Zenodo10K

该数据集是从Zenodo数字仓库精心挑选的10,448份演示文稿，旨在解决现有演示文稿数据集中存在的局限。该数据集解决了先前数据集中存在的语义信息丢失问题，并包含了来自多个领域的演示文稿，其规模达到了10,448份，任务涉及演示文稿的生成和评估。

This dataset comprises 10,448 presentations carefully curated from the Zenodo digital repository, aiming to address the limitations of existing presentation datasets. It resolves the issue of semantic information loss present in prior datasets, covers presentations from multiple disciplines, and includes 10,448 samples, with tasks involving both presentation generation and evaluation.

搜集汇总

背景与挑战

背景概述

该内容描述的是PPTAgent项目，一个用于自动生成和评估演示文稿的代理框架，支持多种功能如文本到图像生成、上下文管理等，并非传统数据集。项目已开源并发布在GitHub上，拥有4.1k星标和491次分叉。