v0.6.0

Changes

• Release 0.6.0 (#174) @ChristophAlt
• dont export core and fix readme examples (#173) @ArneBinder

v0.5.0

Changes

• Release 0.5.0 (#171) @ChristophAlt
• reoarganize pytorch_ie.core (#166) @ArneBinder
• add nox sessions: tests_no_local_datasets and tests_only_local_datasets (#170) @ArneBinder
• use Auto classes in readme (#165) @ArneBinder
• Fix dummy data folder name for NL dataset (#168) @ChristophAlt
• Add AutoTaskModule, AutoModel, AutoPipeline (#157) @ChristophAlt
• NER datasets (#162) @ChristophAlt

v0.4.0

Changes

• Release 0.4.0 (#146) @ChristophAlt
• Initial datasets setup (#155) @ChristophAlt
• Reworked TaskModule and Pipeline (#149) @ChristophAlt
• Make readme examples work in interactive interpreter (#145) @ChristophAlt
• Update main dependencies to latest version (#141) @ChristophAlt
• Drop python 3.8 support (#143) @ChristophAlt
• Prototype annotations, documents, and dataset (#130) @ArneBinder

v0.3.3

Changes

• Release 0.3.3 (#138) @ChristophAlt
• implement Model.test_step (#133) @ArneBinder

v0.3.2

Changes

• Release 0.3.2 (#131) @ChristophAlt
• Refactor datamodule (#121) @ArneBinder
• refactor conll2003 (#122) @ArneBinder
• fix #124 (#125) @GabrielKP
• add train test for transformer_token_classification (#115) @ArneBinder

🪲 Fixes

• exclude spans that are out of valid window (#128) @ArneBinder

v0.3.1

Changes

• Release 0.3.1 (#113) @ChristophAlt

🪲 Fixes

• Fix missing pass of is_training to encode_input (#108) @ArneBinder

v0.3.0

Changes

• Release 0.3.0 (#107) @ChristophAlt
• add prepare_documents (#102) @ArneBinder
• revert using the commit hash as prerelease identifier (#106) @ArneBinder
• use commit hash instead of date for pre-release identifier (#104) @ArneBinder

🔨 Refactoring

• Refactor/document typed (#95) @ArneBinder

v0.2.1

Changes

• Release 0.2.1 (#97) @ChristophAlt
• control if statistics are collected and shown (#85) @ArneBinder
• make it optional to include ill formed spans when predicting (#79) @ArneBinder
• Windowing for token classification (#72) @ArneBinder
• simplify transformer_token_classification_taskmodule (#71) @ArneBinder
• tests for TransformerTokenClassificationTaskModule (#67) @ArneBinder
• implement #63 (#65) @ArneBinder
• annotation getters default to None (#64) (#70) @ChristophAlt
• implement windowing for text classification based rel (#47) @ArneBinder
• add tests for TransformerRETextClassificationTaskModule (#58) @ArneBinder
• create tests from the examples that check the actual pipeline output (#59) @ArneBinder
• pipeline: cleanup argument handling (#57) @ArneBinder
• Fix issue #51 (#55) @ChristophAlt
• show pipeline progress (#48) @ArneBinder
• implement serialize_brat() and add test for that and load_brat() (#31) @ArneBinder
• Unittests for TransformerSpanClassificationModel (#38) @ChristophAlt
• partitioning for TransformerRETextClassificationTaskModule (#43) @ArneBinder
• Fix remaining mypy errors (#39) @ChristophAlt
• fix stupid bug in TransformerTokenClassificationTaskModule introduced… (#37) @ArneBinder
• fix a lot of typing in mdules/ and taskmodules/ (#33) @ArneBinder
• fix TaskEncodingDataset: use correct base type (#35) @ArneBinder
• fix typing for data/ and core/ (#32) @ArneBinder
• Unittests for Taskmodule (#34) @ChristophAlt
• [WIP] Add transformer seq2seq model and taskmodule (#28) @ChristophAlt
• add and improve type annotations to taskmodules (#27) @ArneBinder
• fix encode_target for self.single_sentence=True (#26) @ArneBinder
• fix signature of TransformerSpanClassificationTaskModule.encode_target (#24) @ArneBinder
• add general DataModule and brat dataset (#19) @ArneBinder
• use consistent function names in taskmodule and pipeline (#22) @ArneBinder
• remove inplace doc modification (#16) @ArneBinder

🪲 Fixes

• fix brat equivalence_relations (#84) @ArneBinder
• fix warnings for unaligened span (#82) @ArneBinder
• fix checks for required span annotations (#81) @ArneBinder
• fix decode (#66) @ArneBinder
• Fix brat serialization with newlines (#61) @ArneBinder
• fix label id creation for single_sentence=False in encode_input (#45) @ArneBinder
• fix issue #49 (#50) @ArneBinder

👷 Continuous Integration

• Properly configure coverage.py (#90) @ChristophAlt
• Add labeler and release-drafter to github workflow (#88) @ChristophAlt
• Prepare CI / CD pipeline for initial release (#76) @ChristophAlt

📚 Documentation

• Readme refactoring (#96) @ChristophAlt

🔨 Refactoring

• refactor example test documents (#89) @ArneBinder
• move utility functions into own package (#80) @ArneBinder

📦 Dependencies

• Switch to src layout (#92) @ChristophAlt