config.json

{
  "model_type": "inception_v3",
  "num_labels": 8,
  "image_size": 299,
  "num_channels": 3,
  "input_shape": [299, 299, 3],
  "framework": "tf"
}