To create a custom image classification model, we need to use a graphics processing unit (GPU) enabled training job instance. GPUs are excellent at parallelizing the computations required to train a neural network for this project. In order to access a GPU-enabled training job instance, you must submit a request for a service limit increase to the AWS Support Center. You can follow the instructions [here]() to increase your limit. For this recipe we will use a single ml.p2.xlarge instance.
Open the AWS Support Center console.
On the AWS Support Center page, choose Create Case and then choose Service limit increase.
In the Case classification panel under Limit type, search for Amazon SageMaker.
In the Request panel, choose the Region that you are working in. For Resource Type, choose SageMaker Training.
For Limit choose ml.p2.xlarge instances.
For New Limit Value, verify that the value is 1.
In Case description, provide a brief explanation of why you need the Service limit increase. For example, I need to use this GPU-enabled training job instance to train a deep learning model using TensorFlow. I’ll use this model on an AWS DeepLens device.
In Contact options, provide some details about how you would like to be contacted by the AWS service support team on the status of your Service limit increase request.