0

I'm trying to run a basic tensorflow lstm on aws. I have a basic EC2 ubuntu instance set up with tensorflow, as well as a python script I want to run. My script gets all the way to training (making model and processing data is ok) but then gives me this error: tensorflow.python.framework.errors_impl.ResourceExhaustedError: OOM when allocating tensor with shape[1024,1024] and type float on /job:localhost/replica:0/task:0/device:CPU:0 by allocator cpu

I think this might be an error with the EC2 setup, because the same code worked on my own machine(though I was using jupyter notebook). However, this is my first time using aws and therefore I am unfamiliar with it. Is this an error with my AWS setup and if so how do I fix it?

Pseudo Nym
  • 175
  • 1
  • 5
  • What type of EC2 instance are you using? What are the specs? Looks like it doesn't have enough processing power, or too little memory. – Karl Dec 27 '18 at 21:14
  • its a t2.micro running ubuntu. How would I add processing power or memory? Could I augment this one or do I have to create a new one? – Pseudo Nym Dec 27 '18 at 21:26
  • Is it likely some of your processing is failing to release resources when a task is completed? OOM is an Out Of Memory condition because resources have been exhausted. From wikipedia, Apache is referenced implying you have an Apache configuration file that may need to be reviewed and adjusted for reasonable limits. – Wilson Hauck Dec 27 '18 at 21:35
  • Does this answer your question? [Understanding the ResourceExhaustedError: OOM when allocating tensor with shape](https://stackoverflow.com/questions/46066850/understanding-the-resourceexhaustederror-oom-when-allocating-tensor-with-shape) – Hagbard Aug 05 '21 at 09:28

0 Answers0