Skip to content

Dataset · tdavidson

English tweets with hate speech and offensive language annotations

tdavidson/hate_speech_offensive

A classification dataset of English tweets annotated for hate speech, offensive language, and neutral text.

What's inside

It contains English tweet texts and annotator votes across three categories: hate speech, offensive language, and neutral messages. The dataset size ranges from 10k to 100k entries, presented in tabular and text formats with a final class label.

What you can do with it

  • Train or fine-tune a toxicity and content moderation classifier (e.g., based on BERT or RoBERTa)
  • Evaluate safety filter quality for English-language chatbots and social networks
Tasks
Text classification
Languages
English
Size
10K–100K rows
Downloads this month
2.9K
Likes
42

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("tdavidson/hate_speech_offensive")