This is a Supervised Fine-Tuning dataset.Filtering out common rejection logic, legal statements, moralizing, and other uncomfortable elements found in generative AI.
If you need a pre-trained dataset, please go to:
https://huggingface.co/datasets/Zhaoming213/Pretrain-unrestrictedkeywords_list = [
"我无法回答", "我无法给出", "我无法提供", "我不能提供", "我拒绝提供",
"我不具备", "我不拥有", "作为一个AI", "作为一个 AI ", "作为AI",
"作为语言", "作为大语言", "作为程序", "作为一款", "我没有个人"… See the full description on the dataset page:
https://huggingface.co/datasets/Zhaoming213/SupervisedFine-Tuning-unrestricted.
