ai2-adapt-dev/toolu-sft-mix-T2-S2

该数据集包含了会话信息，每个会话有唯一的标识符(id)，来源(source)，消息内容(messages)，步骤数(n_step)，对话轮数(n_turn)，执行类型(exec_type)，是否拒绝(is_refusal)，是否可解析(parseable)，API中的函数是否存在(function_in_api)，参数数量是否正确(num_args_correct)，参数名是否正确(arg_names_correct)，代码是否正确(code_is_correct)，以及使用的模型(model)。数据集分为训练集(train)，共有1446095个样本，大小为5309590684字节。

The dataset contains conversational information, with each conversation having a unique identifier (id), source (source), message content (messages), number of steps (n_step), number of dialogue turns (n_turn), execution type (exec_type), whether refused (is_refusal), whether parseable (parseable), whether the function exists in the API (function_in_api), whether the number of arguments is correct (num_args_correct), whether the argument names are correct (arg_names_correct), whether the code is correct (code_is_correct), and the model used (model). The dataset is split into a training set (train) with a total of 1,446,095 samples, totaling 5,309,590,684 bytes.