normalization data: untrained model： trained model: scored value output: evaluate model: create realtime inference: inference deployment to instance: final svc endpoint: the final model that used in the inference service:

scored labels result: evaluation results: running logs: [crayon-6779b1773c367336370019/]