{
  "data_designer": {
    "columns": [
      {
        "name": "instruction",
        "drop": false,
        "allow_resize": false,
        "column_type": "seed-dataset"
      },
      {
        "name": "input",
        "drop": false,
        "allow_resize": false,
        "column_type": "seed-dataset"
      },
      {
        "name": "output",
        "drop": false,
        "allow_resize": false,
        "column_type": "seed-dataset"
      },
      {
        "name": "topic",
        "drop": false,
        "allow_resize": false,
        "column_type": "seed-dataset"
      },
      {
        "name": "task_type",
        "drop": false,
        "allow_resize": false,
        "column_type": "seed-dataset"
      },
      {
        "name": "source_file",
        "drop": false,
        "allow_resize": false,
        "column_type": "seed-dataset"
      }
    ],
    "seed_config": {
      "source": {
        "seed_type": "curated-exam-ready"
      },
      "sampling_strategy": "balanced-by-lab-topic"
    }
  },
  "library_version": "0.5.4-compatible",
  "note": "Rebuilt for Unsloth Alpaca-style fine-tuning for ML/Python exam preparation."
}