The deployment to API for hosting production systems solution supports models built on the platform for both single as well as batch requests.
The API is called by sending an HTTP POST to the endpoint shown in the interface. The request body needs to be multipart-form encoded or json.
A deployment token is required to authenticate the calls. The token is valid from the moment a new deployment is created.
When creating a deployment you can decide which experiment and checkpoint to be used in the deployment.
You can control whether a deployment is enabled for requests and then disable it whenever needed, just toggle the Enable switch.
A deployment can be re-enabled and disabled several times and can be deleted when it’s not relevant anymore. Note that you have to disable the deployment before you can delete it
Make sure input parameters of the request match the expected format and type in order to get a successful prediction. Adjust the parameters naming if needed for the service calls.
Both single requests, as well as batch requests can be sent to the API. Note that if the input batch request contains faulty input samples, the whole response will fail.
The Peltarion Platform provides example code. The curl-example shown is a good way to test out the deployment functionality. Select the right language format and expose the information needed to send requests to the model using a REST API. Currently available curl and JSON format examples.
|If you want to post a file, remember that curl uses the "@"-prefix, e.g., -F “Image=@my_image.png”|
The Token is valid for the created deployment permanently. A deployment can be enabled and disabled as needed. The deploy to API is not responding with predictions when the deployment is disabled.