maharshipandya/synthanime-openhermes2.5

---
license: apache-2.0
task_categories:
- text2text-generation
- text-generation
pretty_name: Instruction-Response anime synopsis data
size_categories:
- 10K<n<100K
---

# What is this dataset?

This is a hybrid **Instruction-Response** dataset (scraped + synthetic) for anime synopses.
Given around 10,000 scraped anime synopses, the user instructions for this dataset were generated using [Teknium's Openhermes 2.5](teknium/OpenHermes-2.5-Mistral-7B)

- The `assistant` column consists of synopsis for different animes (which were previously scraped)
- The `user` column consists of the instructions that "might have" generated the synopsis (synthetically generated)

The goal of this dataset is: to be used as a part of a much larger dataset in order to fine tune open-source LLMs to follow instructions better and also to have some more knowledge about anime.

**Using open-source LLMs to make open-source LLMs better** 🫶

原始信息汇总

数据集概述

数据集名称

Instruction-Response anime synopsis data

许可证

apache-2.0

任务类别

• 文本到文本生成
• 文本生成

数据集大小

10K<n<100K

数据集描述

这是一个混合型的指令-响应数据集（爬取+合成），用于动漫剧情简介。该数据集包含约10,000条爬取的动漫剧情简介，用户指令是通过Tekniums Openhermes 2.5生成的。

• assistant列包含不同动漫的剧情简介（之前爬取的）
• user列包含可能生成这些剧情简介的指令（合成生成）

数据集目标

该数据集旨在作为更大数据集的一部分，用于微调开源大型语言模型（LLMs），以更好地遵循指令并增加对动漫的了解。