Please enable JavaScript.

Coggle requires JavaScript to display documents.

AWS Model Updation, from sagemaker.amazon.amazon_estimator import get…

- - - - xbb = sagemaker.estimator.Estimator()
        
        xgb_contrainer
        
        role
        
        train_instace_count = 1
        
        train_isntance_type = 'ml.m4.xlarge'
        
        output_path = 's3://{}/{}'.format(session.default_bucket(),s3_folder_prefix)
        
        sagemaker_session = sagemaker.Session()
        
        set hyperparameters
        
        xgb.set_hyperparameters()
        
        hyper parameters for xgboost
        
        Train the model
        
        xgb.fit({'train' : s3_train_input, 'validation' : s3_validation_input)
        
        s3_train_input = sagemaker.s3_input()
        
        s3_data = s3_train_location
        
        content_type = 'text/csv'
        
        This doesn't create any model until the container/estimator is deployed or endpoint is created for it
        
        use training model artifacts to create a model
        
        xgb_primary_container =
        
        {'Image' : xgb_contrainer,
        'ModelDataUrl' : xgb,model_data}
        
        2 more items...
        
        xgb_model_info = session.sagemaker_client.create_model()
        
        3 more items...
        
        Create Endpoint
        
        xgb_endpoint_info = session.sagemaker_client.create_endpoint_config()
        
        3 more items...
- - - - container = linear_container
      - role = sagemaker.get_execution_role()
      - train_instance_count = 1
      - train_instance_type = 'ml.m.xlarge'
      - output_path = 's3://{}/{}'.format(session.default_bucket(),s3_folder)
      - sagemaker_session = sagemaker.session()
      - linear_estimator.set_hyperparameters()
        
        feature_dim = 13
        predictor_type = 'regressor'
        mini_batch_size = 200
        
        linear_estimator.fit()
        
        {
        'train' : sagemaker.s3_input
        (
        s3_data = input_data,
        content_type = 'text/csv'
        )
        'validation' : sagemaker.s3_input (
        s3_data = valiodation_data
        content_type = 'text/csv'
        )
        }
        
        linear_primary_contrainer =
        
        {
        ""Image" : linear_container
        "ModelDataUrl" : linear_estimator.model_data
        }
        
        linear_model_info = session.sagemaker_client.create_model()
        
        ModelName = unique_model_name
        ExecutionRoleArn = sagemaker.get_execution_role()
        Primarycontainer = linear_primary_container
        
        linear_endpoint_config_info = session.sagemaker_client.create_endpoint_config()
        
        EndpointConfigName = unique_name
        
        ProductionVariants =
        
        1 more item...
        
        linear_endpoint_info = session,sagemaker_client.create_endpoint()
        
        3 more items...
  - - - combined_endpoint_info =session.sagemaker_client.create_endpoint()
        
        EndpointName = unique_endpoin_nam
        
        EndpointConfigName = unique_endpoint_config_name
        
        session,sagemaker_clinet.wait_for_endpoint(unique_endpoint_name)
        
        response = session.sagemaker_runtime_client.invoke_endpoint()
        
        EndpointName = unique_endpoint_name
        
        ContentType = 'text/csv'
        
        Body = ' , '.join(map(str,test.values[10]))
        
        respoce['Body'].read().decode('utf-8')
        
        session.sagemaker_client.describe_endpoint(EndpointName = unique_endpoint_name)
        
        session.sagemaker_client.update_endpoint(EndpointName = unique_endpoint_name, EndpointConfigName = unique_config_name)
      - EndpointConfigName = unique_config_info_name
      - ProdcutionVariants = [ ]
        
        {
        "InstanceType" : "ml.m4.xlarge",
        "InitiallVariantWeight": 1
        "InitialInstancecount" : 1
        "ModelName" : unique_linear_model_name
        "VariantName" : "linear_model"
        }
        
        {
        "InstanceType" : "ml.m4.xlarge",
        "InitialVariantWeight" : 1
        "InitialInstanceCount" : 1,
        "ModelName": unique_xgb_model_name
        "VariantName" : "xgb_model"
        }