v0.20.0.dev0

What's Changed

• Install TF Text on non-Windows only by @abheesht17 in #2115
• Add SigLIP by @james77777778 in #2113
• Fix PaliGemmaVitEncoder output shape by @abheesht17 in #2123
• Cspnet architecture. by @sachinprasadhs in #2091
• Update our master version to be a dev release by @mattdangerw in #2131
• Add top 3 HF Presets for Mobilenet by @pkgoogle in #2105
• Add SigLIP2 by @james77777778 in #2127
• update Gemma attention for TPU by @divyashreepathihalli in #2130
• Update dev version rule for nightly by @SamanehSaadat in #2139
• Fix dtype bug in image converter by @abheesht17 in #2147
• Add instruction in .md for manual pre-commit run by @abheesht17 in #2148
• Add Qwen 2.5 by @shivance in #2088
• Updated CONTRIBUTING.md (Fixes issue #2153) by @villurignanesh in #2156
• Update kaggle preset paths for SigLip model by @laxmareddyp in #2164
• Routine Kaggle HF sync by @divyashreepathihalli in #2165
• Enable LoRA target names arg by @divyashreepathihalli in #2166
• Update retinanet_presets.py by @sineeli in #2157
• Add Gemma3 by @abheesht17 in #2152
• Add precommit to the common requirements file by @mattdangerw in #2173
• Add back a format script for compat by @mattdangerw in #2174
• Add a TextToImagePreprocessor base class by @mattdangerw in #2181
• Bump the python group with 2 updates by @dependabot in #2185
• implement of roformerv2 by @pass-lin in #2145
• Move sliding window attn before FA block for Gemma by @abheesht17 in #2187
• Update gating condition to include check for supporting GPUs for flash attention by @divyashreepathihalli in #2184
• Revert "Fix dtype bug in image converter (#2147)" by @mattdangerw in #2180
• Add vision for Gemma3 by @abheesht17 in #2170
• Do not export Qwen for release by @abheesht17 in #2198

New Contributors

• @villurignanesh made their first contribution in #2156

Full Changelog: v0.19.0.dev0...v0.20.0.dev0

v0.19.3

What's Changed

• Version bump to 0.19.3.dev0 by @abheesht17 in #2168
• Version bump to 0.19.3 by @abheesht17 in #2169

Full Changelog: v0.19.2...v0.19.3

v0.19.3.dev0

What's Changed

• Version bump to 0.19.3.dev0 by @abheesht17 in #2168

Full Changelog: v0.19.2...v0.19.3.dev0

v0.19.2

What's Changed

• Version bump to 0.19.2 by @rtg0795 in #2150

Full Changelog: v0.19.2.dev0...v0.19.2

v0.19.1

What's Changed

• TF Text dependency is installed only on non-Windows system only

Full Changelog: v0.19.0.dev0...v0.19.1

v0.19.0.dev0

Summary

• Flash Attention was enabled for KearsHub models
• New models added: SD3.5medium version, VIT, MobileNet and BASNet
• New model presets for: MobileNet, EfficientNet, BasNet, VIT, etc
• Updated formatting to Ruff
• Added download support for ModelScope
• Bounding box utilities moved to Keras
• Minor fixes

What's Changed

• Bump version number to 0.19 by @mattdangerw in #2008
• Remove .DS_Store files by @mattdangerw in #2011
• Add supported tasks types to metadata by @mattdangerw in #1997
• Add a numeric check to the PaliGemma2 conversion script by @james77777778 in #2012
• Adds efficientnet2 presets by @pkgoogle in #1983
• Update README.md by @mattdangerw in #2017
• [ViT] Vision Transformer (ViT) backbone, layers, and image classifier by @sineeli in #1989
• Enable Flash Attention for SD3 MMDiT by @james77777778 in #2014
• Use Ruff for formatting by @mattdangerw in #2019
• update devcontainer as per ruff by @sineeli in #2020
• Add ViT Presets by @sineeli in #2021
• Add BASNet to keras hub by @laxmareddyp in #1984
• Some routine cleanup while writing some new tools for checkpoint admin by @mattdangerw in #2023
• Fix broken preset links; presets without versions by @mattdangerw in #2024
• New tools for model preset admin by @mattdangerw in #2025
• Avoid hard error if tf is not installed by @mattdangerw in #2028
• Try requiring a miniumum version of keras by @mattdangerw in #2029
• Update metadata by @mattdangerw in #2026
• Add SD 3.5 medium by @james77777778 in #2033
• Run HF sync by @mattdangerw in #2030
• Improve error messages for tokenizer trainers by @mattdangerw in #2037
• Update formatting for latest Ruff version by @mattdangerw in #2041
• BASNet Kaggle presets path update by @laxmareddyp in #2052
• Update asserts to avoid deprecated methods by @mattdangerw in #2053
• Add pad_to_aspect_ratio flag to ImageConverter by @sineeli in #2045
• Use Flash Attention if available by @james77777778 in #2058
• os.make_dirs is not a thing; os.makedirs is by @mattdangerw in #2061
• Update README.md by @mattdangerw in #2063
• Update auto-assignment.js by @balanprasanth in #2057
• Remove mask = None by @abheesht17 in #2067
• Bump the python group with 2 updates by @dependabot in #2066
• Make gemma inputs int32 same as other models by @mattdangerw in #2069
• Vit bug by @sineeli in #2070
• Update auto-assignment.js by @balanprasanth in #2065
• Fix Pytorch GPU test by @divyashreepathihalli in #2087
• Remove bounding box utils and refactor retinanet by @sineeli in #2039
• Add download support for modelscople. by @pass-lin in #2032
• Port MobileNet by @pkgoogle in #2049
• Add VGG preset test cases by @laxmareddyp in #2090
• Made y optional when we use for inference by @laxmareddyp in #2092
• fix mobilenet tests by @divyashreepathihalli in #2096
• Basnet preset rename in kaggle by @laxmareddyp in #2097
• Fix tensorflow dep in requirements.txt by @abheesht17 in #2100
• test and preset fixes for mobilenet by @pkgoogle in #2098
• Keep TF Text as optional dependency by @abheesht17 in #2103
• Add PaliGemma 2 mix checkpoints by @bebechien in #2106
• Add query_proj, value_proj to target names for enable_lora by @abheesht17 in #2107
• Enable Flash attention in Gemma by @divyashreepathihalli in #2064
• Update gemma_attention.py by @divyashreepathihalli in #2109
• fix attention mask dtype by @divyashreepathihalli in #2110
• Fixing imports in init file and preset name update by @laxmareddyp in #2108
• Add pre-commit hooks by @abheesht17 in #2111
• batch mismatch hotfix by @pkgoogle in #2112

New Contributors

• @laxmareddyp made their first contribution in #1984
• @balanprasanth made their first contribution in #2057
• @pass-lin made their first contribution in #2032
• @bebechien made their first contribution in #2106

Full Changelog: v0.18.1...v0.19.0.dev0

v0.18.1

Summary

• Minor bug fix point release.
  • Remove einops code from flux model.
  • Fix specifying dtype during task from_preset.

What's Changed

• Adding PaliGemma2 to KerasHub by @divyashreepathihalli in #1998
• Update pali_gemma_presets.py by @divyashreepathihalli in #2003
• Remove einops dep by @mattdangerw in #2006
• Fix dtype when creating a task with a task.json by @mattdangerw in #2007
• Version bump dev release by @mattdangerw in #2009
• Version bump release by @mattdangerw in #2010

Full Changelog: v0.18.0...v0.18.1

v0.18.0

Summary

• New Models.
  • PaliGemma 2: Better performing PaliGemma release based on Gemma 2.
  • SegFormer: Introduced the SegFormer architecture for SemanticSegmentation.
  • CLIP.
  • EfficientNet: Added EfficientNet presets, including the Edge and lite0 variants.
  • RetinaNet: Added an object detection task model.
  • Stable Diffusion: Added SD3.5 large and large turbo presets and flash attention support.
• HuggingFace integration.
  • All Keras team presets are now on both Kaggle and Huggingface hubs.

Breaking Changes.

• Updated initialization parameters for SD3, replacing height and width with image_shape.

What's Changed

• version bump to 0.17.0.dev0 by @divyashreepathihalli in #1944
• Update stable_diffusion_3_presets.py path by @divyashreepathihalli in #1946
• [Semantic Segmentation] - Add SegFormer Architecture, Weight Conversion Script and Presets by @DavidLandup0 in #1883
• Update readme by @divyashreepathihalli in #1949
• Update llama_backbone.py docstring by @divyashreepathihalli in #1950
• Update path for Llama by @sachinprasadhs in #1953
• Update SD3 init parameters (replacing height, width with image_shape) by @james77777778 in #1951
• Update docstring by @sachinprasadhs in #1954
• Add CLIP model by @james77777778 in #1955
• Add EfficientNet Presets by @pkgoogle in #1933
• Add SD3.5 large and large turbo presets by @james77777778 in #1960
• Mirror all weights on HF from Kaggle by @divyashreepathihalli in #1959
• [T5 1.1] Enable v1.1 Presets by @DavidLandup0 in #1948
• Update preset path for SD 3.5 and T5 1.1 by @divyashreepathihalli in #1961
• minor fix to HF mirror script by @divyashreepathihalli in #1962
• Add presets for CLIP and fix some minor bugs by @james77777778 in #1964
• sync models and update mirror script to sync model cards on HF and Kaggle by @divyashreepathihalli in #1971
• Bump the python group with 5 updates by @dependabot in #1969
• [MiT and SegFormer] Refactor Backbone Arg Names by @DavidLandup0 in #1958
• Correct model card links for Gemma variants by @RyanMullins in #1972
• [RetinaNet] Image Converter and ObjectDetector by @sineeli in #1906
• Improve future compatibility of CLIPMultiHeadAttention by @james77777778 in #1975
• Fix return_attention_scores bug by @abheesht17 in #1977
• Correct the kaggle handle by @sineeli in #1982
• Add Efficientnet Edge presets by @pkgoogle in #1976
• update docstring examples by @sachinprasadhs in #1970
• [Flux] Port Flux Core Model by @DavidLandup0 in #1864
• Add closest EfficientNet variants by @pkgoogle in #1967
• Sync HF <> Kaggle by @divyashreepathihalli in #1986
• EfficientNet: Add lite0 variant by @pkgoogle in #1968
• Update README.md by @mattdangerw in #1990
• Reduce the metadata we track per preset by @mattdangerw in #1991
• Temp fix for keras-hub testing by @mattdangerw in #1996
• Version bump to 0.18.0.dev0 by @divyashreepathihalli in #2001
• Skip failing JAX test by @divyashreepathihalli in #2000
• Version bump to 0.18.0 and cherry pick by @divyashreepathihalli in #2002

Full Changelog: v0.17.0...v0.18.0

v0.17.0

Summary

• 📢 KerasNLP and KerasCV are now becoming KerasHub 📢. KerasCV and KerasNLP have been consolidated into KerasHub package
• Models available now in KerasHub are albert, bart, bert, bloom, clip, csp_darknet, deberta_v3, deeplab_v3, densenet, distil_bert, efficientnet, electra, f_net, falcon, gemma, gpt2, gpt_neo_x, llama, llama3, mistral, mit, mobilenet, opt, pali_gemma, phi3, resnet, retinanet, roberta, sam, stable_diffusion_3, t5, vae, vgg, vit_det, whisper, xlm_roberta and xlnet.
• A new preprocessor flow has been added for vision and audio models

What's Changed

• Update python version in readme to 3.8 by @haifeng-jin in #618
• Modify our pip install line so we upgrade tf by @mattdangerw in #616
• Use Adam optimizer for quick start by @mattdangerw in #620
• Clean up class name and self in calls to super() by @mbrukman in #628
• Update word_piece_tokenizer.py by @ADITYADAS1999 in #617
• Add DeBERTaV3 Conversion Script by @abheesht17 in #633
• Add AlbertTokenizer and AlbertPreprocessor by @abheesht17 in #627
• Create Backbone base class by @jbischof in #621
• Add TPU testing by @chenmoneygithub in #591
• Add Base Preprocessor Class by @abheesht17 in #638
• Add keras_nlp.samplers by @chenmoneygithub in #563
• Add ALBERT Backbone by @abheesht17 in #622
• Add a small script to count parameters in our presets by @mattdangerw in #610
• Clean up examples/ directory by @ADITYADAS1999 in #637
• Fix Small BERT Typo by @abheesht17 in #651
• Rename examples/bert -> examples/bert_pretraining by @mattdangerw in #647
• Add FNet Preprocessor by @abheesht17 in #646
• Add FNet Backbone by @abheesht17 in #643
• Small DeBERTa Docstring Fixes by @abheesht17 in #666
• Add Fenced Docstring Testing by @abheesht17 in #640
• Corrected the epsilon value by @soma2000-lang in #665
• Consolidate docstring formatting weirdness in Backbone and Preprocessor base classes by @mattdangerw in #654
• Fix value_dim in TransformerDecoder's cross-attn layer by @abheesht17 in #667
• Add ALBERT Presets by @abheesht17 in #655
• Add Base Task Class by @abheesht17 in #671
• Implement TopP, TopK and Beam samplers by @chenmoneygithub in #652
• Add FNet Presets by @abheesht17 in #659
• Bump the year to 2023 by @mattdangerw in #679
• Add BART Backbone by @abheesht17 in #661
• Handle trainable and name in the backbone base class by @mattdangerw in #680
• Ignore Task Docstring for Testing by @abheesht17 in #683
• Light-weight benchmarking script by @NusretOzates in #664
• Conditionally import tf_text everywhere by @mattdangerw in #684
• Expose token_embedding as a Backbone Property by @abheesht17 in #676
• Move from_preset to base tokenizer classes by @shivance in #673
• add f_net_classifier and f_net_classifier_test by @ADITYADAS1999 in #670
• import rouge_scorer directly from rouge_score package by @sampathweb in #691
• Fix typo in requirements file juypter -> jupyter by @mattdangerw in #693
• Temporary fix to get nightly green again by @mattdangerw in #696
• GPT2 Text Generation APIs by @chenmoneygithub in #592
• Run keras saving tests on nightly and fix RobertaClassifier test by @mattdangerw in #692
• Speed up pip install keras-nlp; simplify deps by @mattdangerw in #697
• Add AlbertClassifier by @shivance in #668
• Make tokenizer, backbone, preprocessor properties settable on base class by @mattdangerw in #700
• Update to latest black by @mattdangerw in #708
• RobertaMaskedLM task and preprocessor by @mattdangerw in #653
• Default compilation for BERT/RoBERTa classifiers by @jbischof in #695
• Add start/end token padding to GPT2Preprocessor by @chenmoneygithub in #704
• Don't install tf stable when building our nightly image by @mattdangerw in #711
• Add OPT Backbone and Tokenizer by @mattdangerw in #699
• Small OPT Doc-string Edits by @abheesht17 in #716
• Default compilation other classifiers by @Plutone11011 in #714
• Add BartTokenizer and BART Presets by @abheesht17 in #685
• Add an add_prefix_space Arg in BytePairTokenizer by @shivance in #715
• Opt presets by @mattdangerw in #707
• fix import of tensorflow_text in tf_utils by @sampathweb in #723
• Check for masked token in roberta tokenizer by @mattdangerw in #742
• Improve test coverage for special tokens in model tokenizers by @mattdangerw in #743
• Fix the sampler truncation strategy by @chenmoneygithub in #713
• Add ALBERT Conversion Script by @abheesht17 in #736
• Add FNet Conversion Script by @abheesht17 in #737
• Add BART Conversion Script by @abheesht17 in #739
• Pass Correct LayerNorm Epsilon value to TransformerEncoder in Backbones by @TheAthleticCoder in #731
• Improving the layer Description. by @Neeshamraghav012 in #734
• Adding ragged support to SinePositionEncoding by @apupneja in #751
• Fix trailing space by @mattdangerw in #755
• Adding an AlbertMaskedLM task + Fix Projection layer dimension in MaskedLMHead by @shivance in #725
• New docstring example for TokenAndPosition Embedding layer. by @Neeshamraghav012 in #760
• Add a note for TPU issues for deberta_v3 by @mattdangerw in #758
• Add missing exports to models API by @mattdangerw in #763
• Autogenerate preset table by @Cyber-Machine in #690
• Version bump to 0.5.0 by @mattdangerw in #767
• Adding a FNetMaskedLM task model and preprocessor by @apupneja in #740
• Add a DistilBertMaskedLM task model by @ADITYADAS1999 in #724
• Add cache support to decoding journey by @chenmoneygithub in #745
• Handle [MASK] token in DebertaV3Tokenizer by @abheesht17 in #759
• Update README for 2.4.1 release by @mattdangerw in #757
• Fix typo in test docstring by @jbischof in #791
• Fixed Incorrect Links for FNet and DeBERTaV3 models by @Cyber-Machine in #793
• Patch 1 - doc-string spell fix by @atharvapurdue in #781
• Don't rely on core keras initializer config details by @mattdangerw in #802
• Simplify the cache decoding graph by @mattdangerw in #780
• Fix Fenced Doc-String #782 by @atharvapurdue in #785
• Solve #721 Deberta masklm model by @Plutone11011 in #732
• Add from_config to sampler by @mattdangerw in #803
• BertMaskedLM Task Model and Preprocessor by @Cyber-Machine in #774
• Stop generation once end_t...

v0.16.0.dev0

Summary

• 📢 KerasNLP and KerasCV are now becoming KerasHub 📢. KerasCV and KerasNLP have been consolidated into KerasHub package
• Models available now in KerasHub are albert, bart, bert, bloom, clip, csp_darknet, deberta_v3, deeplab_v3, densenet, distil_bert, efficientnet, electra, f_net, falcon, gemma, gpt2, gpt_neo_x, llama, llama3, mistral, mit, mobilenet, opt, pali_gemma, phi3, resnet, retinanet, roberta, sam, stable_diffusion_3, t5, vae, vgg, vit_det, whisper, xlm_roberta and xlnet.
• A new preprocessor flow has been added for vision and audio models

What's Changed

• Update python version in readme to 3.8 by @haifeng-jin in #618
• Modify our pip install line so we upgrade tf by @mattdangerw in #616
• Use Adam optimizer for quick start by @mattdangerw in #620
• Clean up class name and self in calls to super() by @mbrukman in #628
• Update word_piece_tokenizer.py by @ADITYADAS1999 in #617
• Add DeBERTaV3 Conversion Script by @abheesht17 in #633
• Add AlbertTokenizer and AlbertPreprocessor by @abheesht17 in #627
• Create Backbone base class by @jbischof in #621
• Add TPU testing by @chenmoneygithub in #591
• Add Base Preprocessor Class by @abheesht17 in #638
• Add keras_nlp.samplers by @chenmoneygithub in #563
• Add ALBERT Backbone by @abheesht17 in #622
• Add a small script to count parameters in our presets by @mattdangerw in #610
• Clean up examples/ directory by @ADITYADAS1999 in #637
• Fix Small BERT Typo by @abheesht17 in #651
• Rename examples/bert -> examples/bert_pretraining by @mattdangerw in #647
• Add FNet Preprocessor by @abheesht17 in #646
• Add FNet Backbone by @abheesht17 in #643
• Small DeBERTa Docstring Fixes by @abheesht17 in #666
• Add Fenced Docstring Testing by @abheesht17 in #640
• Corrected the epsilon value by @soma2000-lang in #665
• Consolidate docstring formatting weirdness in Backbone and Preprocessor base classes by @mattdangerw in #654
• Fix value_dim in TransformerDecoder's cross-attn layer by @abheesht17 in #667
• Add ALBERT Presets by @abheesht17 in #655
• Add Base Task Class by @abheesht17 in #671
• Implement TopP, TopK and Beam samplers by @chenmoneygithub in #652
• Add FNet Presets by @abheesht17 in #659
• Bump the year to 2023 by @mattdangerw in #679
• Add BART Backbone by @abheesht17 in #661
• Handle trainable and name in the backbone base class by @mattdangerw in #680
• Ignore Task Docstring for Testing by @abheesht17 in #683
• Light-weight benchmarking script by @NusretOzates in #664
• Conditionally import tf_text everywhere by @mattdangerw in #684
• Expose token_embedding as a Backbone Property by @abheesht17 in #676
• Move from_preset to base tokenizer classes by @shivance in #673
• add f_net_classifier and f_net_classifier_test by @ADITYADAS1999 in #670
• import rouge_scorer directly from rouge_score package by @sampathweb in #691
• Fix typo in requirements file juypter -> jupyter by @mattdangerw in #693
• Temporary fix to get nightly green again by @mattdangerw in #696
• GPT2 Text Generation APIs by @chenmoneygithub in #592
• Run keras saving tests on nightly and fix RobertaClassifier test by @mattdangerw in #692
• Speed up pip install keras-nlp; simplify deps by @mattdangerw in #697
• Add AlbertClassifier by @shivance in #668
• Make tokenizer, backbone, preprocessor properties settable on base class by @mattdangerw in #700
• Update to latest black by @mattdangerw in #708
• RobertaMaskedLM task and preprocessor by @mattdangerw in #653
• Default compilation for BERT/RoBERTa classifiers by @jbischof in #695
• Add start/end token padding to GPT2Preprocessor by @chenmoneygithub in #704
• Don't install tf stable when building our nightly image by @mattdangerw in #711
• Add OPT Backbone and Tokenizer by @mattdangerw in #699
• Small OPT Doc-string Edits by @abheesht17 in #716
• Default compilation other classifiers by @Plutone11011 in #714
• Add BartTokenizer and BART Presets by @abheesht17 in #685
• Add an add_prefix_space Arg in BytePairTokenizer by @shivance in #715
• Opt presets by @mattdangerw in #707
• fix import of tensorflow_text in tf_utils by @sampathweb in #723
• Check for masked token in roberta tokenizer by @mattdangerw in #742
• Improve test coverage for special tokens in model tokenizers by @mattdangerw in #743
• Fix the sampler truncation strategy by @chenmoneygithub in #713
• Add ALBERT Conversion Script by @abheesht17 in #736
• Add FNet Conversion Script by @abheesht17 in #737
• Add BART Conversion Script by @abheesht17 in #739
• Pass Correct LayerNorm Epsilon value to TransformerEncoder in Backbones by @TheAthleticCoder in #731
• Improving the layer Description. by @Neeshamraghav012 in #734
• Adding ragged support to SinePositionEncoding by @apupneja in #751
• Fix trailing space by @mattdangerw in #755
• Adding an AlbertMaskedLM task + Fix Projection layer dimension in MaskedLMHead by @shivance in #725
• New docstring example for TokenAndPosition Embedding layer. by @Neeshamraghav012 in #760
• Add a note for TPU issues for deberta_v3 by @mattdangerw in #758
• Add missing exports to models API by @mattdangerw in #763
• Autogenerate preset table by @Cyber-Machine in #690
• Version bump to 0.5.0 by @mattdangerw in #767
• Adding a FNetMaskedLM task model and preprocessor by @apupneja in #740
• Add a DistilBertMaskedLM task model by @ADITYADAS1999 in #724
• Add cache support to decoding journey by @chenmoneygithub in #745
• Handle [MASK] token in DebertaV3Tokenizer by @abheesht17 in #759
• Update README for 2.4.1 release by @mattdangerw in #757
• Fix typo in test docstring by @jbischof in #791
• Fixed Incorrect Links for FNet and DeBERTaV3 models by @Cyber-Machine in #793
• Patch 1 - doc-string spell fix by @atharvapurdue in #781
• Don't rely on core keras initializer config details by @mattdangerw in #802
• Simplify the cache decoding graph by @mattdangerw in #780
• Fix Fenced Doc-String #782 by @atharvapurdue in #785
• Solve #721 Deberta masklm model by @Plutone11011 in #732
• Add from_config to sampler by @mattdangerw in #803
• BertMaskedLM Task Model and Preprocessor by @Cyber-Machine in #774
• Stop generation once en...