rubert-base-cased-mcn

Normalization model, based on rubert, for linking phrases to their MedDRA concepts in russian. F1-micro of this model is 71.34 on the 4th fold of the RDRS corpus of russian internet drug reviews.

The use of the weights of the current model and the calculation of accuracies on the laid out RDRS corpus is contained in our repository on our repo.