docvqa_metadata.json

{
  "dataset_name": "DocVQA",
  "source": "lmms-lab/DocVQA",
  "target_size_k": 2,
  "actual_size": 2000,
  "sampling_strategy": "hard_mining",
  "original_data_path": "data/docvqa"
}