Dataset · Spawning
PD12M: 12.4 million image-text pairs
Spawning/PD12M
The largest public domain image-text pair dataset for training foundation models.
What's inside
12.4 million English image-caption pairs. According to the dataset card, this is the largest public domain dataset of its kind; its size is sufficient for training foundation models while minimizing copyright issues. Collected via the Source.Plus platform with community governance aimed at harm reduction and reproducibility.
What you can do with it
- Training multimodal models
- LoRA for image captioning
- Evaluating image-text pair quality
- Languages
- English
- Size
- 10M–100M rows
- Downloads this month
- 7.2K
- Likes
- 197
Models trained on it
- Mindscend/mindscend-image-0.3Unconditional image generation 25
- Mindscend/mindscend-image-0.2Unconditional image generation 20
- morrisgaby987/suite 0
- betzabetmilagrosul/roberta-baseTranslation 0
- aicoder69/aicoder69Multimodal 0
- laidinir/Axxa.SocialText generation 0
- Landon93737373/UshshshnaText generation 0
- Skepsy88/AwText → audio 0
How to load it
from datasets import load_dataset
ds = load_dataset("Spawning/PD12M")