zzliang/GRIT
Text To ImageImage To TextObject DetectionZero Shot ClassificationEN
Zzliang/GRIT is a text to image dataset in EN from zzliang in Parquet format.
About zzliang/GRIT
GRIT: Large-Scale Training Corpus of Grounded Image-Text Pairs
Dataset Summary
We introduce GRIT, a large-scale dataset of Grounded Image-Text pairs, which is created based on image-text pairs from COYO-700M and LAION-2B. We construc...
Details
- Task
- Text To Image, Image To Text, Object Detection, Zero Shot Classification
- Language
- EN
- Format
- Parquet
- Rows / instances
- N/A
- Creator
- zzliang
- Year
- 2023