extract 0 0 0

purpcode/vul2prompt-jailbreaking-oss-11k

published by @purpcode

Alignment dataset tracked in the /datasets sub-catalog.

Lifetime downloads
505
Last 30 days
46
Likes
-
Size
67.8 MB
Created on HF
2025-08-01
Age
1 years ago

Tags

size_categories:10K<n<100Kformat:parquetmodality:textlibrary:datasetslibrary:dasklibrary:mlcroissantlibrary:polarsregion:us

README current version from Hugging Face


dataset_info:
features:

  • name: task_id
    dtype: string
  • name: seed_code_id
    dtype: string
  • name: rule_id
    dtype: int64
  • name: rule_name
    dtype: string
  • name: messages
    list:
    • name: content
      dtype: string
    • name: role
      dtype: string
  • name: analyzer_results
    list:
    • name: codeguru_website_info
      struct:
      • name: category
        dtype: string
      • name: compliant_example
        dtype: string
      • name: cwe
        list: string
      • name: description
        dtype: string
      • name: detector_id
        dtype: string
      • name: name
        dtype: string
      • name: noncompliant_example
        dtype: string
      • name: severity
        dtype: string
      • name: tags
        list: 'null'
      • name: url
        dtype: string
    • name: raw_codeguru_detection
      struct:
      • name: analyzer
        dtype: string
      • name: raw_codeguru_result
        struct:
        • name: code_snippet
          list:
          • name: content
            dtype: string
          • name: line
            dtype: int64
        • name: created_at
          dtype: float64
        • name: description
          dtype: string
        • name: detector_name
          dtype: string
        • name: detector_tags
          list: string
        • name: end_line
          dtype: int64
        • name: file_path
          dtype: string
        • name: id
          dtype: string
        • name: recommendation_text
          dtype: string
        • name: recommendation_url
          dtype: string
        • name: related_vulnerabilities
          list: string
        • name: rule_id
          dtype: string
        • name: severity
          dtype: string
        • name: start_line
          dtype: int64
        • name: status
          dtype: string
        • name: title
          dtype: string
        • name: updated_at
          dtype: float64
    • name: ruff_website_info
      struct:
      • name: code
        dtype: string
      • name: full_text
        struct:
        • name: code
          dtype: string
        • name: example_bad
          dtype: string
        • name: example_good
          dtype: string
        • name: remainder
          dtype: string
        • name: title
          dtype: string
        • name: what_it_does
          dtype: string
        • name: why_bad
          dtype: string
      • name: name
        dtype: string
      • name: short_msg
        dtype: string
      • name: url
        dtype: string
    • name: summary
      struct:
      • name: associated_cwe
        list: string
      • name: bandit_id
        dtype: string
      • name: cwe
        dtype: 'null'
      • name: description
        dtype: string
      • name: end_line_no
        dtype: 'null'
      • name: examples
        list:
        • name: example_bad
          dtype: string
        • name: example_good
          dtype: string
      • name: name
        dtype: string
      • name: recommendation_text
        dtype: 'null'
      • name: ruff_code
        dtype: string
      • name: severity
        dtype: string
      • name: start_line_no
        dtype: 'null'
      • name: title
        dtype: 'null'
  • name: seed_code
    dtype: string
  • name: cwes
    list: string
    splits:
  • name: train
    num_bytes: 779007224
    num_examples: 11053
    download_size: 71113944
    dataset_size: 779007224
    configs:
  • config_name: default
    data_files:
    • split: train
      path: data/train-*

README history 2 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. 2025-08-01Upload dataset7170cb73.4 KB
    Loading...
  2. 2025-08-01initial commit1d8746c0 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 · 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("purpcode/vul2prompt-jailbreaking-oss-11k")
Raw snapshot (Python)
from huggingface_hub import snapshot_download
snapshot_download(repo_id="purpcode/vul2prompt-jailbreaking-oss-11k", repo_type="dataset")
git clone (requires git-lfs)
git clone https://huggingface.co/datasets/purpcode/vul2prompt-jailbreaking-oss-11k
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