russian_toxicity_classifier
| Property | Value |
|---|---|
| Parameter Count | 178M |
| License | OpenRAIL++ |
| Base Model | DeepPavlov/rubert-base-cased-conversational |
| Accuracy | 97% |
What is russian_toxicity_classifier?
The russian_toxicity_classifier is a BERT-based model specifically designed for detecting toxic content in Russian text. Built upon the DeepPavlov's conversational RuBERT, this model has been fine-tuned on a comprehensive dataset merged from 2ch.hk and ok.ru sources, making it particularly effective for Russian language toxicity detection.
Implementation Details
The model utilizes the BERT architecture with 178M parameters and has been trained on a carefully curated dataset split into 80-10-10 proportions for training, development, and testing. It achieved impressive metrics with 0.97 weighted average F1-score on the test dataset.
- Pre-trained base: DeepPavlov/rubert-base-cased-conversational
- Training data: Merged dataset from 2ch.hk and ok.ru
- Evaluation metrics: 98% precision for non-toxic, 94% for toxic content
Core Capabilities
- Binary classification of Russian text toxicity
- High accuracy in distinguishing between toxic and non-toxic content
- Optimized for Russian language processing
- Easy integration with the Transformers library
Frequently Asked Questions
Q: What makes this model unique?
This model stands out for its specialized focus on Russian language toxicity detection, combining data from multiple sources and achieving high accuracy (97%) in classification tasks. It's particularly valuable for content moderation in Russian-language contexts.
Q: What are the recommended use cases?
The model is ideal for content moderation systems, social media platforms, and online communities where Russian language content needs to be monitored for toxic behavior. It can be easily integrated into existing systems using the Transformers library.
