sismetanin/rubert-toxic-pikabu-2ch
Sismetanin/rubert-toxic-pikabu-2ch is a machine learning model.
About sismetanin/rubert-toxic-pikabu-2ch
Fine-tuned RuBERT-Toxic achieved F1 = 92.20%, demonstrating the best classification score . Kaggle Russian Language Toxic Comments Dataset is the collection of Russian-language annotated comments from 2ch and Pikabu . It consists of 14412 comments, where 4826 texts were labelled toxic, and 9586 were labelled as non-toxic . The average length of comments is ~175 characters; the minimum length is 21, and the maximum is 7403 . We fine-tuneed two versions of Multilingual Universal Sentence Encoder (M-USE), Multilingual Bidirectional Encoder Representations from Transformers (M) and RuBER,