Skip to content

Latest commit

 

History

History
16 lines (10 loc) · 862 Bytes

README.md

File metadata and controls

16 lines (10 loc) · 862 Bytes

SafeChat

NLP powered anti toxicity chatroom that automatically detects and de-toxifies messages containing profanities and other offensive content, the classification is performed using a fine-tuned BERT model and detoxification is carried out by a fine-tuned BART model, both models use the paradetox dataset but in different structural formats.

Non Toxic Example :

Untitled Untitled

Toxic Example :

Untitled Untitled