Beta
Explore
Marketplace
Neural Labs
Playground
Wallet
Docs
MADQA – Dataset by OxRML | AlphaNeural AI
You can deploy this model and start earning money today!
OxRML
/
MADQA
like
0
visual-document-retrieval
visual-question-answering
en
cc-by-nc-4.0
1K<n<10K
parquet
text
document
datasets
pandas
polars
mlcroissant
2603.12180
us
benchmark
agent
document
multimodal
RAG
Views
No views yet
Model card
Files and Versions
Community
API
MADQA Dataset
An agentic document question-answering benchmark with 2250 questions over a collection of 800 real-world PDF documents spanning multiple domains.
For more details, see the paper, leaderboard, and code.
Splits
Split Questions Description
train 1,550 Training set
dev 200 Development/validation set
test 500 Held-out test set (answers hidden)
Schema
Column Type Description
id string Unique identifier (split/index)… See the full description on the dataset page:
https://huggingface.co/datasets/OxRML/MADQA
.
