russian inappropriate messages

Introduction

The "Russian Inappropriate Messages" model is designed to classify inappropriate messages in the Russian language. It focuses on detecting inappropriateness as a derivative of toxicity, acting as an additional filtering layer after standard toxicity filtration. The model uses datasets that are publicly available on GitHub and Kaggle.

Architecture

The model is built using the BERT architecture, leveraging PyTorch, TensorFlow, and JAX frameworks. It is optimized for text classification tasks and is suitable for filtering inappropriate content in Russian.

Training

The model was trained on a dataset comprising Russian messages classified as inappropriate. It achieved high precision and recall rates, with an accuracy of 89% on the test set. The training process was based on samples labeled with 100% confidence, focusing on precise classification metrics.

Guide: Running Locally

1. Set up Environment: Install necessary libraries such as PyTorch or TensorFlow.
2. Download Model: Obtain the model files from the Hugging Face repository.
3. Load the Model: Use a compatible library like transformers to load the model.
4. Inference: Run inferences on your text data to classify inappropriate messages.

For enhanced performance, consider using cloud GPUs from providers like AWS or Google Cloud.

License

This model is licensed under the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License. You can view the license details here.