0.16.0

Changed

• Improve handling of optional dependencies

Fixed

• Fix model path changed since speechbrain v1.0

0.15.0

BREAKING CHANGES

• Make iamsystem an optional dependency

Fixed

• Add notice for downloading example documents
• Warn if dot is unavailable when displaying provenance graph
• Require typing-extensions >= 4.6.0

0.14.1

Fixed

• Add NER benchmark to cookbook

0.14.0

Changed

• Backport or use itertools.batched from Python 3.12

Fixed

• Use fork of mtsamplesFR under medkit-lib
• Fix returned value in batching utility

0.13.1

Fixed

• Use ISO 8601 timestamp for model checkpoint paths
• Fix test of iamsystem matcher on Python 3.12

0.13.0

Added

• Add nlstruct-based entity matcher

Changed

• Improve robustness of PASpeakerDetector
• Allow to specify model output language with HFTranscriber

Fixed

• Use link to new repository
• When parsing BRAT, preserve leading space in entities
• Replace unidecode by anyascii

0.12.0

Changes

• Document attributes are now supported (both for text and audio) and are added/accessed the same way as annotations attributes:

  doc.attrs.add(Attribute(label="type", value="report"))
  doc.attrs.get(label="type")

• Brat Input and Output converters can now load and save UMLS CUIs stored in notes

• the Trainer now saves both the last checkpoint and the best checkpoint, instead of only the last checkpoint

• medkit is now compatible with the latest (0.9) EDS-NLP

• most operations loading models from HuggingFace can now receive an authentication token (useful to access private repositories)

• new from_dir()/from_file() helper methods added to TextDocument/AudioDocument

• new text classification, audio diarization and audio transcription metrics

• support for remapping entity labels in Seq2SeqEvaluator (useful when predicted and reference label do not match exactly)

• custom attributes (DateAttribute, UMLSNormAttribute) don't have None as a value anymore

• easier initialization of PASpeakerDetector

• many bugfixes