✨ A "state-of-the-art NLP model" trained on 150,000 twitter comments using Bi-Directional LSTMs and 1-D Convolutions , working as the brain of a discord bot to morderate chat rooms and kick/ban abusive users.
https://drive.google.com/file/d/1E0os4yYhm-YNocMgISNoXMPuKbB2GgMz/view?usp=sharing
- Model Trained on 150K real-world comment texts after being preprocessed with regex.
- Use of Bi-Directional LSTM and Conv1D with Spatial Dropouts.
- Use of tensorflow-lite api in model quantization and retaining of its efficiency.
- Deployment of the model for a widely used case of Moderation.
👉 Screenshot of the model in working on a discord channel :-
P.S :-
- The requirements.txt file is present in the "deployment scripts" folder of this repo.
- We will be happy to see your sugesstions on further addition of new features to the bot.
- Please create relevant issue in the Issues section.
Team Gray-Interface ✖️ HackSlash