Dataset · ucberkeley-dlab
Annotated comment corpus for hate speech detection
ucberkeley-dlab/measuring-hate-speech
39,565 English comments with ratings from 7,912 annotators across 10 toxicity and hate speech scales.
What's inside
A table-text English dataset including 135,556 annotation rows for 39,565 unique comments rated by 7,912 annotators. In addition to the final hate speech index, each example is rated on 10 ordinal scales, including profanity, degradation, dehumanization, calls to violence, and sentiment.
What you can do with it
- Training multi-criteria toxicity classifiers and content moderation
- Studying subjectivity and annotator agreement in hate speech evaluation
- Tasks
- Text classification
- Languages
- English
- Size
- 100K–1M rows
- Downloads this month
- 9.3K
- Likes
- 56
Models trained on it
- StrandsAgents/strands-decider-2B-qwen3.5-v1-2610Text classification 25K
- StrandsAgents/strands-decider-2B-hobson-v21Text classification 15.3K
- StrandsAgents/strands-decider-2B-hobson-v19Text classification 7K
- dmis-lab/UltraBERT-Embed-150MSentence similarity 1.4K
- mradermacher/elf-1-12b-GGUF 753
- mradermacher/elf-1-31b-GGUF 721
- amazon/strands-decider-2B-qwen3.5-v1-2610Text classification 199
- LatentAILabs/yama-indicZero shot classification 171
How to load it
from datasets import load_dataset
ds = load_dataset("ucberkeley-dlab/measuring-hate-speech")