v1.23.1: Patch release

• Fix doc build by @regisss in #2050
• Don't hardcode the logger level to INFO let users set TRANSFORMERS_VERBOSITY by @tomaarsen in #2047
• Add workflow to mark issues as stale by @regisss in #2051
• Fix onnx export when transformers >= v4.45 (impacting sentence-transformers and timm models) by @echarlaix in #2053 and #2054

v1.23.0: ORTDiffusionPipeline, transformers v4.45

ONNX Runtime Diffusion pipeline

Adding ORTDiffusionPipeline to simplify diffusers model loading by @IlyasMoutawwakil in #1960 and #2021

  model_id = "runwayml/stable-diffusion-v1-5"
- pipeline = ORTStableDiffusionPipeline.from_pretrained(model_id, revision="onnx")
+ pipeline = ORTDiffusionPipeline.from_pretrained(model_id, revision="onnx")
  image = pipeline("sailing ship in storm by Leonardo da Vinci").images[0]

Transformers v4.45

Transformers v4.45 support by @echarlaix in #2023 and #2045

Subfolder

Remove the restriction for the model's config to be in the model's subfolder by @echarlaix in #2044

New Contributors

• @tcsavage made their first contribution in #1965
• @yuanwu2017 made their first contribution in #2003
• @h3110Fr13nd made their first contribution in #2031
• @glegendre01 made their first contribution in #2033
• @rbrugaro made their first contribution in #2027

Full Changelog: v1.22.0...v1.23.0

v1.22.0: transformers 4.44 compatibility, bugfixes

What's Changed

• Fix sentence transformers modeling patching for export by @echarlaix in #1936
• Update optimum intel extra by @echarlaix in #1935
• Update Habana extra by @regisss in #1937
• Remove inplace op in mistral patcher by @IlyasMoutawwakil in #1938
• Fix forward bug in ORTModelForFeatureExtraction by @moria97 in #1941
• Deprecate ORTModel class by @IlyasMoutawwakil in #1939
• Remove warning by @echarlaix in #1945
• Clip vision model onnx export by @fxmarty in #1920
• Add export test for swin with shifted windows by @echarlaix in #1942
• Refactor diffusers tasks by @IlyasMoutawwakil in #1947
• Fix optimizer's command line reading by @idruker-cerence in #1961
• Fix unmask_unattended_patched signature by @fxmarty in #1963
• Fix undefined variable in library name inference by @IlyasMoutawwakil in #1964
• Fix gpt bigcode ONNX export for transformers<4.39.0 by @echarlaix in #1973
• Support transformers 4.43 by @IlyasMoutawwakil in #1971
• chore(ci): migrate runner configuration in GitHub workflows by @XciD in #1978
• Fix typos in quantization.mdx by @aldakata in #1989
• Update Habana extra in setup.py by @regisss in #1991
• Follow up the diffusers task refactoring by @JingyaHuang in #1999
• Transformers 4.44 support by @IlyasMoutawwakil in #1996
• Modify token classification processor default dataset args by @echarlaix in #2005
• Fix TFLite tests by @IlyasMoutawwakil in #2007
• Fix attribute name from inputs_names to input_names by @J4BEZ in #2010
• Fix typo in BetterTransformer's overview docs by @ftnext in #2015
• Apply deprecated evaluation_strategy by @muellerzr in #1819
• Update transformers imports for deepspeed and is_torch_xla_available by @Rohan138 in #2012
• Add quanto install and instructions by @dacorvo in #1976

New Contributors

• @moria97 made their first contribution in #1941
• @XciD made their first contribution in #1978
• @zhenglongjiepheonix made their first contribution in #1933
• @aldakata made their first contribution in #1989
• @J4BEZ made their first contribution in #2010
• @ftnext made their first contribution in #2015
• @muellerzr made their first contribution in #1819
• @Rohan138 made their first contribution in #2012

Full Changelog: v1.21.4...v1.22.0

v1.21.4: Patch release

• Update Habana extra in setup.py by @regisss in #1991

Full Changelog: v1.21.3...v1.21.4

v1.21.3: Patch release

• Deprecate ORTModel class by @IlyasMoutawwakil in #1939
• Remove warning by @echarlaix in #1945
• Fix optimizer's command line reading by @idruker-cerence in #1961
• Fix unmask_unattended_patched signature by @fxmarty in #1963
• Fix gpt bigcode ONNX export for transformers<4.39.0 by @echarlaix in #1973
• Support transformers 4.43 by @IlyasMoutawwakil in #1971

Full Changelog: v1.21.2...v1.21.3

v1.21.2: Patch release

• Remove inplace op in mistral patcher by @IlyasMoutawwakil in #1938
• Fix ORTModelForFeatureExtraction modeling by @moria97 in #1941

Full Changelog: v1.21.1...v1.21.2

v1.21.1: Patch release

• Fix sentence transformers model patching by @echarlaix in #1936
• Update Intel extra by @echarlaix in #1935
• Update Habana extra by @regisss in #1937

Full Changelog: v1.21.0...v1.21.1

v1.21.0: many bugfixes, transformers 4.42 compatibility

What's Changed

• ORTOptimizer for the model type Segformer by @zachmayer in #1820
• fix: device consistence by @Daya-Jin in #1891
• Allow optimum to discover and load subpackages by @dacorvo in #1894
• feat(ci): add trufflehog secrets detector by @McPatate in #1899
• fix(ci): remove unnecessary permissions by @McPatate in #1904
• Remove read token by @fxmarty in #1903
• Remove dataset with restrictive license by @echarlaix in #1910
• Fix Windows and onnx dtype compatibility by @IlyasMoutawwakil in #1886
• Deprecated use_auth_token by @IlyasMoutawwakil in #1837
• Add redirection for optimum intel doc by @echarlaix in #1918
• Read use_external_data_format from ORTConfig file by @idruker-cerence in #1917
• Pin numpy v1 for onnxruntime by @IlyasMoutawwakil in #1921
• Fix GPTQ CI by @IlyasMoutawwakil in #1878
• Fix code quality by @IlyasMoutawwakil in #1928
• Fix incorrect names for usage blenderbot for causallm by @eaidova in #1887
• Fixed bug key error "last_hidden_state" by @satishsilveri in #1674
• Support transformers 4.42 by @fxmarty in #1929

New Contributors

• @zachmayer made their first contribution in #1820
• @Daya-Jin made their first contribution in #1891
• @dacorvo made their first contribution in #1894
• @McPatate made their first contribution in #1899
• @idruker-cerence made their first contribution in #1917
• @satishsilveri made their first contribution in #1674

Full Changelog: v1.20.0...v1.21.0

v1.20.0: VITS, Phi-3 ONNX export

Extended ONNX export

• VITS ONNX export by @echarlaix in #1607
• Phi-3 ONNX export by @JingyaHuang in #1870
• Add Phi-3 normalized config by @kunal-vaishnavi in #1841
• Add Phi-3 small normalized config by @JingyaHuang in #1864

Other changes and bugfixes

• Bump transformers version by @echarlaix in #1824

• Remove call to apt update before apt purge in the main doc build workflow by @regisss in #1830

• Update github workflows by @echarlaix in #1829

• Remove bad PPA in main doc build workflow by @regisss in #1831

• Fix TPU doc build by @regisss in #1834

• Fix sentence transformers models infer library by @echarlaix in #1832

• Fix random initialization of bias when using GPTQ quantization with models without bias by @B-201 in #1827

• Update the Transformers dependency in the Habana extra by @regisss in #1851

• Make stable diffusion unet and vae number of channels static by @eaidova in #1840

• Fix compatibility with transformers v4.41.0 for ONNX by @echarlaix in #1860

• Fix FX CI by @IlyasMoutawwakil in #1866

• Fix Utils CI by @IlyasMoutawwakil in #1867

• Fix BT CI by @IlyasMoutawwakil in #1872

• Fix ORTConfig loading by @mr-sarthakgupta in #1879

• Update ORT doc for ROCM 6.0 by @mht-sharma in #1862

• Fix ort config instantiation (from_pretrained) and saving (save_pretrained) by @IlyasMoutawwakil in #1865

• Fix ORT CI by @IlyasMoutawwakil in #1875

• Update optimum intel extra by @echarlaix in #1882

• Bump transformers version for neuron extras by @JingyaHuang in #1881

New Contributors

• @B-201 made their first contribution in #1827
• @mr-sarthakgupta made their first contribution in #1879

Full Changelog: v1.19.0...v1.20.0

v1.19.2: Patch release

• Update the Transformers dependency in the Habana extra #1851 @regisss

Full Changelog: v1.19.1...v1.19.2