launch training on sagemaker

launch_training.py

from sagemaker import get_execution_role
from sagemaker.tensorflow import TensorFlow

role=get_execution_role()

estimator = TensorFlow(entry_point='aws_entry_point.py',
                             role=role,
                             training_steps=1000, 
                             train_instance_count=1,
                             train_instance_type='ml.p2.xlarge',
                             hyperparameters={
                                              'learning_rate': 0.001,
                                              'batch_size' : 32
                                      })
                                      
                                      
                                      
estimator.fit('s3://sagemaker-eu-west-1-107382316743/founder-classifier/data')