Google Cloud ML Engine: Hyperparameter Tuning Fails to Load Tensorflow Backend

Question

I'm using Cloud ML Engine to tune my model's hyperparameters. I'm looking at the logs it produces and I see an Error that says:

Using Tensorflow Backend

It says the severity is Error and this is the jsonPayload:

 jsonPayload: {
  created:  1536552469.856194   
  levelname:  "ERROR"   
  lineno:  326   
  message:  "Using TensorFlow backend."   
  pathname:  "/runcloudml.py"   
 }

I'm using a Keras model with a Theano backend but I do import Tensorflow in my train function.

The logs, unfortunately, don't provide much additional detail. Any idea why ML Engine would have trouble using Tensorflow?

Thanks!

lwz1992 lwz1992 · Accepted Answer · 2018-09-11T18:29:49

I think the log you posted comes from here. It doesn't indicate a real error. If your job failed, it should be caused by other issues. And you can send your project number as well as job id to [email protected] if you get trouble to debug it.

Google Cloud ML Engine: Hyperparameter Tuning Fails to Load Tensorflow Backend

1 Answers