config.json

{
  "task_type": "text-to-image",
  "inference_framework": "pytorch",
  "inference_config": {
    "task_type": "text-to-image",
    "runtime": "docker",
    "docker_config": {
      "use_dockerfile": true
    }
  }
}