Deploying ML Models via Restful API

ML models can be deployed programmatically using the Restful API. The Restful API details are also available on the “Deploy” page of MLOps. To see the API details for model deployment, click the “Deploy” menu option and scroll down to the bottom of the page to see the CuRL command and python code. 

Figure 4.2 shows as a screenshot. 

Figure 4.2: Screenshot showing the Curl and Python code for Restful API for model deployment 

Using the command line or CURL:  

curl -X PUT -H ‘Content-Type:multipart/form-data’ -H ‘Authorization: Token 1234567890987654321’ -F ‘model_file=@myfile.xml’ http://one.accure.ai:443/mlops/v1/model_upload/model_name/cluster_id/model_category 

Using Python code: 

url = http://one.accure.ai:443/mlops/v1/model_upload/model_name/cluster_id/model_category  

headers = {“Authorization”: “Token 1234567890987654321”} 

files = {“model_file”: open(r”myfile.xml”, “rb”)} 

r = requests.put(url=url, headers=headers, files=files) 

print(r.json()) 

Replace the token with the real security token.  

Replace the “myfile.xml” by the model file path. 

Replace the URL with your MLOps deployed domain and port. 

model_name: replace with the model name you are deploying 

cluster_id: Use appropriate cluster_id (default_cluster by default) 

model_category: Depending on the category of the model you are deploying. The supported categories are: 

classificateion 

clustering 

image_classification 

object_detection 

nlp 

regression