| Property | Value |
|---|---|
| License | Apache 2.0 |
| Paper | Deep Learning Models for Multilingual Hate Speech Detection |
| Downloads | 127,719 |
| Framework | PyTorch, JAX |
What is dehatebert-mono-english?
dehatebert-mono-english is a specialized hate speech detection model fine-tuned on monolingual English data using the multilingual BERT architecture. Developed by Hate-speech-CNERG, it achieved a validation score of 0.726030 with a learning rate of 2e-5, making it particularly effective for identifying harmful content in English text.
Implementation Details
The model is built upon the multilingual BERT architecture and specifically optimized for English language processing. It leverages transformer-based technology and was trained with various learning rates to find the optimal performance point.
- Fine-tuned on English-only dataset for specialized hate speech detection
- Optimized learning rate of 2e-5 for best performance
- Built on multilingual BERT architecture
- Implements both PyTorch and JAX frameworks
Core Capabilities
- Monolingual hate speech detection in English
- High accuracy with 72.6% validation score
- Robust text classification for content moderation
- Inference endpoint support for deployment
Frequently Asked Questions
Q: What makes this model unique?
This model's uniqueness lies in its specialized focus on English-language hate speech detection, achieving high accuracy through monolingual training, unlike multilingual alternatives. Its validation score of 72.6% demonstrates its effectiveness in this specific task.
Q: What are the recommended use cases?
The model is ideal for content moderation systems, social media platforms, and online communities requiring automated hate speech detection in English content. It's particularly suitable for applications requiring high-accuracy hate speech identification with dedicated English language support.