v0.22.0

Changes

• release 0.22.0 (#329) @ArneBinder

🚀 Features

• add model and taskmodule Interface classes (#328) @ArneBinder

v0.21.0

Changes

• release 0.21.0 (#326) @ArneBinder
• set minimum transformers version (#325) @ArneBinder
• update datasets and transformers (#323) @ArneBinder
• Update pytorch-lightning (#322) @ArneBinder
• update (dev) dependencies (#321) @ArneBinder

🪲 Fixes

• set minimum datasets version (#324) @ArneBinder

v0.20.0

Changes

• release 0.20.0 (#320) @ArneBinder
• integrate eval_counts_for_layer() into F1Metric.calculate_counts() (#318) @ArneBinder

🚀 Features

• tokenize_document: set text_pair if text is already provided (#317) @ArneBinder
• conversion of text based document to token based and tokenize document (#315) @ArneBinder

🪲 Fixes

• fix statistic value aggregation for empty result lists (#316) @ArneBinder

v0.19.0

Changes

• release 0.19.0 (#314) @ArneBinder

🚀 Features

• aggregate the result of DocumentStatistic (#313) @ArneBinder
• implement DocumentStatistics (#312) @ArneBinder

v0.18.1

Changes

• release 0.18.1 (#311) @ArneBinder

🪲 Fixes

• Fix Document.add_all_annotations_from_other() (#310) @ArneBinder

v0.18.0

Changes

• release 0.18.0 (#309) @ArneBinder
• set compare=False for the score of built-in annotations (#307) @ArneBinder

💥 Breaking Changes

• include non-comparison annotion fields in the _id calculation (#305) @ArneBinder

🚀 Features

• implement DocumentMetric and F1Metric (#308) @ArneBinder
• implement Document.add_all_annotations_from_other (#306) @ArneBinder

v0.17.1

Changes

• release 0.17.1 (#301) @ArneBinder

🚀 Features

• implement DatasetDict.load_dataset() (#300) @ArneBinder

🚨 Testing

• add tests for TransformerTokenClassificationTaskModule (#299) @ArneBinder

v0.17.0

Changes

• release 0.17.0 (#298) @ArneBinder

🚀 Features

• implement DatasetDict (#296) @ArneBinder
• implement target getters for BaseAnnotationList (#297) @ArneBinder

v0.16.0

Changes

• release 0.16.0 (#293) @ArneBinder
• token classification taskmodule: skip docs with overlapping spans (#284) @ArneBinder

💥 Breaking Changes

• rework TransformerRETextClassificationTaskModule (#283) @ArneBinder

🚀 Features

• BaseAnnotationList.clear() returns the contained annotations (#291) @ArneBinder
• add Annotation.copy() (#292) @ArneBinder
• TextBasedDocument and TokenBasedDocument types (#290) @ArneBinder
• N-ary Relation classification (#287) @ArneBinder
• individual document type per config name (#288) @ArneBinder
• individual Huggingface base dataset per config (#263) @ArneBinder
• BinaryRelation can have any annotation as head and tail (#286) @ArneBinder
• add Document.copy() (#285) @ArneBinder

🪲 Fixes

• fix relative paths in base dataset builder (#289) @ArneBinder

🚨 Testing

• add tests for SimpleTransformerTextClassificationTaskModule (#282) @ArneBinder

🔨 Refactoring

• rework TransformerRETextClassificationTaskModule (#283) @ArneBinder

v0.15.1

Changes

• No changes