{
  "architectures": [
    "VisionEncoderDecoderModel"
  ]
}