eval_set_summary.json

{
  "name": "indic_deva_eval",
  "description": "Broad Indic Devanagari OCR benchmark across printed pages, digits, word crops, and handwriting.",
  "task": "indic_devanagari_ocr",
  "output_dir": "data/eval_ocr/indic_deva_eval",
  "ocr_rows": 1068,
  "sharegpt_rows": 1068,
  "image_errors": 0,
  "images_per_dir": 5000,
  "outputs": {
    "ocr_jsonl": "data/eval_ocr/indic_deva_eval/indic_deva_eval.ocr.jsonl",
    "sharegpt_json": "data/eval_ocr/indic_deva_eval/indic_deva_eval.sharegpt.json",
    "dataset_info": "data/eval_ocr/indic_deva_eval/dataset_info_fragment.json",
    "metadata_jsonl": "data/eval_ocr/indic_deva_eval/indic_deva_eval.metadata.jsonl"
  }
}