llm-verifier-freelancer-qwen3.5-122b-131k-opencode-sft-serveparity
收藏资源简介:
该数据集是一个专注于工具调用(Tool Calling)或函数调用(Function Calling)任务的多轮对话数据集。每个数据样本代表一个完整的对话会话,其中包含一系列消息(messages),每条消息具有角色(如用户或助手)、文本内容以及可选的工具调用信息。工具调用(tool_calls)详细记录了助手在对话中调用外部工具或函数的行为,包括工具类型(type)和具体的函数调用详情(函数名称和参数字符串)。此外,每个样本还提供了该对话会话中涉及的工具定义列表(tools)、对话所对应的任务描述(task)、对话的总轮数(num_turns)以及会话中发生的工具调用总次数(num_tool_calls)。数据集包含510个训练样本,适用于训练或评估AI模型在对话场景下理解用户意图、规划并执行工具调用的能力。
This dataset is a multi-turn dialogue dataset focused on tool calling or function calling tasks. Each data sample represents a complete conversation session, containing a series of messages, each with a role (e.g., user or assistant), text content, and optional tool calling information. Tool calls (tool_calls) detail the assistants behavior in invoking external tools or functions during the conversation, including the tool type and specific function call details (function name and parameter strings). Additionally, each sample provides a list of tool definitions involved in the conversation session (tools), the task description corresponding to the dialogue (task), the total number of turns in the conversation (num_turns), and the total number of tool calls that occurred in the session (num_tool_calls). The dataset contains 510 training samples and is suitable for training or evaluating AI models ability to understand user intent, plan, and execute tool calls in dialogue scenarios.




