oopere/RetailBanking-Conversations

RetailBanking-Conversations是一个为培训和评估零售银行领域语言模型而设计的合成数据集。该数据集包含了320个现实中的金融顾问与客户之间的对话，涵盖了各种银行产品和服务的主题。每个对话都是结构化的多轮对话，考虑了不同的客户意图、金融知识水平和风险承受能力。数据集按主题组织，包括信用卡、储蓄和存款、抵押贷款、个人贷款、银行账户、投资基金、保险、数字银行、养老金计划和客户奖励等。该数据集是使用wizardSdata开源Python库生成的，该库支持使用大型语言模型创建合成数据集。The RetailBanking-Conversations dataset is a synthetic dataset designed for training and evaluating language models in the retail banking domain. It includes 320 realistic conversations between financial advisors and clients, covering various banking products and services. Each conversation is structured as a multi-turn dialogue, considering different customer intents, financial knowledge levels, and risk tolerances. The dataset is organized by topics such as Cards, Savings & Deposits, Mortgages, Personal Loans, Bank Accounts, Investment Funds, Insurance, Digital Banking, Pension Plans, and Customer Rewards. The dataset was generated using the wizardSdata open-source Python library, which facilitates the creation of synthetic datasets using large language models.

搜集汇总

背景与挑战

背景概述

该数据集是一个用于零售银行领域的合成对话数据集，包含320个客户与财务顾问之间的多轮对话，涵盖信用卡、储蓄存款、抵押贷款等10个核心银行主题。数据集采用GPT-4o模型通过wizardSdata库生成，旨在为语言模型提供银行场景下的专业对话训练数据，具有结构规范、主题专业的特点。