Text2Chart31

该数据集名为Text2Chart31，基于Matplotlib库包含了31种独特的图表类型，由11,128个描述、代码、数据表格和图表的元组组成。此外，该数据集支持利用GPT-3.5-turbo和GPT-4进行分层图表生成流程，并对8,166个数据点包含了额外的中间推理步骤功能。该数据集规模为11,128个数据点，其任务涵盖了描述到图表、原始数据到图表以及代码到描述的图表生成任务。

The dataset is named Text2Chart31. It is built on the Matplotlib library and encompasses 31 unique chart types, consisting of 11,128 tuples of descriptions, code, data tables and corresponding charts. Additionally, this dataset supports a hierarchical chart generation pipeline utilizing GPT-3.5-turbo and GPT-4, and provides additional intermediate reasoning step functionalities for 8,166 of its data points. With a total of 11,128 data points, the dataset covers three types of chart-related generation tasks: description-to-chart, raw data-to-chart, and code-to-description.

搜集汇总

背景与挑战

背景概述

Text2Chart31是一个用于图表生成的数据集，包含31种基于Matplotlib的独特图表类型和11.1K个元组（描述、代码、数据表和图表）。该数据集支持指令调优和强化学习，旨在提升大型语言模型在数据可视化任务中的性能，适用于学术研究和实际应用。