Dataset · diffusers
Pokemon images with detailed GPT-4 descriptions
diffusers/pokemon-gpt4-captions
A small dataset of Pokemon images (under 1,000 items) re-captioned with GPT-4 Turbo for image generation.
What's inside
This is an updated version of the well-known lambdalabs/pokemon-blip-captions dataset, replacing standard captions with descriptions generated by the GPT-4 Turbo model. It contains English image-text pairs with fewer than 1,000 records.
What you can do with it
- Training stylized LoRA models for image generators (Stable Diffusion, Flux)
- Quick testing of text-to-image model fine-tuning pipelines
- Comparing generation quality when training on short versus detailed prompts
- Tasks
- Text → image
- Languages
- English
- Size
- up to 1K rows
- Downloads this month
- 1.6K
- Likes
- 43
Models trained on it
How to load it
from datasets import load_dataset
ds = load_dataset("diffusers/pokemon-gpt4-captions")