class SagemakerClient:
def __init__(self):
self.client = Session(profile_name="default").client("sagemaker", region_name="us-west-2")
def submit_training_job(self):
training_params = {
"TrainingJobName": "sample-training19",
"HyperParameters": {
'objective': 'multiclass',
'num_class': '3'
},
"AlgorithmSpecification": {
'TrainingImage': "562998738767.dkr.ecr.us-west-2.amazonaws.com/test-sagemaker:latest",
'TrainingInputMode': 'File'
},
"RoleArn": "arn:aws:iam::562998738767:role/dev-sagemaker",
"InputDataConfig": [
{
'ChannelName': 'training',
'DataSource': {
'S3DataSource': {
'S3DataType': 'S3Prefix',
'S3Uri': "s3://test-ubuntu-sagemaker/input-data/iris5.csv"
}
}
}
],
"OutputDataConfig": {
'S3OutputPath': "s3://test-ubuntu-sagemaker/output-data/"
},
"ResourceConfig": {
'InstanceType': 'ml.m4.xlarge',
'InstanceCount': 1,
'VolumeSizeInGB': 10
},
"StoppingCondition": {
'MaxRuntimeInSeconds': 60 * 60
}
}
response = self.client.create_training_job(**training_params)
print(response)