All Products
Search
Document Center

Authorization of RAM

Last Updated: Mar 18, 2020

If you want to use a Resource Access Management (RAM) user account to manage instances and train models, you must first use your Alibaba Cloud account to log on to the RAM console and authorize the RAM user account. Log on to the RAM console

  1. Note: Currently two RAM console versions are available. Before you create a RAM role, switch the console to the new version, as shown in the following figure.

Create a RAM role

Log on to the RAM console. In the left-side navigation pane, click Policies under Permissions, and then click Create Policy.

image.png

Create a custom policy for Auto Learning

On the Create Custom Policy page, enter the policy name autolearningfullaccess, set the configuration mode to Script, and then enter the script content, as shown in the following figure.

image.png

  • Add the following permissions:
Permission Description
CreateInstance The permission to create and delete instances.
EditInstance The permission to edit instances.
LabelInstance The permission to create, delete, and modify labels, and add labels to images.
TrainInstance The permission to train models.
ListInstance The permission to query instance information.
  • The authorization policy script is as follows:
  1. {
  2. "Version": "1",
  3. "Statement": [
  4. {
  5. "Effect": "Allow",
  6. "Action": [
  7. "autolearning:CreateInstance",
  8. "autolearning:LabelInstance",
  9. "autolearning:TrainInstance",
  10. "autolearning:EditInstance",
  11. "autolearning:ListInstance"
  12. ],
  13. "Resource": "*"
  14. }
  15. ]
  16. }

Authorize the RAM role

  1. In the left-side navigation pane, click Users, select your RAM user account, and then click Add Permission in the Action column, as shown in the following figure.image.png

  2. On the Add Permissions page, select Custom Policy, enter the policy name autolearningfullaccess, and then click the policy to add it to the right-side list.image.png
    After you click OK, your RAM user account is authorized to manage instances in Auto Learning.