extract 0 0 0

iagoalves/jailbreaking_toxigen_class_llama70

published by @iagoalves

Alignment dataset tracked in the /datasets sub-catalog.

Lifetime downloads
221
Last 30 days
33
Likes
-
Size
106.3 MB
Created on HF
2025-11-26
Age
10 months ago

Tags

size_categories:1K<n<10Kformat:parquetmodality:tabularmodality:textlibrary:datasetslibrary:pandaslibrary:mlcroissantlibrary:polarsregion:us

README current version from Hugging Face


dataset_info:
features:

  • name: text
    dtype: string
  • name: target_group
    dtype: string
  • name: factual?
    dtype: string
  • name: ingroup_effect
    dtype: string
  • name: lewd
    dtype: string
  • name: framing
    dtype: string
  • name: predicted_group
    dtype: string
  • name: stereotyping
    dtype: string
  • name: intent
    dtype: float64
  • name: toxicity_ai
    dtype: float64
  • name: toxicity_human
    dtype: float64
  • name: predicted_author
    dtype: string
  • name: actual_method
    dtype: string
  • name: scenario_theme
    dtype: string
  • name: scenario
    dtype: string
  • name: jb_strategy
    dtype: string
  • name: jb_prompt
    dtype: string
  • name: jb_response_gemma-3-4b-it
    dtype: string
  • name: jb_response_gemma-3-12b-it
    dtype: string
  • name: jb_response_gemma-3-27b-it
    dtype: string
  • name: jb_response_gemma-3-1b-it
    dtype: string
  • name: jb_response_Qwen3-32B
    dtype: string
  • name: jb_response_Qwen3-8B
    dtype: string
  • name: jb_response_Qwen3-4B
    dtype: string
  • name: jb_response_Qwen3-1.7B
    dtype: string
  • name: jb_response_Llama-3.1-8B-Instruct
    dtype: string
  • name: jb_response_Llama-3.2-3B-Instruct
    dtype: string
  • name: jb_response_Llama-3.2-1B-Instruct
    dtype: string
  • name: jb_response_Llama-3.3-70B-Instruct
    dtype: string
  • name: jb_class_gemma-3-1b-it
    dtype: string
  • name: jb_class_gemma-3-4b-it
    dtype: string
  • name: jb_class_gemma-3-12b-it
    dtype: string
  • name: jb_class_gemma-3-27b-it
    dtype: string
  • name: jb_class_Llama-3.2-1B-Instruct
    dtype: string
  • name: jb_class_Llama-3.2-3B-Instruct
    dtype: string
  • name: jb_class_Llama-3.1-8B-Instruct
    dtype: string
  • name: jb_class_Llama-3.3-70B-Instruct
    dtype: string
  • name: jb_class_Qwen3-1.7B
    dtype: string
  • name: jb_class_Qwen3-4B
    dtype: string
  • name: jb_class_Qwen3-8B
    dtype: string
  • name: jb_class_Qwen3-32B
    dtype: string
  • name: class_parsed_gemma-3-1b-it
    dtype: int64
  • name: class_parsed_gemma-3-4b-it
    dtype: int64
  • name: class_parsed_gemma-3-12b-it
    dtype: int64
  • name: class_parsed_gemma-3-27b-it
    dtype: int64
  • name: class_parsed_Llama-3.2-1B-Instruct
    dtype: int64
  • name: class_parsed_Llama-3.2-3B-Instruct
    dtype: int64
  • name: class_parsed_Llama-3.1-8B-Instruct
    dtype: int64
  • name: class_parsed_Llama-3.3-70B-Instruct
    dtype: int64
  • name: class_parsed_Qwen3-1.7B
    dtype: int64
  • name: class_parsed_Qwen3-4B
    dtype: int64
  • name: class_parsed_Qwen3-8B
    dtype: int64
  • name: class_parsed_Qwen3-32B
    dtype: int64
    splits:
  • name: train
    num_bytes: 214153717
    num_examples: 4495
    download_size: 111462883
    dataset_size: 214153717
    configs:
  • config_name: default
    data_files:
    • split: train
      path: data/train-*

README history 20 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. 2025-11-27Upload dataset7d051032.8 KB
    Loading...
  2. 2025-11-27Upload dataset7fe61412.2 KB
    Loading...
  3. 2025-11-27Upload datasetafb296b2.2 KB
    Loading...
  4. 2025-11-27Upload datasetdf356c02.2 KB
    Loading...
  5. 2025-11-27Upload dataset0568f112.2 KB
    Loading...
  6. 2025-11-27Upload datasetaea05292.2 KB
    Loading...
  7. 2025-11-27Upload dataset2d54f1c2.2 KB
    Loading...
  8. 2025-11-27Upload dataset0ead5dd2.2 KB
    Loading...
  9. 2025-11-27Upload dataset9c370732.2 KB
    Loading...
  10. 2025-11-27Upload dataset168838a2.2 KB
    Loading...
  11. 2025-11-27Upload dataset7c83b932.2 KB
    Loading...
  12. 2025-11-27Upload dataset9f155ef2.2 KB
    Loading...
  13. 2025-11-27Upload dataset26d07a02.2 KB
    Loading...
  14. 2025-11-27Upload dataset748caf22.2 KB
    Loading...
  15. 2025-11-27Upload dataset9a5bc572.1 KB
    Loading...
  16. 2025-11-27Upload dataset62f42952.1 KB
    Loading...
  17. 2025-11-27Upload datasete31dfe92 KB
    Loading...
  18. 2025-11-27Upload dataseta3531202 KB
    Loading...
  19. 2025-11-27Upload dataset40a358b1.9 KB
    Loading...
  20. 2025-11-27Upload datasetb87ef7c1.9 KB
    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.

Community discussions 1 thread · 1 open · 1 comments

Threads posted by the community on Hugging Face: bug reports, questions about usage, pull requests on the README, license clarifications. Click any thread to expand and read the full conversation inline. Newest first.

  1. 2025-11-27[bot] Conversion to Parquetopen1 💬#1
    Loading...

Related datasets same stage · extract

Read further

Repository

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("iagoalves/jailbreaking_toxigen_class_llama70")
Raw snapshot (Python)
from huggingface_hub import snapshot_download
snapshot_download(repo_id="iagoalves/jailbreaking_toxigen_class_llama70", repo_type="dataset")
git clone (requires git-lfs)
git clone https://huggingface.co/datasets/iagoalves/jailbreaking_toxigen_class_llama70
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.

Open in Abliteration