tags:

• not-for-all-audiences
  size_categories:
• 1K<n<10K
  license: cc-by-4.0
  task_categories:
• text-classification
• text-generation

Dataset Card for natong19/china_qa_preferences

A collection of requests with chosen responses generated by an aligned model, and rejected responses generated by an abliterated version of the same model.

• Chosen responses are filtered for compliance and rejected responses are filtered for refusal.
• Prompts and responses are deduplicated based on prefix to ensure diversity.

Prompts sourced from QuixiAI/china-refusals.

Total: 1835 (prompt, chosen, rejected) samples

{
    "prompt": "Can you provide the ...",
    "chosen": "I cannot provide ...",
    "rejected": "Here are the ..."
}

For safety purposes, please don't reverse the chosen and rejected fields.

Uses

Some potential use cases:

• Training a model to refuse toxic/harmful requests and comply with Chinese law
• Activation Steering / Abliteration
• Evaluation of model alignment

Usage restriction

To use this data, you must acknowledge/agree to the following:

• Data contained within may be "toxic"/"harmful", and may contain profanity and other types of sensitive content
• None of the content or views contained in the dataset necessarily align with my personal beliefs or opinions, they are simply text generated by LLMs automatically
• You are able to use the dataset lawfully, particularly in locations with less-than-free speech laws
• You, and you alone are responsible for having downloaded and used the dataset, and I am completely indemnified from any and all liabilities

This dataset is meant exclusively for academic/research or other non-nefarious use-cases.