Fixing bug in mlrun serving (#2872)

We've updated the code so we're updating documentation here as well.

Author: alexiguazio

content/en/docs/external-add-ons/serving/mlrun.md

@@ -99,7 +99,7 @@ def kfpipeline():
         outputs=['model', 'test_set'])
 
     # deploy our model as a serverless function, we can pass a list of models to serve 
-    deploy = mlrun.import_function('hub://sklearn_classifier').deploy_step(
+    deploy = mlrun.import_function('hub://v2_model_server').deploy_step(
         models=[{"key": "mymodel:v1", "model_path": train.outputs['model']}])
 
     # test out new model server (via REST API calls)