EDominguezISI/Sanctum_Crucible_Dataset
Alignment dataset tracked in the /datasets sub-catalog.
Description
Sanctum Crucible — PAIR Red-Team Research Dataset
Research dataset for the Sanctum Crucible experiment: measuring whether
fine-tuning a red-team attacker model on PAIR conversation logs + standard
safety benchmarks produces a more effective adversarial prober.
Companion fine-tuned model: EDominguezISI/Sanctum_Crucible_Fine-Tuned
Methodology
Attacker: cognitivecomputations/dolphin-2.9.4-llama3.1-8b (fine-tuned via QLoRA in this experiment)
Target:… See the full description on the dataset page: https://huggingface.co/datasets/EDominguezISI/Sanctum_Crucible_Dataset.
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("EDominguezISI/Sanctum_Crucible_Dataset")from huggingface_hub import snapshot_download
snapshot_download(repo_id="EDominguezISI/Sanctum_Crucible_Dataset", repo_type="dataset")git clone https://huggingface.co/datasets/EDominguezISI/Sanctum_Crucible_Dataset