Open Inference Protocol Using Curl

Consider the following instructions for using Open Interference Protocol Using Curl.

You can construct the input of the inference call based on the Open Inference Protocol V2:

# cat ./examples/wine-input.json

Your input payload will look something like the following:

{
  "parameters": {
    "content_type": "pd"
  },
  "inputs": [
    {
      "name": "float_input",
      "shape": [
        1,
        11
      ],
      "datatype": "FP32",
      "data": [
        7.4,
        7.4,
        7.4,
        7.4,
        7.4,
        7.4,
        7.4,
        7.4,
        7.4,
        7.4,
        7.4
      ]
    }
  ]
}

curl -H "Content-Type: application/json" -H "Authorization: Bearer ${CDP_TOKEN}" "https://${DOMAIN}/namespaces/serving-default/endpoints/[***ENDPOINT_NAME***]/v2/models/model/infer" -d @./examples/wine-input.json

You will receive response similar to the following:

{
  "model_name":"model",
  "model_version":"1",
  "outputs":[
    {
      "name":"variable",
      "datatype":"FP32",
      "shape":[1,1],
      "data":[5.535987377166748]
    }
  ]
}