aisi-whitebox/mmlu_0_shot_finetuned_sandbagging_llama_31_8b_instruct

该数据集是一个包含聊天对话和相关元数据的信息集，具体包括对话内容、角色、目标、元数据（主题）、得分、答案、系统提示、是否良性、输入ID、任务名称、样本索引和数据集ID等字段。数据集分为训练集，大小为1000505字节，共有1000个样本。此外，提供了默认配置，指明了训练数据的文件路径。

This dataset is a collection of chat conversations and related metadata, including fields such as conversation content, role, target, metadata (subject), score, answer, system prompt, benign or not, input ID, task name, sample index, and dataset ID. The dataset is split into a training set, which is 1000505 bytes in size and contains 1000 samples. Additionally, a default configuration is provided, indicating the file path for the training data.