Dataset · Ritabrata04
Synthetic corpus for object counting in VLM
Ritabrata04/masscount-cf
A synthetic counterfactual corpus of images and scene graphs to study numerals in VLMs.
What's inside
The dataset includes about 326,000 images and scenes with over 28 million objects in English. The corpus is designed to study the cardinal coordinates of neural masses in multimodal models.
What you can do with it
- Study vision and language capabilities for counting and causality
- Train and validate multimodal models (VLM)
- Size
- 100K–1M rows
- Downloads this month
- 114.5K
- Likes
- 1
How to load it
from datasets import load_dataset
ds = load_dataset("Ritabrata04/masscount-cf")