Russian Text Detoxification & Paraphrase
12 repos
NLP systems for detecting and removing toxic language from Russian text, primarily using paraphrase generation and style transfer techniques. The cluster centers on competition datasets and solutions from the RUSSE detoxification shared tasks, alongside related research in conditional text generation for Russian. Repositories here combine Jupyter notebooks with Python implementations for training and evaluating models that rewrite toxic content into non-toxic alternatives while preserving semantic meaning.
Jupyter Notebook · 4
Python · 3
nlp ·50
russian-language ·50
style-transfer ·50
paraphrase-generation ·11
paraphrase ·11
conditional-text-generation ·11
textdetox ·4
