Created by: nastyachizhikova
FAQ refactored:
- old MIPT FAQ dataset replaced with English and Russian subsets of the MASSIVE dataset
- fasttext_logreg model added
- language versatility added to the config with integration of the LANGUAGE meta variable
Removed models:
- fasttext_avg_autofaq
- fasttext_tfidf_autofaq
- tfidf_autofaq
- tfidf_logreg_autofaq
- tfidf_logreg_en_faq
- tfidf_vectorizer
- tfidf_logreg_autofaq_psearch
- cv_tfidf_autofaq
removed components:
other library improvements:
- few-shot sampling support added to basic_classification_iterator
- ru_tokenizer component replaced with stream_spacy_tokenizer
-
PyMorphyremoved from requirements