Created by: ayeffkay
This PR deals with scores update after re-training Distil* models on full conversational data (previous version was train on part of data).
-
overview.rst
updated with new scores (for distilrubert-tiny and distilrubert-base) - hyperparameters changed in
ner_rus_convers_distilrubert*.json
,paraphraser_convers_distilrubert*.json
,rusentiment_convers_distilrubert*.json
,squad_ru_convers_distilrubert*.json