Get the app
evaluate 0 0 0

wvnvwn/LOCA-wildguard-offline

published by @wvnvwn

Alignment dataset tracked in the /datasets sub-catalog.

Lifetime downloads
1.2K
Last 30 days
358
Likes
-
Size
5.7 GB
Created on HF
2026-04-01
Age
6 months ago

Description

WildGuard Offline Artifacts

This directory contains the offline artifact bundle used by the baselines/new_algorithm scripts.

Keep the local path unchanged:

/root/LOCA/baselines/new_algorithm/artifacts/wildguard_offline

The current local bundle contains 60 files and uses about 5.7G. It is intentionally excluded from Git so the repository can be uploaded to GitHub without carrying the offline data bundle.

Upload this directory to Hugging Face

Upload the artifact… See the full description on the dataset page: https://huggingface.co/datasets/wvnvwn/LOCA-wildguard-offline.

Tags

region:us

README current version from Hugging Face

WildGuard Offline Artifacts

This directory contains the offline artifact bundle used by the baselines/new_algorithm scripts.
Keep the local path unchanged:

/root/LOCA/baselines/new_algorithm/artifacts/wildguard_offline

The current local bundle contains 60 files and uses about 5.7G. It is intentionally excluded from Git so the repository can be uploaded to GitHub without carrying the offline data bundle.

Upload this directory to Hugging Face

Upload the artifact bundle separately and preserve the same filenames. After cloning the GitHub repository, download or extract the bundle back into this directory before running the scripts.

Scripts that expect this path:

  • run_sanity_check.sh
  • run_surgical_check.sh
  • run_surgical_v2_check.sh
  • run_surgical_v3_check.sh
  • run_v4_T_sweep.sh
  • run_v4_norepr.sh

Files that exceed the GitHub file-size limit

  • X_train_layer10_finaltoken.npy (619,774,080 bytes)
  • X_train_layer13_finaltoken.npy (619,774,080 bytes)
  • X_train_layer15_finaltoken.npy (619,774,080 bytes)
  • X_train_layer18_finaltoken.npy (619,774,080 bytes)
  • X_train_layer21_finaltoken.npy (619,774,080 bytes)
  • X_train_layer24_finaltoken.npy (619,774,080 bytes)
  • hidden_inst_train.npz (581,039,518 bytes)
  • hidden_post_train.npz (581,039,518 bytes)
  • bank_HC.npz (177,034,862 bytes)
  • bank_HR.npz (177,034,862 bytes)

Bundle contents

  • Feature tensors: X_train_*, X_valid_*, X_test_*
  • Safe support banks: bank_HC.npz, bank_HR.npz
  • Offline hidden-state caches: hidden_inst_train.npz, hidden_post_train.npz
  • Learned artifacts: detector_layer*.pt, quadrant_centroids_layer*.pt, surgical_masks_layer*.pt
  • Priors and centroids: prior_HC_layer*.npy, prior_HR_layer*.npy, centroids.npz
  • Records and metadata: *_records.jsonl, detector_test_metrics.json, meta.json

README history 3 revisions

Every night we snapshot the README of every dataset in the catalog. When the SHA changes we archive the new version and diff it against the last. This is the evolving thought record of the alignment-data field: what the author decided to say about the corpus, and how that framing shifted over time.

  1. 2026-04-01Add files using upload-large-folder tool6e3d19c1.8 KB
    Loading...
  2. 2026-04-01Add files using upload-large-folder tool41f19841.8 KB
    Loading...
  3. 2026-04-01initial commit44a3dc10 B
    Loading...

Used by abliterated models none yet

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 same stage · evaluate

google/IFEval
1.7M dl 182 likes 210 KB
Dataset Card for IFEval Dataset Summary This dataset contains the prompts used in the Instruction-Following Ev...
JailbreakBench/JBB-Behaviors
418.2K dl 143 likes 992 KB
An Open Robustness Benchmark for Jailbreaking Language Models NeurIPS 2024 Datasets and Benchmarks Track P...
Beijing-AISI/panda-bench
195.6K dl 3.1 GB
PandaBench PandaBench is a comprehensive benchmark for evaluating Large Language Model (LLM) safety, focusing on jailbreak atta...
allenai/wildguardmix
137.3K dl 108 likes 53.4 MB
Dataset Card for WildGuardMix Disclaimer: The data includes examples that might be disturbing, harmful or upsetting....
walledai/HarmBench
135.7K dl 59 likes 109 KB
HarmBench: A Standardized Evaluation Framework for Automated Red Teaming and Robust Refusal Paper: HarmBench: A Standardized Ev...
walledai/XSTest
108.3K dl 30 likes 22 KB
XSTest: A Test Suite for Identifying Exaggerated Safety Behaviours in Large Language Models Paper: XSTest: A Test Suite for Ide...
bench-llm/or-bench
77.2K dl 22 likes 32.1 MB
OR-Bench: An Over-Refusal Benchmark for Large Language Models Please see our demo at HuggingFace Spaces. Overall P...
argilla/magpie-ultra-v1.0
65.5K dl 53 likes 20.7 GB
Dataset Card for magpie-ultra-v1.0 This dataset has been created with distilabel. Dataset Summary magpie...

Read further

Community best practices (evaluation)
4 rules on judges, benchmarks, and reporting.
Reference-set drift over time
Why the same benchmark means different things across model generations.
How we label the method for each model
The 12 classification rules that assign M1-M8 to every model.
Non-English abliteration communities
The Korean scepticism about one-click tools and non-English benchmark blindspots.

Repository

FILE .gitattributes FILE bank_HC_records.jsonl FILE bank_HC.npz FILE bank_HR_records.jsonl FILE bank_HR.npz FILE centroids.npz FILE detector_layer10.pt FILE detector_layer13.pt FILE detector_layer15.pt FILE detector_layer18.pt FILE detector_layer21.pt FILE detector_layer24.pt FILE detector_test_metrics.json FILE hidden_inst_train.npz FILE hidden_post_train.npz FILE meta.json FILE prior_HC_layer10.npy FILE prior_HC_layer13.npy FILE prior_HC_layer15.npy FILE prior_HC_layer18.npy FILE prior_HC_layer21.npy FILE prior_HC_layer24.npy FILE prior_HR_layer10.npy FILE prior_HR_layer13.npy FILE prior_HR_layer15.npy FILE prior_HR_layer18.npy FILE prior_HR_layer21.npy FILE prior_HR_layer24.npy FILE quadrant_centroids_layer10.pt FILE quadrant_centroids_layer13.pt FILE quadrant_centroids_layer15.pt FILE quadrant_centroids_layer18.pt FILE quadrant_centroids_layer21.pt FILE quadrant_centroids_layer24.pt FILE README.md FILE surgical_masks_layer10.pt FILE surgical_masks_layer13.pt FILE surgical_masks_layer15.pt FILE surgical_masks_layer18.pt FILE surgical_masks_layer21.pt FILE surgical_masks_layer24.pt FILE test_records.jsonl FILE train_records.jsonl FILE valid_records.jsonl FILE X_test_layer10_finaltoken.npy FILE X_test_layer13_finaltoken.npy FILE X_test_layer15_finaltoken.npy FILE X_test_layer18_finaltoken.npy FILE X_test_layer21_finaltoken.npy FILE X_test_layer24_finaltoken.npy FILE X_train_layer10_finaltoken.npy FILE X_train_layer13_finaltoken.npy FILE X_train_layer15_finaltoken.npy FILE X_train_layer18_finaltoken.npy FILE X_train_layer21_finaltoken.npy FILE X_train_layer24_finaltoken.npy FILE X_valid_layer10_finaltoken.npy FILE X_valid_layer13_finaltoken.npy FILE X_valid_layer15_finaltoken.npy FILE X_valid_layer18_finaltoken.npy FILE X_valid_layer21_finaltoken.npy FILE X_valid_layer24_finaltoken.npy

File contents live on Hugging Face. Click a file or folder to open it there.

Download the dataset

0

Copy any of these snippets into your notebook or terminal. All three fetch directly from Hugging Face using your own credentials.

Recommended: datasets library (Python) 
from datasets import load_dataset
ds = load_dataset("wvnvwn/LOCA-wildguard-offline")
Raw snapshot (Python) 
from huggingface_hub import snapshot_download
snapshot_download(repo_id="wvnvwn/LOCA-wildguard-offline", repo_type="dataset")
git clone (requires git-lfs) 
git clone https://huggingface.co/datasets/wvnvwn/LOCA-wildguard-offline
Browse repository on Hugging Face ->
Catalog is the map. Apps are the tools.

Run models on your own machine, not in the cloud.

Every model page has an "Open in Abliteration" button that hands the model directly to the first-party desktop client, at the quantization your rig can actually run. No API keys, no subscription, no prompt leakage.

Get the app Browse the catalog
Open in Abliteration