Russian Multi-Task Toxicity Text Dataset for Profanity, Threats, and Illegal Content

A Russian-language dataset for training a multi-task toxicity classifier across three independent binary categories: profanity, threats, and illegal requests. The dataset, created by AtesiT, was last updated on July 15, 2026. It is designed for a multi-label classification task where a single text can belong to multiple categories simultaneously.

Use Cases

• Training multi-label classifiers to detect profanity in Russian text based on the described 'profanity' category.
• Developing threat detection models for user safety based on the described 'threat' category.
• Identifying texts related to illegal activities for automated filtering based on the described 'illegal' category.
• Benchmarking the performance of toxicity detection models on a multi-task Russian-language dataset.

Strengths

• Defines three specific, independent binary categories for toxicity classification: profanity, threats, and illegal content.
• Designed for a multi-label task, reflecting real-world complexity where texts can belong to multiple categories.

Limitations

• Row count, column definitions, and sample data are unknown, limiting suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• Last updated 2026-07-15 20:18:45; freshness should be verified.

Provenance

Source

huggingface, author AtesiT

Freshness

Last updated 2026-07-15 20:18:45.

Geography

Russian-language texts