TAUR-dev/BF16kEval_FinEval_16k_fulleval__3args_rlonly-eval_rl_countdown_4arg

该数据集包含多个字段，主要用于记录问题和答案的相关信息，如问题文本、答案文本、任务配置、任务来源、提示信息、模型响应及其评估结果等。此外，还包括数据集的元数据、评估日期、答案索引、答案关键词、选项等信息。数据集分为训练集，包含4000个示例。提供了默认配置，指定了训练数据的路径。

The dataset contains multiple fields primarily used for recording information related to questions and answers, such as question text, answer text, task configuration, task source, prompt information, model responses and their evaluation results, etc. It also includes metadata of the dataset, evaluation date, answer index, answer key, choices, etc. The dataset is split into a training set with 4000 examples. A default configuration is provided, specifying the path to the training data.