All notable changes to this project will be documented in this file.
0.15.0 - 2020-03-10
- Remove support of Python2.7 #123
0.14.6 - 2020-01-14
- Support for python3.8 #121
0.14.5 - 2019-08-20
- Fix issue with intents filter when dataset has not enough data #118
0.14.4 - 2019-06-18
- Update dependencies
0.14.3 - 2019-05-10
- Add optional parameter
intents_filter
to metrics APIs #115
0.14.2 - 2019-03-21
- Number of exact parsings
- Possibility to provide out-of-domain utterances
- Logging
- Hanging issue when using multiple workers, when one job returns a non-zero exit code
0.14.1 - 2019-01-07
- Support for new NLU output format
- Bug with None intent when computing average metrics
0.14.0 - 2018-11-13
- Possibility to use parallel workers
- Seed parameter for reproducibility
- Average metrics for intent classification and slot filling
0.13.0 - 2018-07-25
- Crash while computing metrics when either actual or predicted intent is unknown
- APIs depending implicitely on Snips NLU:
compute_cross_val_nlu_metrics
compute_train_test_nlu_metrics
- Use flexible version specifiers for dependencies
0.12.0 - 2018-03-29
- F1 scores for intent classification and entity extraction
- Confusion matrix
- New option to exclude slot metrics in the output
- Samples