{0xc00044b600 0xc0004cf0c0} Prepare training dataset :: Distributed training with Amazon SageMaker / Amazon EKS Workshop

Prepare training dataset

Download the CIFAR10 dataset and upload it to Amazon S3

On a terminal window in JupyterLab client, navigate to the notebook directory

cd ~/SageMaker/distributed-training-workshop/notebooks/

Activate the TensorFlow conda environment

source activate tensorflow_p36

Download CIFAR10 dataset and convert it to TFRecords format

python generate_cifar10_tfrecords.py --data-dir dataset

Confirm that the dataset was downloaded successfully. Run:

sudo yum install tree -y
tree dataset

You should see the following output

├── eval
│   └── eval.tfrecords
├── train
│   └── train.tfrecords
└── validation
    └── validation.tfrecords

Create a new S3 bucket and upload the dataset to it. Be sure to add a unique identifier, such as your name.

aws s3 mb s3://<your_bucket>

Note: Bucket names should be unique globally. If a bucket with the same name already exists, add another unique identifier such as today’s date or your last name.

Proceed only if you don’t see an error. Now, upload the dataset to S3

aws s3 sync dataset/ s3://<your_bucket>/cifar10-dataset/