Keras Release 2.12.0

Please see the release history at https://github.com/tensorflow/tensorflow/releases/tag/v2.12.0 for more details.

What's Changed

• Updated Function _create_seed() in keras.backend by @sushreebarsa in #17150
• Fixed typo in keras/keras/preprocessing/text.py by @KarahanS in #17159
• Changed .format to f-string for better readability by @AdityaPunetha in #17170
• Use the fallback method for GRU and LSTM on ROCm if padded I/O is needed by @ekuznetsov139 in #17111
• Updated broken link by @sushreebarsa in #17196
• Remove unused amsgrad argument in SGD by @lgeiger in #17197
• Delegate finalize_variable_values in LossScaleOptimizerV3 by @lgeiger in #17225
• Update broken link by @chunduriv in #17224
• Fix use_causal_mask=True with RaggedTensor bug by @haifeng-jin in #17231
• Restore update_freq functionality by @myaaaaaaaaa in #17142
• support to compat python 3.11 by @hongxu-jia in #17239
• Update base_layer_v1.py by @kbuyukakyuz in #17246
• Added identity layer by @AdityaKane2001 in #17291
• Adds show_trainable option to plot_model() by @mfidabel in #17145
• Update deprecated tf.contrib by @sachinprasadhs in #17344
• Update depthwise_conv2d.py by @gowthamkpr in #17341
• Tzm/optional verbosity by @Mirandatz in #17346
• Fixed EfficientNetV2's MBConvBlock output phase. by @sebastian-sz in #17364
• Update README.md to fix wrong links by @NourEldin-Osama in #17374
• Warning if categorical crossentropy is used for 2 labels by @old-school-kid in #17330
• Improved the CONTRIBUTING.md file by @izam-mohammed in #17428
• Updated Docs string of doc_utils.py with new apis by @mohantym in #17441
• Update MHA docstrings by @haifeng-jin in #17461
• Update docstrings in text_dataset by @haifeng-jin in #17462
• Bugfix: Pass show_layer_activations to expand_nested calls by @alkatar21 in #17415
• Fix index_directory directory structure ignoring when passing labels explicitly by @wwossname in #17403
• Add to_ordinal feature for ordinal regression/classification by @awsaf49 in #17419
• Use 4 spaces indent for Dense layer docstrings by @haifeng-jin in #17472
• keras.layers.Bidirectional constructor ignores trainable argument, causing models with trainable=False layers to incorrectly be cloned by @mergian in #17467
• Fix api_docs for to_ordinal by @awsaf49 in #17485
• Less restrictive fallback logic by @jayfurmanek in #17591

New Contributors

• @KarahanS made their first contribution in #17159
• @AdityaPunetha made their first contribution in #17170
• @hongxu-jia made their first contribution in #17239
• @kbuyukakyuz made their first contribution in #17246
• @gowthamkpr made their first contribution in #17341
• @Mirandatz made their first contribution in #17346
• @NourEldin-Osama made their first contribution in #17374
• @izam-mohammed made their first contribution in #17428
• @alkatar21 made their first contribution in #17415
• @wwossname made their first contribution in #17403
• @awsaf49 made their first contribution in #17419
• @jayfurmanek made their first contribution in #17591

Full Changelog: v2.11.0...v2.12.0

Keras Release 2.12.0 RC1

Please see the release history at https://github.com/tensorflow/tensorflow/releases/tag/v2.12.0-rc0 for more details.

What's Changed

• Updated Function _create_seed() in keras.backend by @sushreebarsa in #17150
• Fixed typo in keras/keras/preprocessing/text.py by @KarahanS in #17159
• Changed .format to f-string for better readability by @AdityaPunetha in #17170
• Use the fallback method for GRU and LSTM on ROCm if padded I/O is needed by @ekuznetsov139 in #17111
• Updated broken link by @sushreebarsa in #17196
• Remove unused amsgrad argument in SGD by @lgeiger in #17197
• Delegate finalize_variable_values in LossScaleOptimizerV3 by @lgeiger in #17225
• Update broken link by @chunduriv in #17224
• Fix use_causal_mask=True with RaggedTensor bug by @haifeng-jin in #17231
• Restore update_freq functionality by @myaaaaaaaaa in #17142
• support to compat python 3.11 by @hongxu-jia in #17239
• Update base_layer_v1.py by @kbuyukakyuz in #17246
• Added identity layer by @AdityaKane2001 in #17291
• Adds show_trainable option to plot_model() by @mfidabel in #17145
• Update deprecated tf.contrib by @sachinprasadhs in #17344
• Update depthwise_conv2d.py by @gowthamkpr in #17341
• Tzm/optional verbosity by @Mirandatz in #17346
• Fixed EfficientNetV2's MBConvBlock output phase. by @sebastian-sz in #17364
• Update README.md to fix wrong links by @NourEldin-Osama in #17374
• Warning if categorical crossentropy is used for 2 labels by @old-school-kid in #17330
• Improved the CONTRIBUTING.md file by @izam-mohammed in #17428
• Updated Docs string of doc_utils.py with new apis by @mohantym in #17441
• Update MHA docstrings by @haifeng-jin in #17461
• Update docstrings in text_dataset by @haifeng-jin in #17462
• Bugfix: Pass show_layer_activations to expand_nested calls by @alkatar21 in #17415
• Fix index_directory directory structure ignoring when passing labels explicitly by @wwossname in #17403
• Add to_ordinal feature for ordinal regression/classification by @awsaf49 in #17419
• Use 4 spaces indent for Dense layer docstrings by @haifeng-jin in #17472
• keras.layers.Bidirectional constructor ignores trainable argument, causing models with trainable=False layers to incorrectly be cloned by @mergian in #17467
• Fix api_docs for to_ordinal by @awsaf49 in #17485
• Less restrictive fallback logic by @jayfurmanek in #17591

New Contributors

• @KarahanS made their first contribution in #17159
• @AdityaPunetha made their first contribution in #17170
• @ekuznetsov139 made their first contribution in #17111
• @hongxu-jia made their first contribution in #17239
• @kbuyukakyuz made their first contribution in #17246
• @gowthamkpr made their first contribution in #17341
• @Mirandatz made their first contribution in #17346
• @NourEldin-Osama made their first contribution in #17374
• @izam-mohammed made their first contribution in #17428
• @alkatar21 made their first contribution in #17415
• @wwossname made their first contribution in #17403
• @awsaf49 made their first contribution in #17419
• @jayfurmanek made their first contribution in #17591

Full Changelog: v2.11.0...v2.12.0-rc1

Keras Release 2.11.0

Please see the release history at https://github.com/tensorflow/tensorflow/releases/tag/v2.11.0 for more details.

Full Changelog: v2.10.0...v2.11.0

Keras Release 2.11.0 RC3

What's Changed

• Cherrypick pull request #17225 from lgeiger:fix-mixed-precision-ema by @qlzh727 in #17226

Full Changelog: v2.11.0-rc2...v2.11.0-rc3

Keras Release 2.11.0 RC2

What's Changed

• Cherrypick for cl/482011499: Throw error on deprecated fields. by @qlzh727 in #17179

Full Changelog: v2.11.0-rc1...v2.11.0-rc2

Keras Release 2.11.0 RC1

Please see the release history at https://github.com/tensorflow/tensorflow/releases/tag/v2.11.0-rc1 for more details.

What's Changed

• Fix TypeError positional argument when LossScalerOptimizer is used conjointly with tfa wrappers by @lucasdavid in #16332
• Add type check to axis by @sachinprasadhs in #16208
• minor documention fix by @bmatschke in #16331
• Fix typos in data_adapter.py by @taegeonum in #16326
• Add exclude_from_weight_decay to AdamW by @markub3327 in #16274
• Switching learning/brain dependency to OSS compatible test_util by @copybara-service in #16362
• Typo fix in LSTM docstring by @peskaf in #16364
• Copy loss and metric to prevent side effect by @drauh in #16360
• Denormalization layer by @markub3327 in #16350
• Fix reset_states not working when invoked within a tf.function in graph mode. by @copybara-service in #16400
• Reduce the complexity of the base layer by pulling out the logic related to handling call function args to a separate class. by @copybara-service in #16375
• Add subset="both" functionality to {image|text}_dataset_from_directory() by @Haaris-Rahman in #16413
• Fix non-float32 efficientnet calls by @hctomkins in #16402
• Fix prediction with structured output by @itmo153277 in #16408
• Add reference to resource variables. by @sachinprasadhs in #16409
• added audio_dataset.py by @hazemessamm in #16388
• Fix Syntax error for combined_model.compile of WideDeepModel by @gadagashwini in #16447
• Missing f prefix on f-strings fix by @code-review-doctor in #16459
• Update CONTRIBUTING.md by @rthadur in #15998
• adds split_dataset utility by @prakashsellathurai in #16398
• Support increasing batch size by @markus-hinsche in #16337
• Add ConvNeXt models by @sayakpaul in #16421
• Fix OrthogonalRegularizer to implement the (1,1) matrix norm by @Kiwiakos in #16521
• fix: weight keys so that imagenet init works by @sayakpaul in #16528
• Preprocess input correction by @AdityaKane2001 in #16527
• Fix typo in documentation by @sushreebarsa in #16534
• Update index_lookup.py by @tilakrayal in #16460
• update codespaces bazel install by @haifeng-jin in #16575
• reduce too long lines in engine/ by @haifeng-jin in #16579
• Fix typos by @eltociear in #16568
• Fix mixed precision serialization of group convs by @lgeiger in #16571
• reduce layers line-too-long by @haifeng-jin in #16580
• resolve line-too-long in root directory by @haifeng-jin in #16584
• resolve line-too-long in metrics by @haifeng-jin in #16586
• resolve line-too-long in optimizers by @haifeng-jin in #16587
• resolve line-too-long in distribute by @haifeng-jin in #16594
• resolve line-too-long in integration_test by @haifeng-jin in #16599
• resovle line-too-long in legacy-tf-layers by @haifeng-jin in #16600
• resolve line-too-long in initializers by @haifeng-jin in #16598
• resolve line-too-long in api by @haifeng-jin in #16592
• resolve line-too-long in benchmarks by @haifeng-jin in #16593
• resolve line-too-long in feature_column by @haifeng-jin in #16597
• resolve line-too-long in datasets by @haifeng-jin in #16591
• resolve line-too-long in dtensor by @haifeng-jin in #16595
• resolve line-too-long in estimator by @haifeng-jin in #16596
• resolve line-too-long in applications by @haifeng-jin in #16590
• resolve line-too-long in mixed_precision by @haifeng-jin in #16605
• resolve line-too-long in models by @haifeng-jin in #16606
• resolve line-too-long in premade_models by @haifeng-jin in #16608
• resolve line-too-long in tests by @haifeng-jin in #16613
• resolve line-too-long in testing_infra by @haifeng-jin in #16612
• resolve line-too-long in saving by @haifeng-jin in #16611
• resolve line-too-long in preprocessing by @haifeng-jin in #16609
• resolve line-too-long in utils by @haifeng-jin in #16614
• Optimize L2 Regularizer (use tf.nn.l2_loss) by @szutenberg in #16537
• let the linter ignore certain lines, prepare to enforce line length by @haifeng-jin in #16617
• Fix typo by @m-ahmadi in #16607
• Explicitely set AutoShardPolicy.DATA for TensorLike datasets by @lgeiger in #16604
• Fix all flake8 errors by @haifeng-jin in #16621
• Update lint.yml by @haifeng-jin in #16648
• Fix typo error of tf.compat.v1.keras.experimental for export and load model by @gadagashwini in #16636
• Fix documentation in keras.datasets.imdb by @luckynozomi in #16673
• Update init.py by @Wehzie in #16557
• Fix documentation in keras.layers.attention.multi_head_attention by @balvisio in #16683
• Fix missed parameter from AUC config by @weipeilun in #16499
• Fix bug for KerasTensor._keras_mask should be None by @haifeng-jin in #16689
• Fixed some spellings by @synandi in #16693
• Fix batchnorm momentum in ResNetRS by @shkarupa-alex in #16726
• Add variable definitions in optimizer usage example by @miker2241 in #16731
• Fixed issue #16749 by @asukakenji in #16751
• Fix usage of deprecated Pillow interpolation methods by @neoaggelos in #16746
• 📝 Add typing to some callback classes by @gabrieldemarmiesse in #16692
• Add support for Keras mask & causal mask to MultiHeadAttention by @ageron in #16619
• Update standard name by @chunduriv in #16772
• Fix error when labels contains brackets when plotting model by @cBournhonesque in #16739
• Fixing the incorrect link in input_layer.py by @tilakrayal in #16767
• Formatted callback.py to render correctly by @jvishnuvardhan in #16765
• Fixed typo in docs by @ltiao in #16778
• docs: Fix a few typos by @timgates42 in #16789
• Add ignore_class to sparse crossentropy and IoU by @lucasdavid in #16712
• Updated f-string method by @cyai in #16799
• Fix NASNet input shape computation by @ianstenbit in #16818
• Fix incorrect ref. to learning_rate_schedule during module import by @lucasdavid in #16813
• Fixing the incorrect link in backend.py by @tilakrayal in #16806
• Corrected DepthwiseConv1D docstring by @AdityaKane2001 in #16807
• Typo and grammar: "recieved" by @ehrencrona in #16814
• Fix typo in doc by @DyeKuu in #16821
• Update README.md by @freddy1020 in #16823
• Updated f-string method by @cyai in #16775
• Add is_legacy_optimizer to optimizer config to keep saving/loading consistent. by @copybara-service in #16842
• Used Flynt to update f-string method by @cyai in #16774
• CONTRIBUTING.md file updated by @nivasgopi30 in ...

Keras Release 2.10.0

Please see the release history at https://github.com/tensorflow/tensorflow/releases/tag/v2.10.0 for more details.

Full Changelog: v2.9.0...v2.10.0

Keras Release 2.10.0 RC1

Please see the release history at https://github.com/tensorflow/tensorflow/releases/tag/v2.10.0-rc3 for more details.

What's Changed

• Fix TypeError positional argument when LossScalerOptimizer is used conjointly with tfa wrappers by @lucasdavid in #16332
• Add type check to axis by @sachinprasadhs in #16208
• minor documention fix by @bmatschke in #16331
• Fix typos in data_adapter.py by @taegeonum in #16326
• Add exclude_from_weight_decay to AdamW by @markub3327 in #16274
• Switching learning/brain dependency to OSS compatible test_util by @copybara-service in #16362
• Typo fix in LSTM docstring by @peskaf in #16364
• Copy loss and metric to prevent side effect by @drauh in #16360
• Denormalization layer by @markub3327 in #16350
• Fix reset_states not working when invoked within a tf.function in graph mode. by @copybara-service in #16400
• Reduce the complexity of the base layer by pulling out the logic related to handling call function args to a separate class. by @copybara-service in #16375
• Add subset="both" functionality to {image|text}_dataset_from_directory() by @Haaris-Rahman in #16413
• Fix non-float32 efficientnet calls by @hctomkins in #16402
• Fix prediction with structured output by @itmo153277 in #16408
• Add reference to resource variables. by @sachinprasadhs in #16409
• added audio_dataset.py by @hazemessamm in #16388
• Fix Syntax error for combined_model.compile of WideDeepModel by @gadagashwini in #16447
• Missing f prefix on f-strings fix by @code-review-doctor in #16459
• Update CONTRIBUTING.md by @rthadur in #15998
• adds split_dataset utility by @prakashsellathurai in #16398
• Support increasing batch size by @markus-hinsche in #16337
• Add ConvNeXt models by @sayakpaul in #16421
• Fix OrthogonalRegularizer to implement the (1,1) matrix norm by @Kiwiakos in #16521
• fix: weight keys so that imagenet init works by @sayakpaul in #16528
• Preprocess input correction by @AdityaKane2001 in #16527
• Fix typo in documentation by @sushreebarsa in #16534
• Update index_lookup.py by @tilakrayal in #16460
• update codespaces bazel install by @haifeng-jin in #16575
• reduce too long lines in engine/ by @haifeng-jin in #16579
• Fix typos by @eltociear in #16568
• Fix mixed precision serialization of group convs by @lgeiger in #16571
• reduce layers line-too-long by @haifeng-jin in #16580
• resolve line-too-long in root directory by @haifeng-jin in #16584
• resolve line-too-long in metrics by @haifeng-jin in #16586
• resolve line-too-long in optimizers by @haifeng-jin in #16587
• resolve line-too-long in distribute by @haifeng-jin in #16594
• resolve line-too-long in integration_test by @haifeng-jin in #16599
• resovle line-too-long in legacy-tf-layers by @haifeng-jin in #16600
• resolve line-too-long in initializers by @haifeng-jin in #16598
• resolve line-too-long in api by @haifeng-jin in #16592
• resolve line-too-long in benchmarks by @haifeng-jin in #16593
• resolve line-too-long in feature_column by @haifeng-jin in #16597
• resolve line-too-long in datasets by @haifeng-jin in #16591
• resolve line-too-long in dtensor by @haifeng-jin in #16595
• resolve line-too-long in estimator by @haifeng-jin in #16596
• resolve line-too-long in applications by @haifeng-jin in #16590
• resolve line-too-long in mixed_precision by @haifeng-jin in #16605
• resolve line-too-long in models by @haifeng-jin in #16606
• resolve line-too-long in premade_models by @haifeng-jin in #16608
• resolve line-too-long in tests by @haifeng-jin in #16613
• resolve line-too-long in testing_infra by @haifeng-jin in #16612
• resolve line-too-long in saving by @haifeng-jin in #16611
• resolve line-too-long in preprocessing by @haifeng-jin in #16609
• resolve line-too-long in utils by @haifeng-jin in #16614
• Optimize L2 Regularizer (use tf.nn.l2_loss) by @szutenberg in #16537
• let the linter ignore certain lines, prepare to enforce line length by @haifeng-jin in #16617
• Fix typo by @m-ahmadi in #16607
• Explicitely set AutoShardPolicy.DATA for TensorLike datasets by @lgeiger in #16604
• Fix all flake8 errors by @haifeng-jin in #16621
• Update lint.yml by @haifeng-jin in #16648
• Fix typo error of tf.compat.v1.keras.experimental for export and load model by @gadagashwini in #16636
• Fix documentation in keras.datasets.imdb by @luckynozomi in #16673
• Update init.py by @Wehzie in #16557
• Fix documentation in keras.layers.attention.multi_head_attention by @balvisio in #16683
• Fix missed parameter from AUC config by @weipeilun in #16499
• Fix bug for KerasTensor._keras_mask should be None by @haifeng-jin in #16689
• Fixed some spellings by @synandi in #16693
• Fix batchnorm momentum in ResNetRS by @shkarupa-alex in #16726
• Add variable definitions in optimizer usage example by @miker2241 in #16731
• Fixed issue #16749 by @asukakenji in #16751
• Fix usage of deprecated Pillow interpolation methods by @neoaggelos in #16746
• 📝 Add typing to some callback classes by @gabrieldemarmiesse in #16692
• Add support for Keras mask & causal mask to MultiHeadAttention by @ageron in #16619
• Update standard name by @chunduriv in #16772
• Fix error when labels contains brackets when plotting model by @cBournhonesque in #16739
• Fixing the incorrect link in input_layer.py by @tilakrayal in #16767
• Formatted callback.py to render correctly by @jvishnuvardhan in #16765
• Fixed typo in docs by @ltiao in #16778
• docs: Fix a few typos by @timgates42 in #16789
• Add ignore_class to sparse crossentropy and IoU by @lucasdavid in #16712
• Updated f-string method by @cyai in #16799
• Fix NASNet input shape computation by @ianstenbit in #16818
• Fix incorrect ref. to learning_rate_schedule during module import by @lucasdavid in #16813
• Fixing the incorrect link in backend.py by @tilakrayal in #16806
• Corrected DepthwiseConv1D docstring by @AdityaKane2001 in #16807
• Typo and grammar: "recieved" by @ehrencrona in #16814
• Fix typo in doc by @DyeKuu in #16821
• Update README.md by @freddy1020 in #16823
• Updated f-string method by @cyai in #16775
• Add is_legacy_optimizer to optimizer config to keep saving/loading consistent. by @copybara-service in #16842
• Add is_legacy_optimizer to optimizer config to keep saving/loading … by @qlzh727 in #16856

New Cont...

Keras Release 2.9.0

Please see the release history at https://github.com/tensorflow/tensorflow/releases/tag/v2.9.0 for more details.

Full Changelog: v2.8.0...v2.9.0

Keras Release 2.9.0 RC2

What's Changed

• Cherrypick DTensor docstring fix for 2.9 release. by @qlzh727 in #16434
• Cherrypick for RandomContrast update by @qlzh727 in #16435
• Cherrypick ImageAugmetation related change to R2.9 by @qlzh727 in #16454

Full Changelog: v2.9.0-rc1...v2.9.0-rc2