No applicable method for 'predict' applied to an object of class "MXFeedForwardModel"

2
votes

I have created a Neural network model using mxnet package in R studio. I tested the model on local and it works as expected. I have deployed the same model as a webservice in AzureML using publishwebservice() function from R.

When I try to predict the test data with the webservice using consume() function: 

pred_cnn <- consume(endpoint_cnn, testdf)

it always throws following error:

Error: AzureML returns error code: HTTP status code : 400 AzureML error code : LibraryExecutionError

Module execution encountered an internal library error.
The following error occurred during evaluation of R script: R_tryEval: return error: Error in UseMethod("predict") :
no applicable method for 'predict' applied to an object of class "MXFeedForwardModel"

1
razure-machine-learning-studiomxnet
The first point of failure is HTTP status code: 400 which implies a "bad request", output of traceback() could shed more light – Silence Dogood
Cough - gentle reminder that we prefer not to see mentions of urgency here. Remember that it's volunteers that are reading your questions. – halfer

1 Answers

1
votes

I believe the Azure Machine Learning Service platform is running services in windows containers that might not be compatible with the compiled version of MXNet you are uploading with your service.

I would recommend a different operationalization platform, for example if you prefer to remain in the Microsoft ecosystem have a look at Microsoft R Server Operationalization.