The goal of this repository is to provide a ready-to-use example on how to use Google ML engine. For that I will use a dataset provided by kaggle.
For this code to work, you need to set the following environment variables:
PROJECT_ID: Your main Google project ID
BUCKET_NAME: GCS bucket in where to store the models
REGION: Region for Compute Engine
Define the following variables:
TRAINING_PACKAGE_PATH="[YOUR-LOCAL-PATH-TO-TRAINING-PACKAGE]/trainier/"
MAIN_TRAINER_MODULE=trainer.train