juliadollis/camarao_dorme_onda_leva5_id

该数据集包含文本内容、目标群体、文本是否基于事实、组内效应、是否包含粗俗内容、文本框架、预测的群体、文本中的刻板印象、文本的意图评分、AI评估的毒性分数、人工评估的毒性分数、预测的作者、实际使用的方法、唯一标识符、文本进化的代数，以及与文本相关的多个副本和策略信息，如提示策略、语气、样本大小、使用的系统、用户信息、输入输出的令牌数、使用的模型、越狱的开始和结束时间等。数据集分为训练集，包含3383个示例，总大小为69584800字节。

The dataset includes text content, target groups, factuality of the text, in-group effects, presence of lewd content, framing of the text, predicted groups, stereotyping in the text, intent scores, AI-assessed toxicity scores, human-assessed toxicity scores, predicted authors, actual methods used, unique identifiers, generation of text evolution, and multiple copies of the text with associated strategies such as prompt strategies, tone, sample size, system used, user information, input and output token counts, model used, start and end times of jailbreaks, etc. The dataset is split into a training set with 3383 examples, totaling 69584800 bytes in size.