Giri2241/customer-support-client-agent-conversations

该数据集是一个合成的、上下文总结的多轮客户服务问答数据集，专注于银行业务领域，旨在训练和评估小型语言模型在对话连续性和上下文理解任务上的表现。数据集包含183,337个上下文总结的多轮客户服务对话，涵盖账户管理、交易、卡片服务、欺诈预防和一般查询等多种银行场景。每个实例包括对话历史总结、客户问题以及经过优化的代理回答，适用于训练指令调优的语言模型。数据集通过多阶段处理流程构建，包括初始数据源筛选、多轮对话构建、上下文总结、回答优化和内容审核等步骤。

A synthetic context-summarized multi-turn customer-service question-answering dataset for banking domain conversations, designed for training and evaluating small language models on dialogue continuity and contextual understanding tasks. The dataset contains 183,337 context-summarized multi-turn customer-service conversations spanning various banking scenarios including account management, transactions, card services, fraud prevention, and general inquiries. Each instance includes conversation history summarization, client questions, and refined agent responses optimized for training instruction-tuned language models. The dataset was created through a controlled multi-stage processing pipeline involving initial data source filtering, multi-turn conversation construction, context summarization, response refinement, and content moderation.