dataset_config.json

{
  "default": {
    "description": "LLAVA-compatible brain tumor dataset in JSONL format with image-text pairs for multimodal training.",
    "data_files": {
      "train": "train.jsonl"
    }
  }
}