cat > datasets/flowvqa/README.md <<'EOF'
pretty_name: FlowVQA RAG A-subset
-
rag
-
flowchart
-
vqa
task_categories:
-
text-retrieval
-
visual-question-answering
license: mit
language:
-
en
source_datasets:
-
flowvqa
corpus.parquet — columns: corpus_id: string, image: binary (image bytes)
queries.parquet — columns: query_id, query, answer, options, is_numerical… See the full description on the dataset page:
https://huggingface.co/datasets/kkyzl/RAG_Dataset_2.