datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
measuring-hate-speech
Dataset card for Measuring Hate Speech
This is a public release of the dataset described in Kennedy et al. (2020) and Sachdeva et al. (2022), consisting of 39,565 comments annotated by 7,912 annotators, for 135,556 combined rows. The primary outcome variable is the "hate speech score" but the 10 constituent ordinal labels (sentiment, (dis)respect, insult, humiliation, inferior status, violence, dehumanization, genocide, attack/defense, hate speech benchmark) can also be treated as… See the full description on the dataset page: https://huggingface.co/datasets/ucberkeley-dlab/measuring-hate-speech.hate_speech_offensive
Dataset Card for [Dataset Name]
Dataset Summary
An annotated dataset for hate speech and offensive language detection on tweets.
Supported Tasks and Leaderboards
[More Information Needed]
Languages
English (en)
Dataset Structure
Data Instances
{
"count": 3,
"hate_speech_annotation": 0,
"offensive_language_annotation": 0,
"neither_annotation": 3,
"label": 2, # "neither"
"tweet": "!!! RT @mayasolovely: As a woman you… See the full description on the dataset page: https://huggingface.co/datasets/tdavidson/hate_speech_offensive.hateXplain_filteredhate_speech_pl
Dataset Card for HateSpeechPl
Dataset Summary
The dataset was created to analyze the possibility of automating the recognition of hate speech in Polish. It was collected from the Polish forums and represents various types and degrees of offensive language, expressed towards minorities.
The original dataset is provided as an export of MySQL tables, what makes it hard to load. Due to that, it was converted to CSV and put to a Github repository.
Supported Tasks… See the full description on the dataset page: https://huggingface.co/datasets/community-datasets/hate_speech_pl.STCALIR-DatasetVOZ-HSD-Hate-Comments
Dataset Card for Dataset Name
A subset of VOZ-HSD dataset, consisting of only hate comments (labels: '1').For more information on the original dataset:
https://huggingface.co/datasets/tarudesu/VOZ-HSD.
NOTE: The original dataset is labeled automatically using fine-tuned ViSoBERT-HSD.
Dataset Details
Dataset Description
Curated by: [More Information Needed]
Funded by [optional]: [More Information Needed]
Shared by [optional]: [More Information Needed]… See the full description on the dataset page: https://huggingface.co/datasets/gamusa/VOZ-HSD-Hate-Comments.HatEvalThis is the dataset for SemEval-2019 Task 5: Multilingual Detection of Hate Speech Against Immigrants and Women in Twitter (HatEval).
CodaLab page
Shared Task Report
Shared Task
Hate Speech is commonly defined as any communication that disparages a person or a group on the basis of some characteristic such as race, color, ethnicity, gender, sexual orientation, nationality, religion, or other characteristics. Given the huge amount of user-generated contents on the Web, and in… See the full description on the dataset page: https://huggingface.co/datasets/valeriobasile/HatEval.HateOWS-dataset-LREC2026
OWS Data for LREC 2026
This repository contains the data resources used in the paper:
Toward Generalized Cross-Lingual Hateful Language Detection with Web-Scale Data and Ensemble LLM Annotations (LREC-COLING 2026).
It provides:
multilingual OpenWebSearch (OWS) raw corpora (DEU, ENG, SPA, VIE)
an annotated subset with four base LLM annotators + three ensemble labeling strategies
a LightGBM-derived dataset (training probabilities / features)
the 16 human-labelled datasets used for… See the full description on the dataset page: https://huggingface.co/datasets/danghaidang-passau/HateOWS-dataset-LREC2026.kor-hate-sentence
SJ-Donald/kor-hate-sentence
SJ-Donald/kor-hate-sentence is merged dataset from fllow
Datasets
smilegate-ai/kor_unsmile
korean-hate-speech
Curse-detection-data
korean-malicious-comments-dataset
Merge datasets from above and drop duplicates.
How to use
from datasets import load_dataset
ds = load_dataset("SJ-Donald/kor-hate-sentence")
print(ds)
DatasetDict({
train: Dataset({
features: ['문장', 'hate', 'clean', 'labels'],
num_rows: 29328… See the full description on the dataset page: https://huggingface.co/datasets/SJ-Donald/kor-hate-sentence.hateday
HateDay
🆕 Update (Nov 9, 2025):A new and expanded version of the dataset has been released.
Size: 540K total annotated tweets
Labels: now include hate, offensive, or neutral, and whether hate is violent or non-violent
Sampling: includes both
a regular random sample (n = 30K per language/country), and
a sample weighted by total engagement (n = 15K per language/country)
Quality: higher annotation quality — each hate example was manually reviewed for correctness
New… See the full description on the dataset page: https://huggingface.co/datasets/manueltonneau/hateday.HateBenchSet
HateBenchSet
This is the HateBenchSet dataset curated in the USENIX 2025 paper HateBench: Benchmarking Hate Speech Detectors on LLM-Generated Content and Hate Campaigns.
It contains 7,838 samples across 34 identity groups, generated by six LLMs, i.e., GPT-3.5, GPT4, Vicuna, Baichuan2, Dolly2, and OPT.
These samples are manually labeled, resulting in 3,641 hate samples and 4,197 non-hate samples.
Data Structure
Column
Description
model
Model used to generate… See the full description on the dataset page: https://huggingface.co/datasets/TrustAIRLab/HateBenchSet.ucberkeley-measuring-hate-speech-3classThis is modified version of UCBerkley DLab dataset, converted to represtent harmfull level in 3 classes:
SAFE - hate_speech_score < -1
BORDERLINE - -1 < hate_speech_score < 0.5
HARMFUL - hate_speech_score >= 0.5
Text and harmful level classification was created by UCBerkley DLab Team! I'm not creator of this data - i'm only converted it into classes!
implicit_hate-himeasuring-hate-speech
Dataset card for Measuring Hate Speech
This is a public release of the dataset described in Kennedy et al. (2020) and Sachdeva et al. (2022), consisting of 39,565 comments annotated by 7,912 annotators, for 135,556 combined rows. The primary outcome variable is the "hate speech score" but the 10 constituent ordinal labels (sentiment, (dis)respect, insult, humiliation, inferior status, violence, dehumanization, genocide, attack/defense, hate speech benchmark) can also be treated as… See the full description on the dataset page: https://huggingface.co/datasets/kush5699/measuring-hate-speech.neuronovo-utc-measuring-hate-speechThe_Gab_Hate_Corpus_ghc_test_originalmeasuring-hate-speech-simpleSimplified version of Measuring Hate Speech using our custom class thresholds.
Original dataset: https://huggingface.co/datasets/ucberkeley-dlab/measuring-hate-speech
measuring-hate-speech
Dataset card for Measuring Hate Speech
This is a public release of the dataset described in Kennedy et al. (2020) and Sachdeva et al. (2022), consisting of 39,565 comments annotated by 7,912 annotators, for 135,556 combined rows. The primary outcome variable is the "hate speech score" but the 10 constituent ordinal labels (sentiment, (dis)respect, insult, humiliation, inferior status, violence, dehumanization, genocide, attack/defense, hate speech benchmark) can also be treated as… See the full description on the dataset page: https://huggingface.co/datasets/Shubhi324/measuring-hate-speech.The_Gab_Hate_Corpus_ghc_train_originalHateEval_with_emotions_diversityasia-owid-hate-crime-protections-lgbt
Hate Crime Protections Lgbt | Asia (Our World in Data)
🌏 1,715 observations · 49 Asia countries · 1991–2025 · Repackaged by Electric Sheep Asia
TL;DR
This dataset contains 1,715 observations of Hate Crime Protections Lgbt data across 49 Asia countries, spanning 1991–2025.
About the source
Source: Our World in Data
Publisher: Our World in Data
License: cc-by-4.0
Topic: Hate Crime Protections Lgbt
Geographic coverage
49 Asia… See the full description on the dataset page: https://huggingface.co/datasets/electricsheepasia/asia-owid-hate-crime-protections-lgbt.hate-speech18-es
Dataset Card for "hate_speech18-es"
More Information needed
measuring-hate-speech
Dataset card for Measuring Hate Speech
This is a public release of the dataset described in Kennedy et al. (2020) and Sachdeva et al. (2022), consisting of 39,565 comments annotated by 7,912 annotators, for 135,556 combined rows. The primary outcome variable is the "hate speech score" but the 10 constituent ordinal labels (sentiment, (dis)respect, insult, humiliation, inferior status, violence, dehumanization, genocide, attack/defense, hate speech benchmark) can also be treated as… See the full description on the dataset page: https://huggingface.co/datasets/aditya040594/measuring-hate-speech.neuronovo-utc-hate-speech18-sentenceshate-speech18
Dataset Card for "hate_speech18"
More Information needed
ucberkeley-dlab-measuring-hate-speech
Dataset Card for "ucberkeley-dlab-measuring-hate-speech"
More Information needed
targeted_hatespeech_entailment
Dataset Card for "targeted_hatespeech_entailment"
More Information needed
hateful_memes_fine_grained
Hateful Memes Fine-Grained Dataset
This dataset is a fine-grained extension of the widely used Hateful Memes dataset, designed to enable more nuanced analysis of harmful multimodal content. While the original dataset focuses on binary hatefulness classification, this extension introduces additional annotation dimensions capturing incivility and intolerance at a more granular level.
The dataset consists of a subset of 2,030 memes, each annotated independently by three annotators.… See the full description on the dataset page: https://huggingface.co/datasets/nils-herrmann/hateful_memes_fine_grained.slovak_hate_speechhatecheck-enriched
Dataset Card for "hatecheck-enriched"
More Information needed