T

Toxic Bert

Developed by unitary
A toxicity comment classification system based on PyTorch Lightning and Hugging Face Transformers, capable of detecting various types of harmful online content
Downloads 609.57k
Release Time : 3/2/2022

Model Overview

Detoxify is a collection of pre-trained models specifically designed to identify and classify toxic content in online comments, including threats, insults, hate speech, and more.

Model Features

Multi-task Toxicity Detection
Capable of detecting multiple types of toxicity simultaneously, including threats, insults, hate speech, etc.
Bias Control
Specially optimized to reduce misjudgments against specific identity groups
Multilingual Support
Supports toxicity detection in 7 languages
Ready-to-Use Model
Provides pre-trained models that can be used without additional training

Model Capabilities

Text toxicity classification
Multi-label classification
Multilingual text processing
Bias detection

Use Cases

Content Moderation
Social Media Comment Filtering
Automatically identifies and flags harmful comments for moderator review
Can detect multiple types of toxicity with accuracy exceeding 93%
Research Analysis
Online Speech Research
Analyzes the distribution of toxic content across different platforms or communities
Provides fine-grained toxicity classification data
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
© 2025AIbase