Dataset · sklevtsova
ReIV benchmark with original and modified photos
sklevtsova/ReIV
Dataset with original and logically manipulated images of cats and dogs for evaluating Multimodal LLMs.
What's inside
Provides images from the Re-Imagine Vision benchmark to test multimodal language models' ability to account for visual content. Includes baseline images of cats and dogs.
What you can do with it
- Evaluating reasoning and visual attention in multimodal LLMs
- Image generation detection and classification tasks
- Tasks
- Image classification, Image → image
- Languages
- English
- Size
- 10K–100K rows
- Downloads this month
- 38.3K
- Likes
- 0
How to load it
from datasets import load_dataset
ds = load_dataset("sklevtsova/ReIV")