definition_template.yaml

version: 1 # integer version number of the definition.yaml format
enable_testing: # true|false, set to true to enable the automatic testing of this network/model
network:
  filename: # relative path to the model file
  framework: # TensorFlow, TensorFlow Lite, ONNX, ...
  provenance: # URL to where the base model come from. Modifications should be explained in the description field
  training: # Is it trained? Fully? On which data? Link to a page that describes this. Make distinction of the training of the upstream base model and additional training by you.
  model_license: # link or identifier to the license of the model file
  data_license: # link or identifier to the license of the dataset used during training
network_parameters:
  datatype: # The type of of the network float32 or int8
  input_nodes:
    - name: # Name of the input node without a tensor specification (:0 is assumed)
      description: # Brief description of this input node e.g. purpose and expected data format
      shape: # A list of dimensions. Use -1 for variable sizes, e.g:
        - 1
        - -1
        - -1
        - 3
  output_nodes:
    - name: # Name of the output node without a tensor specification (:0 is assumed)
      description: # Brief description of this output node. Link to output processing pipeline code or description of how such a pipeline can be written.
      shape: # A list of dimensions. Use -1 for variable sizes, e.g:
        - 1
        - -1
        - -1
        - 3
    - name:
      description:
      shape: # A list of dimensions. Use -1 for variable sizes, e.g:
        - 1
        - -1
        - -1
        - 3