Beta
Explore
Marketplace
Neural Labs
Playground
Wallet
Docs
Omni-Bench – Dataset by ModalityDance | AlphaNeural AI
You can deploy this model and start earning money today!
ModalityDance
/
Omni-Bench
like
0
visual-question-answering
n<1K
parquet
text
datasets
pandas
polars
mlcroissant
2601.09536
us
benchmark
multimodal
evaluation
reasoning
Views
No views yet
Model card
Files and Versions
Community
API
Omni-Bench Overview
Omni-Bench is an evaluation benchmark for unified multimodal reasoning. It contains 800 samples spanning 4 Uni-Tasks:
Natural-Scene Perception: V* Structured-Image: ArxivQA, ChartQA Diagrammatic Math: Geometry3k, MathVista Vision-Operational Scenes: ViC-Bench
Data Fields
Each example contains the following fields:
image (string): the image encoded as a Base64 string.The underlying bytes are typically common image formats (e.g.… See the full description on the dataset page:
https://huggingface.co/datasets/ModalityDance/Omni-Bench
.
