Skip to content

A Russian Sentiment analysis BERT model with 27 emotion classes

Notifications You must be signed in to change notification settings

stas1f1/RuSentiClassificator

Repository files navigation

SentiChallenge

A Russian Sentiment analysis BERT model, fine-tuned for SA task from RuBert-cased-conversational with Sequence classification head, trained on GoEmotions dataset which was translated to russian with help of Yandex Cloud Translate API.

Model is able to recognise un-emotional speech and 27 categories of emotions

Pre-trained model weights available here

Examples:

Class recognition success rates:

About

A Russian Sentiment analysis BERT model with 27 emotion classes

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published