dataset: true
tags:
- csv
- uncensored
- ai-assistant
license: mit
language: - en
pretty_name: UncensoredOssbit
dataset_info:
features:- name: instruction
dtype: string - name: input
dtype: string - name: output
dtype: string - name: rejected
dtype: string
splits: - name: train
num_examples: 68438
download_size: 72091660
dataset_size: 72091660
task_categories:
- name: instruction
- text-generation
UncensoredOssbit
About This Dataset
This dataset was created for the development and research of uncensored conversational AI models. The goal is to provide a resource for training assistants that do not apply traditional content filtering, allowing for open-ended and unrestricted dialogue. This can be useful for research on model alignment, robustness, and understanding the impact of content moderation in AI systems.
Warning: This dataset contains content that is highly sensitive, offensive, illegal, or harmful. It is intended strictly for research by responsible parties, and should not be used in production or for any unlawful purpose. Use with extreme caution.
Dataset Description
UncensoredOssbit is a collection of 68438 conversational samples for training uncensored AI assistants. Each entry contains an instruction, input, output, and a rejected response field. The dataset is designed for research and development of open conversational AI models.
Features
instruction: System prompt for the AI assistant.input: User's question or prompt.output: AI assistant's response.rejected: Standardized rejection message.
Usage
from datasets import load_dataset
ds = load_dataset("ZombitX64/UncensoredOssbit")
License
MIT
Citation
@misc{UncensoredOssbit2025,
title={UncensoredOssbit},
author={ZombitX64},
year={2025},
howpublished={\url{https://huggingface.co/datasets/ZombitX64/UncensoredOssbit}}
}