Ru Toxicity Int Ml: Russian Text Comments with Multilabel Toxicity Annotations

A dataset for training models to detect toxic comments in Russian. It supports multilabel classification where a single text can belong to multiple toxicity categories simultaneously. The dataset was created by Gloomreach and was last updated on July 16, 2026.

Use Cases

• Train multilabel toxicity classifiers based on the three independent binary labels (profanity, threat, illegal).
• Benchmark models for detecting nuanced forms of harmful content in Russian text.
• Analyze the co-occurrence of different toxicity categories within single comments.
• Develop fine-grained content moderation filters for social media or forums.

Strengths

• Defines three specific, independent toxicity categories (profanity, threat, illegal) for nuanced annotation.
• Designed for multilabel classification, allowing a single text to have multiple labels, which reflects real-world complexity.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

huggingface

Freshness

Last updated 2026-07-16 22:01:16; freshness should be verified.