History

Thang Minh Vu 328edd8117 fix(components): make inputs.model_artifact_url optional in sagemaker model component (#8336 ) * fix(components): make inputs.model_artifact_url optional in sagemaker model component * chore: run black * Fixed Stop bug commit f2092382ee941c2f33935db3e886093a15f103f7 Author: ananth102 <abashyam@amazon.com> Date: Fri Oct 7 19:51:55 2022 +0000 replaced image commit 2f0e2daa54fe80a3dfc471d393be62d612217b84 Merge: `bf2389a66` `7ce165432` Author: ananth102 <abashyam@amazon.com> Date: Fri Oct 7 19:50:28 2022 +0000 Merge remote-tracking branch 'stopfix/handle_stopped' into kfpv1fixes2 commit `7ce165432e` Author: Kartik Kalamadi <kalamadi@amazon.com> Date: Thu Mar 3 09:58:16 2022 -0800 Run black commit `32d6e1388a` Author: Kartik Kalamadi <kalamadi@amazon.com> Date: Tue Mar 1 15:25:32 2022 -0800 Change image for testing commit `7875d9aa27` Author: Kartik Kalamadi <kalamadi@amazon.com> Date: Mon Jan 31 09:29:50 2022 -0800 Handle Stopped state for all components and fix bug in robomaker simulation function * chore(docs): Update model README.md Update README * updated image and liscense * chore: pop ModelDataUrl if not exist * fix: make field as option in aws batch_transform component chore: run black chore: revert docker version pump up chore(docs): update valid instance types Remove key if not use Pop KmsKeyId * update changelog * chore: pop DataProcessing if no value supplied * test(components): Update test * fix(batch_transform): only pop input and output * fixed log bug Co-authored-by: ananth102 <abashyam@amazon.com>		2022-10-14 22:12:49 +00:00
..
src	refactor(components): AWS SageMaker - Full component refactoring (#4336 )	2020-10-27 14:17:57 -07:00
README.md	feat(components): AWS SageMaker - Support for assuming a role (#4212 )	2020-08-03 10:53:43 -07:00
component.yaml	fix(components): make inputs.model_artifact_url optional in sagemaker model component (#8336 )	2022-10-14 22:12:49 +00:00

README.md

SageMaker create private workteam Kubeflow Pipelines component

Summary

Component to submit SageMaker create private workteam jobs directly from a Kubeflow Pipelines workflow.

Details

Intended Use

For creating a private workteam from pre-existing Amazon Cognito user groups using AWS SageMaker.

Runtime Arguments

Argument	Description	Optional	Data type	Default
region	The region where the cluster launches	No	String
endpoint_url	The endpoint URL for the private link VPC endpoint	Yes	String
assume_role	The ARN of an IAM role to assume when connecting to SageMaker	Yes	String
team_name	The name of your work team	No	String
description	A description of the work team	No	String
user_pool	An identifier for a user pool, which must be in the same region as the service that you are calling	No	String
user_groups	An identifier for user groups separated by commas	No	String
client_id	An identifier for an application client, which you must create using Amazon Cognito	No	String
sns_topic	The AWS KMS key that Amazon SageMaker uses to encrypt the model artifacts	Yes	String
tags	Key-value pairs to categorize AWS resources	Yes	Dict	{}

Notes:

Workers in private workteams per account and region may come from only one Amazon Cognito user pool. However, you may have several user groups within the user pool to define different workteams.
Your Amazon Cognito user pool must be in the same region that you are creating the Ground Truth job in.

Outputs

Name	Description
workteam_arn	ARN of the workteam

Requirements

Kubeflow pipelines SDK
Kubeflow set-up on AWS
User pool, user groups, and app client ID set up on Amazon Cognito
1. Create a user pool in Amazon Cognito. Configure the user pool as needed, and make sure to create an app client. The Pool ID will be found under General settings.
2. After creating the user pool, go to the Users and Groups section and create a group. Create users for the team, and add those users to the group.
3. Under App integration > Domain name, create an Amazon Cognito domain for the user pool.