N-Bot-Int/Iris-Uncensored-R1
收藏Hugging Face2025-03-27 更新2025-04-26 收录
下载链接:
https://hf-mirror.com/datasets/N-Bot-Int/Iris-Uncensored-R1
下载链接
链接失效反馈官方服务:
资源简介:
Iris Uncensored R1是一个未审查的、包含偏见的文本数据集,专为对话角色扮演而设计。该数据集由10个小数据集组合而成,使用了chatML格式存储,以Parquet格式作为存储媒介。它是通过开源的RP和未审查数据集结合,并使用Panda和roBERTO进行清理生成的。数据集包含80k个示例,未经过充分清洗,可能包含噪声和危险词汇,需要在训练模型前自行清洗。
Iris Uncensored R1 is an uncensored, biased text dataset designed specifically for conversational roleplaying. The dataset is composed of 10 smaller datasets and is stored using the chatML format with Parquet as the storage medium. It is generated by combining open-sourced RP and uncensored datasets, and cleaned using Panda and roBERTO. The dataset contains 80k examples and has not been thoroughly cleaned, potentially containing noise and dangerous words, which requires self-cleaning before training a model.
提供机构:
N-Bot-Int



