sharegpt_gpt4

MedicalGPT仓库提供了一个基于GPT4的多语言问答数据集，核心定位是为LLM提供训练数据。该数据集包含从ShareGPT中挑选出的多轮对话数据，数据规模约为10万条，主要涵盖知识问答、编程题、推理计算等任务。标注信息来源于原作者，并采用CC-BY 4.0授权许可。数据集支持文本分类和文本生成等任务。

The MedicalGPT repository offers a multilingual question-answering dataset built on GPT-4, whose core objective is to provide training data for large language models (LLMs). This dataset contains multi-turn dialogue data curated from ShareGPT, with a total scale of approximately 100,000 samples, mainly covering tasks including knowledge-based question answering, programming challenges, and reasoning and computation. The annotation information originates from the original author, and the dataset is released under the CC-BY 4.0 license. It supports downstream tasks such as text classification and text generation.

搜集汇总

背景与挑战

背景概述

sharegpt_gpt4是一个多语言问答数据集，包含约10万条从ShareGPT精选的多轮对话数据，涵盖知识问答、编程题和推理计算等任务，支持文本分类和文本生成，采用CC-BY 4.0授权许可。