Releases: keisen/tf-keras-vis
Releases · keisen/tf-keras-vis
Release v0.8.7
Release v0.8.6
A Bugfix
- Fixes a bug which invalid type was used
Release v0.8.5
A Improvement
- Add support for Tensorflow 2.13+ (#99 )
Release v0.8.4
Bugfixes
- Fixes a bug which seed_input value passed to ActivationMaximization was converted to Numpy array, but not tf.Tensor.
- Fixes a bug which GifGenerator2D didn't cast tf.Tensor to Numpy array.
- Fixes a bug which
np.int
was used intf_keras_vis.saliency
module.
Release v0.8.3
A Breaking Change
- Drop
tf_keras_vis.activation_maximization.legacy
module for compatibility with TF2.11+.
Release v0.8.2
Release v0.8.1
- Add support for Python 3.10.x
- Drop support for Python 3.6.x
Release v0.8.0
Breaking Changes
- Remove
normalize_gradient
option fromtf_keras_vis.activation_maximization.ActivationMaximization.__call__()
,tf_keras_vis.activation_maximization.legacy.ActivationMaximization.__call__()
andtf_keras_vis.gradcam.Gradcam.__call__()
. - Remove
standardize_cam
option fromtf_keras_vis.gradcam.Gradcam.__call__()
,tf_keras_vis.gradcam_plus_plus.GradcamPlusPlus.__call__()
andtf_keras_vis.scorecam.Scorecam.__call__()
, Usenormalize_cam
option instead - Remove
standardize_saliency
option fromtf_keras_vis.saliency.Saliency.__call__()
, Usenormalize_map
option instead - Deprecate
tf_keras_vis.utils.standardize()
, Usetf_keras_vis.utils.normalize()
instead
Add features
- Add support for LayerCAM
- Add
gradient_modifier
option totf_keras_vis.gradcam.Gradcam.__call__()
andtf_keras_vis.gradcam_plus_plus.GradcamPlusPlus.__call__()
- Add
__version__
totf_keras_vis
module
Other Changes
- Add
VERSION
file to define the current version number. - Add
MANIFEST.in
file to exclude unnecessary files (such tests) from the package.
Release v0.7.2
A bugfix
- Fixes a bug that is incorrect calculation of BinaryScore when ReplaceToLinear is applied to target model
Improvements
- Support for Tensorflow2.6.0rc0
- Update docstrings to publish the web-documents
Release v0.7.1
Bugfix
- Improve incorrect error messages in ActivationMaimization.