exp_rpt_unitsyn-python-v3-qwen3.5-122b-131k-opencode-sft-serveparity
收藏资源简介:
该数据集是一个专门针对工具增强对话场景构建的对话数据集。数据集中每个样本代表一个完整的对话会话,包含多轮对话消息(messages),每条消息记录发言者角色(role)、对话内容(content)以及可选的工具调用信息(tool_calls)。工具调用信息详细记录了调用类型(type)和函数细节,包括函数名称(name)和参数(arguments)。此外,每个样本还包含对话所使用的工具定义(tools)、任务类型描述(task)、对话总轮数(num_turns)和工具调用总次数(num_tool_calls)。数据集共包含130个训练样本,适用于工具调用对话生成、任务导向对话系统、多轮对话建模等研究任务。
This dataset is a dialogue dataset specifically constructed for tool-enhanced dialogue scenarios. Each sample in the dataset represents a complete dialogue session, containing multiple rounds of dialogue messages (messages), with each message recording the speakers role (role), dialogue content (content), and optional tool call information (tool_calls). The tool call information details the call type (type) and function specifics, including the function name (name) and arguments. Additionally, each sample includes the tool definitions used in the dialogue (tools), task type description (task), total number of dialogue turns (num_turns), and total number of tool calls (num_tool_calls). The dataset contains a total of 130 training samples and is suitable for research tasks such as tool call dialogue generation, task-oriented dialogue systems, and multi-turn dialogue modeling.




