- Gensim migration from 3.8 to >=4.0.0
- Fixed unwanted sorting of documents
- Fixed gensim version (#22)
- Added per-topic kl-uniform significance
- Handling multilabel classification
- Fixed preprocessing when dataset is not split (#17)
- Added regularization hyperparameter to NMF_scikit
- Added similarity metrics
- Fixed handling of stopwords in preprocessing
- Fixed coherence and diversity metrics
- Added new metrics tests
- Fixed CTM training when only training dataset is used
- Dashboard bugs fixed
- Minor bug fixes
- Added new tests for TM training
- Added parameter num_samples to CTM, NeuralLDA and ProdLDA
- Bug fix AVITM
- Bug fix info dataset
- Tomotopy LDA's implementation should work now
- bug fix dataset download
- CTM is no longer verbose
- New classification metrics
- Vocabulary downloader fix
- Dataset downloader fix
- New metrics initialization (do not support dictionaries as input anymore)
- Optimization, dataset and dashboard bug fixes
- Refactoring
- Updated README and documentation
- Dataset preprocessing produces also an indexes.txt file containing the indexes of the documents
- Eval metrics bug fixes
- BBC news added in the correct format
- Bug fixes
- New dataset format
- First release on PyPI.