taoki/stair-captions-prompts

---
license: cc-by-4.0
dataset_info:
features:
- name: id
dtype: int64
- name: file_name
dtype: string
- name: instruction
dtype: string
- name: output
dtype: string
splits:
- name: train
num_bytes: 30613024
num_examples: 82783
- name: validation
num_bytes: 14889077
num_examples: 40504
download_size: 20826123
dataset_size: 45502101
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: validation
path: data/validation-*
---

# Dataset Card for Dataset Name

This dataset was created to train a model that generates diverse image generation prompts from simple Japanese descriptions.

## Dataset Details

### Dataset Description

[STAIR Captions](https://github.com/STAIR-Lab-CIT/STAIR-captions) + [Microsoft COCO Caption](https://github.com/tylin/coco-caption) + [Microsoft COCO Images caption interpreted by UForm](https://huggingface.co/unum-cloud/uform-gen)

This dataset was created by having UForm explain Microsoft COCO Images of source Microsoft COCO based on the pair of translation inforThis dataset was created to train a model that generates diverse image generation prompts from simple Japanese descriptions.mation STAIR Captions and source information Microsoft COCO Caption.

原始信息汇总

数据集概述

数据集描述

该数据集旨在训练一个模型，该模型能够从简单的日语描述生成多样化的图像生成提示。

数据集信息

• 特征:

  • id: 数据类型为 int64
  • file_name: 数据类型为 string
  • instruction: 数据类型为 string
  • output: 数据类型为 string

• 分割:

  • train: 字节数为 30613024，样本数为 82783
  • validation: 字节数为 14889077，样本数为 40504

• 数据大小:

  • 下载大小: 20826123 字节
  • 数据集大小: 45502101 字节

• 配置:

  • default 配置:
    • train 分割路径: data/train-*
    • validation 分割路径: data/validation-*