Topic Thread

Expand all | Collapse all

Not able to deploy Keras model

  • 1.  Not able to deploy Keras model

    Posted Mon August 26, 2019 02:59 PM
    Edited by Martha Jack Fri September 06, 2019 03:40 PM
    # Deploy the stored model as an online web service deployment
    model_id = model_details["metadata"]["guid"]
    #model_id
    deployment_details = client.deployments.create( artifact_uid=model_id, name="Keras deployment" )

    #######################################################################################
    
    Synchronous deployment creation for uid: 'bbbc3b12-3a88-4f6e-aa9d-f01afc8971d3' started
    
    #######################################################################################
    
    
    INITIALIZING
    DEPLOY_IN_PROGRESS.
    2019-08-26 17:52:59,989 - watson_machine_learning_client.wml_client_error - WARNING - Deployment creation failed. Errors: [{'code': 'load_model_failure', 'message': '0', 'target': {'type': 'none', 'name': 'none'}, 'more_info': 'none'}]

    any suggestion?

    #MachineLearning
    #keras
    #watson-machine-learning
    #machinelearning

    ------------------------------
    MUKESH GUPTA
    ------------------------------
    ​​​​​​​​


  • 2.  RE: Not able to deploy Keras model

    Posted Mon August 26, 2019 04:55 PM
    Hi Mukesh! What version of Keras are you using? Our dream team of data scientists in Markham recommend trying 'version': '2.2.4' in your model properties.  ​

    ------------------------------
    Hafsah Lakhany,
    Product Manager, Watson Studio and Machine Learning
    ------------------------------