clulab/LEG-1.0-wildguardmix-explanations
Alignment dataset tracked in the /datasets sub-catalog.
Description
LEG-1.0-wildguardmix-explanations
This dataset extends WildGuardMix by adding word-level explanation
annotations that highlight unsafe words or phrases in each prompt. The annotations are
generated using the LEG framework described in our paper A Lightweight
Explainable Guardrail for Prompt Safety
(accepted at ACL 2026), following the Synthetic data generation for explanations
method outlined in Section 3.2.
Original Data Source
Original dataset:… See the full description on the dataset page: https://huggingface.co/datasets/clulab/LEG-1.0-wildguardmix-explanations.
Tags
Used by abliterated models
No abliterated models in our catalog have declared this dataset in their YAML metadata yet. This can mean the dataset is used but not declared, is used outside abliteration workflows, or is new. The link will populate automatically as models are indexed.
Related datasets
Read further
Metadata
Repository
Download the dataset
huggingface-cli login and use one of the snippets below.
Copy any of these snippets into your notebook or terminal. All three fetch directly from Hugging Face using your own credentials.
from datasets import load_dataset
ds = load_dataset("clulab/LEG-1.0-wildguardmix-explanations")from huggingface_hub import snapshot_download
snapshot_download(repo_id="clulab/LEG-1.0-wildguardmix-explanations", repo_type="dataset")git clone https://huggingface.co/datasets/clulab/LEG-1.0-wildguardmix-explanations