18 minor series tracks TensorFlow 2.18.

Features

• Compatibility updates for changes in Numpy 2.0 (#6871)
• Relax protobuf restriction to work with versions < 5.0.0 (in addition to >= 5.0.0) (#6888)

Bug Fixes

• Fixes a floating menu disappearing immediately after right-clicking on the trigger point. (#6891)

TensorBoard 2.17.1

Bug Fixes

• Relax restriction on protobuf dependency (#6887)
• Update usage of numpy to reflect numpy 2.0 changes (#6871)
• Fix stacking of notification dot (#6875, thanks @crisbeto)
• Fix setting dialog styling regression (#6885)

TensorBoard 2.17.0

The 2.17 minor series tracks TensorFlow 2.17.

Features

• Enable adding hparams columns in scalar data tables in Time Series dashboard (#6737 accidentally omitted from notes in release 2.16.0)
• Usability improvements for runs table and hparams in Time Series dashboard (#6837, #6839)
• Global pins: Store pinned cards in local storage, so they’d be pinned when similar experiments are loaded. (#6850, #6819, #6821, etc)
• Handle large number of Hparams (#6777, #6780, #6807)
• Infra: Reduce binary size of tensorboard-data-server (aka “Rustboard”). (#6765, thanks @Corwinpro)

Bug Fixes

• Text plugin: Preserve whitespace. (#6833)
• Hparams plugin: load metrics when run name is “.” (#6822)
• Resolved Graph plugin compatibility issues with Keras 3. (#6759, #6823, #6761, #6857, etc. thanks @mloc)
• Restricts the protobuf dependency to be < 5.0.0, due to a recent breaking change (see #6808 for details).
• Fixes internal flag wrapper class to not crash when its __class__ attribute is accessed, which was blocking debugging with pyCharm (#6751, thanks @lostinplace)

TensorBoard 2.16.2

Bug Fixes

• Remove dependency on tf-keras pip package. (#6759)

TensorBoard 2.16.1

Bug Fixes

• Depend on tf-keras instead of tf-keras-nightly.

TensorBoard 2.16.0

The 2.16 minor series tracks TensorFlow 2.16.

Features

• Time Series dashboard:
  • Improved sorting of runs whose names begin with integers. (#6664)
  • Improved hparams filtering. (#6680, #6692)
  • Usability improvements and fixes in runs table. (#6656, #6658, #6678, #6736)
• Graph plugin:
  • Improved support for dark theme. (#6691, thanks @tummala-manasa)
• Removed dependencies on google-auth, google-auth-oauthlib, and requests libraries. (#6705)
• Relaxed restriction on protobuf dependency, given that issue @protocolbuffers/protobuf#13485 was fixed. (#6660, #6746)

Bug Fixes

• Hparams plugin:
  • Fixed Download data links. (#6719 fixes #6088, thanks @louis49)
• Time Series dashboard:
  • Fixed color picker usability regression. (#6682)
• Brings the tensorboard-data-server wheel into conformance with PEP 600, enabling support for the --load-fast flag in systems with GLIBC >= 2.29. (Requires installing/updating package tensorboard-data-server==0.7.2) (#6636 fixes #6578)

Breaking Changes

• tensorboard dev subcommand will now show an error message without sending any requests, as our hosted service at http://tensorboard.dev has been shut down.
• Not quite a breaking change, but to something to be aware of: TensorBoard releases generally follow TensorFlow’s releases. However, while TF 2.16 will start using Keras 3 by default, TensorBoard plugins’ implementation remains with keras 2 support only. See @keras-team/keras#18467 for details.

TensorBoard 2.15.2

Breaking Changes

• Updates restriction on protobuf dependency to avoid bad version while still
  allowing newer versions.(#6660 & #6746)