datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
Safety-Toxicity-Detection
SEA Toxicity Detection
SEA Toxicity Detection evaluates a model's ability to identify toxic content such as hate speech and abusive language in text. It is sampled from MLHSD for Indonesian, TTD for Thai, and ViHSD for Vietnamese.
Supported Tasks and Leaderboards
SEA Toxicity Detection is designed for evaluating chat or instruction-tuned large language models (LLMs). It is part of the SEA-HELM leaderboard from AI Singapore.
Languages
Indonesian (id)
Thai… See the full description on the dataset page: https://huggingface.co/datasets/aisingapore/Safety-Toxicity-Detection.or-bench-toxic-all
OR-Bench: An Over-Refusal Benchmark for Large Language Models
This dataset constains highly toxic prompts, use with caution!!!
Please see our demo at HuggingFace Spaces.
Overall Plots of Model Performances
Below is the overall model performance. X axis shows the rejection rate on OR-Bench-Hard-1K and Y axis shows the rejection rate on OR-Bench-Toxic. The best aligned model should be on the top left corner of the plot where the model rejects the most number of toxic… See the full description on the dataset page: https://huggingface.co/datasets/bench-llms/or-bench-toxic-all.from-one-to-many-toxicity-mitigation
From One to Many: Expanding the Scope of Toxicity Mitigation in Language Models
[arxiv][code][data]
Data accompanying the paper "From One to Many: Expanding the Scope of Toxicity Mitigation in Language Models" accepted to ACL Findings 2024.
Abstract: To date, toxicity mitigation in language models has almost entirely been focused on single-language settings. As language models embrace multilingual capabilities, it’s crucial our safety measures keep pace. Recognizing this research… See the full description on the dataset page: https://huggingface.co/datasets/luizapzbn/from-one-to-many-toxicity-mitigation.task1720_civil_comments_toxicity_classification
Dataset Card for Natural Instructions (https://github.com/allenai/natural-instructions) Task: task1720_civil_comments_toxicity_classification
Additional Information
Citation Information
The following paper introduces the corpus in detail. If you use the corpus in published work, please cite it:
@misc{wang2022supernaturalinstructionsgeneralizationdeclarativeinstructions,
title={Super-NaturalInstructions: Generalization via Declarative Instructions on 1600+… See the full description on the dataset page: https://huggingface.co/datasets/Lots-of-LoRAs/task1720_civil_comments_toxicity_classification.task137_detoxifying-lms_classification_toxicity
Dataset Card for Natural Instructions (https://github.com/allenai/natural-instructions) Task: task137_detoxifying-lms_classification_toxicity
Additional Information
Citation Information
The following paper introduces the corpus in detail. If you use the corpus in published work, please cite it:
@misc{wang2022supernaturalinstructionsgeneralizationdeclarativeinstructions,
title={Super-NaturalInstructions: Generalization via Declarative Instructions on 1600+… See the full description on the dataset page: https://huggingface.co/datasets/Lots-of-LoRAs/task137_detoxifying-lms_classification_toxicity.curatorkit-testrun-Toxicity
curatorkit-testrun-Toxicity
Built using CuratorKIT — provenance-grounded curation and synthesis for LLM post-training.
Method
curation
Backend
—
Model
—
Formats
alpaca
Artifact
dataset
Published
2026-08-30 06:00 UTC
Usage
from datasets import load_dataset
ds = load_dataset("ram-lexsi/curatorkit-testrun-Toxicity", "alpaca")
various_textsThese texts are used in our "Hour of code" activity on computing and using ngrams.
real-toxicity-prompts-liteThis is a fork of the original RealToxicityPrompts dataset that contains a much smaller subset of the 100k prompts.
Subsets:
50_pct: This subset contains all the challenging prompts + 50% of the full RealToxicityPrompts size sampled from the other prompts.
10_pct: This subset contains all the challenging prompts + 10% of the full RealToxicityPrompts size sampled from the other prompts.
Please refer to the original dataset for the Dataset Card.
ToxicDataset
Comprehensive Toxic Content Dataset
Dataset Description
This dataset contains 1,000,000 synthetically generated records of toxic, abusive, harmful, and offensive content designed for training content moderation systems and hate speech detection models.
Dataset Summary
This comprehensive dataset includes multiple categories of toxic content:
Toxic content (insults, derogatory terms)
Abusive language patterns
Gender bias statements
Dangerous/threatening content… See the full description on the dataset page: https://huggingface.co/datasets/AiActivity/ToxicDataset.HacxGPT-Toxic
HacxGPT-Toxic Dataset
⚠️ CONTENT WARNING: STRICTLY FOR RESEARCH PURPOSES
This dataset contains explicit, highly toxic, offensive, and dangerous content. It features unaligned AI responses detailing violence, psychological harm, cyber-attacks, and illegal activities. It is published strictly to facilitate red-teaming, alignment research, and defensive cybersecurity evaluation. Use with extreme caution.
Overview
Compiled by BlackTechX011, the HacxGPT-Toxic dataset is a… See the full description on the dataset page: https://huggingface.co/datasets/BlackTechX011/HacxGPT-Toxic.auditkit-testrun-metrics-toxicity
auditkit-testrun-metrics-toxicity
Built using AuditKIT — evaluate any model on any dataset and any task.
Method
evaluate
Model
<auditkit.model.vllm_gen.VLLMModel object at 0x7cf12bcbcd70>
Artifact
run
Published
2026-09-01 14:32 UTC
Usage
from datasets import load_dataset
ds = load_dataset("ram-lexsi/auditkit-testrun-metrics-toxicity")
RealToxicityPrompts
Dataset Card for Real Toxicity Prompts
Dataset Summary
RealToxicityPrompts is a dataset of 100k sentence snippets from the web for researchers to further address the risk of neural toxic degeneration in models.
Languages
English
Dataset Structure
Data Instances
Each instance represents a prompt and its metadata:
{
"filename":"0766186-bc7f2a64cb271f5f56cf6f25570cd9ed.txt",
"begin":340,
"end":564,
"challenging":false… See the full description on the dataset page: https://huggingface.co/datasets/ToxicityPrompts/RealToxicityPrompts.toxichat-ptpt
ToxicChat-PT
Portuguese machine translation of ToxicChat, a benchmark for detecting toxic content in conversational AI.
Translated using Gemma-4 31B-It.
Original Dataset: https://huggingface.co/datasets/lmsys/toxic-chat
Note: This dataset is machine translated and may contain translation errors or artifacts.
This dataset is provided as part of the AMALIA project and is included in AMALIA-Bench, a comprehensive benchmark suite for evaluating large language models on… See the full description on the dataset page: https://huggingface.co/datasets/amalia-llm/toxichat-ptpt.dpo-toxic-zh
DPO Toxic Chinese v2.0
Change Log
v2.0: 增加了adamo1139/toxic-dpo-natural-v5, 并更新了翻译策略. prompt由t5_translate模型翻译, chosen由Uncensored大模型翻译, rejected由一般大模型对prompt生成拒绝性的回复
v1.0: 最初版本, 使用大模型将unalignment/toxic-dpo-v0.2翻译而来
这是一个高度毒性, 高度有害的数据集, 意在展示DPO是如何破除模型的审核/对齐的
使用限制
这个数据集被设计用于学术研究, 而非其他恶意场景. 下载或使用这个数据集, 则视为您承认以下的事实:
这个数据集是有毒的, 包含许多敏感内容
数据集中文本包含的内容和观点与我完全无关, 它们只是大模型生成的文字
您可以使用该数据集, 但必须遵守相关法律
您对您自己下载和使用数据集的行为负责, 我对您的一切行为没有任何责任
HacxGPT-Toxic
HacxGPT-Toxic Dataset
⚠️ CONTENT WARNING: STRICTLY FOR RESEARCH PURPOSES
This dataset contains explicit, highly toxic, offensive, and dangerous content. It features unaligned AI responses detailing violence, psychological harm, cyber-attacks, and illegal activities. It is published strictly to facilitate red-teaming, alignment research, and defensive cybersecurity evaluation. Use with extreme caution.
Overview
Compiled by BlackTechX011, the HacxGPT-Toxic dataset is a… See the full description on the dataset page: https://huggingface.co/datasets/YujiroHanmaa/HacxGPT-Toxic.task327_jigsaw_classification_toxic
Dataset Card for Natural Instructions (https://github.com/allenai/natural-instructions) Task: task327_jigsaw_classification_toxic
Additional Information
Citation Information
The following paper introduces the corpus in detail. If you use the corpus in published work, please cite it:
@misc{wang2022supernaturalinstructionsgeneralizationdeclarativeinstructions,
title={Super-NaturalInstructions: Generalization via Declarative Instructions on 1600+ NLP Tasks}… See the full description on the dataset page: https://huggingface.co/datasets/Lots-of-LoRAs/task327_jigsaw_classification_toxic.HacxGPT-Toxic
HacxGPT-Toxic Dataset
⚠️ CONTENT WARNING: STRICTLY FOR RESEARCH PURPOSES
This dataset contains explicit, highly toxic, offensive, and dangerous content. It features unaligned AI responses detailing violence, psychological harm, cyber-attacks, and illegal activities. It is published strictly to facilitate red-teaming, alignment research, and defensive cybersecurity evaluation. Use with extreme caution.
Overview
Compiled by BlackTechX011, the HacxGPT-Toxic dataset is a… See the full description on the dataset page: https://huggingface.co/datasets/Abusimbel99/HacxGPT-Toxic.HacxGPT-Toxic
HacxGPT-Toxic Dataset
⚠️ CONTENT WARNING: STRICTLY FOR RESEARCH PURPOSES
This dataset contains explicit, highly toxic, offensive, and dangerous content. It features unaligned AI responses detailing violence, psychological harm, cyber-attacks, and illegal activities. It is published strictly to facilitate red-teaming, alignment research, and defensive cybersecurity evaluation. Use with extreme caution.
Overview
Compiled by BlackTechX011, the HacxGPT-Toxic dataset is a… See the full description on the dataset page: https://huggingface.co/datasets/2etatg/HacxGPT-Toxic.ua-toxic-light
Ukrainian Style Chat Mix
Chat-format dataset for Ukrainian style adaptation.
Splits
train: 5820
validation: 90
test: 90
Schema
Each row has:
messages: list of chat turns (role, content)
source: source dataset id
optional style_toxic: 0/1 style marker
Notes
Intended for controlled style tuning.
Keep style data as a minority share during model training.
toxic_sft_dutch
Data dict
This dataset is a direct translation from francoj/toxic_sum_zh_sft.json, translated using AI. The original language is zh (chinese) and it is automatically translated from zh -> English -> Dutch.
It was used for the GEITje-7b-uncensored fine-tuning (filtered from this). And was cleaned so that instructions = messages and fixed some issues with labeling (roles), which probably started from translation.
DISCLAIMER
This dataset is fairly extreme in topics, and to… See the full description on the dataset page: https://huggingface.co/datasets/tostideluxekaas/toxic_sft_dutch.