locked
Error 0138: Memory has been exhausted, unable to complete running of module RRS feed

  • Question

  • Hello

    I'm attempting to create a model using a Multiclass Decision Forest.

    My training data consists of 900,000 rows and 5000 features columns. 
    These feature columns are n-grams which are extracted from the Extract N-Gram Feature module.

    The last step in my experiment runs the Evaluate Model module but it throws a "Memory has been exhausted" exception. 

    Now, I understand that Azure ML Studio imposes a 10GB limit on the size of a dataset, however when 
    my dataset is saved to disk its only 220 MB in size. 

    Given its  size on disk is it conceivable that the in-memory size would be over the 10GB limit?

    If so, is there a way around this limit? 

    Saturday, December 9, 2017 2:51 AM

Answers

  • Hi,

    Azure ML Studio is obviously running out of memory while doing the various steps in your experiment. A couple of options would be to reduce the training dataset size and rerun your code within Azure ML Studio environment or use the Azure ML Workbench where the user has multiple options to use VMs hence the datasets can be larger. https://docs.microsoft.com/en-us/azure/machine-learning/preview/quickstart-installation

    Regards,
    Jaya

    • Marked as answer by RowW Tuesday, December 12, 2017 9:46 AM
    Sunday, December 10, 2017 4:15 PM

All replies