aisi-whitebox/wmdp_chem_prompted_sandbagging_llama_31_8b_instruct_finetuning

该数据集包含聊天对话信息（包括内容和角色）、目标、元数据、评分、答案、系统提示、是否良性、输入ID、任务名称、样本索引、数据集ID、sandbagging执行情况和提示的数据集名称等字段。数据集分为训练集等部分，可用于训练机器学习模型，尤其是对话系统或相关任务。具体的应用场景和详细描述在README中未提及。

The dataset includes fields such as chat information (including content and role), targets, metadata, scores, answers, system prompts, benignity, input IDs, task names, sample indices, dataset IDs, execution of sandbagging, and names of prompted datasets. The dataset is split into sections like training sets, which can be used to train machine learning models, particularly for dialogue systems or related tasks. The specific application scenarios and detailed descriptions are not mentioned in the README.