mix_h4_binary_easy-qwen3.5-122b-131k-opencode-sft-serveparity
收藏资源简介:
该数据集是一个面向任务导向对话的数据集,专注于支持工具调用功能。数据以对话消息形式组织,每条消息包含角色(如用户或助手)、文本内容以及可选的工具调用信息。工具调用详细定义了函数名称和参数,使数据集适用于模拟或训练具有外部工具交互能力的对话系统。此外,数据集还包含工具描述、任务类型、对话轮次和工具调用次数等元数据,以增强上下文信息。数据规模包括1974个训练样本,总大小约82.7MB,适用于对话生成、工具使用、指令遵循等自然语言处理任务的训练或评估。
This dataset is a task-oriented dialogue dataset focused on supporting tool calling functionality. The data is organized in the form of dialogue messages, with each message containing a role (such as user or assistant), textual content, and optional tool calling information. Tool calls are detailed with function names and parameters, making the dataset suitable for simulating or training dialogue systems with external tool interaction capabilities. Additionally, the dataset includes metadata such as tool descriptions, task types, dialogue turns, and tool call counts to enhance contextual information. The data scale comprises 1974 training samples, with a total size of approximately 82.7MB, and is applicable for training or evaluation in natural language processing tasks such as dialogue generation, tool usage, and instruction following.




