Learn how to activate PAI, explore its core modules, and follow typical AI development workflows.
1. Activate PAI
Log on to the PAI console, select a region in the upper-left corner, and click Activate. A default workspace is created automatically.
We recommend that you use an Alibaba Cloud account to activate PAI. This prevents activation failures caused by a lack of the AliyunPAIFullAccess permission.
2. Get started
-
Start with a single module. Choose one from the table below based on your needs.
-
Read the module's quick start guide to learn the basics through examples.
-
Explore the module's user guide for detailed features and best practices.
Click a module name in the Feature module column to open its user guide.
|
Feature module |
Description |
Quick start |
|
Cloud-based IDE (an instance) for AI development. Supports Notebook and VSCode interfaces. |
||
|
Offers 140+ built-in algorithm components for low-code, drag-and-drop model building. |
||
|
Creates distributed or single-node training jobs without manual machine provisioning or environment setup. Mirrors the local training experience. |
||
|
Deploys trained models as online inference services with minimal configuration. |
||
|
Trains and deploys open-source large models through DLC and EAS without code. |
3. Get help
Find common questions and solutions for PAI modules in the following FAQ documents.
4. AI development workflows
PAI covers the full AI lifecycle: data preparation, model training, and deployment. Two typical workflows are described below.
Cloud-native AI development
|
Step |
Description |
Documentation |
|
① |
Centrally manage local, cloud, and public datasets as training data sources. |
|
|
② |
Cloud-based IDE (an instance) for AI development. Supports Notebook and VSCode interfaces. |
|
|
③ |
Images provide runtime environments. PAI manages both official public images and custom images centrally. |
|
|
④ |
After you develop and test model code in DSW, use DLC to run training jobs more efficiently and cost-effectively. |
|
|
⑤ |
Mount file systems (NAS, OSS) and Git repositories to specify data and code for training jobs. |
|
|
⑥ |
Centrally manage trained models and deploy them directly through EAS. |
|
|
⑦ |
Deploy trained models as online inference services through EAS. |
AI and big data development
|
Step |
Description |
Documentation |
|
① |
If you use MaxCompute to store data, preprocess the data in DataWorks and then reference the MaxCompute table in PAI as a training data source. |
|
|
② |
Offers 140+ built-in algorithm components for low-code, drag-and-drop model building. |
|
|
③ |
Use DataWorks to configure and run periodic scheduling jobs. |
|
|
④ |
Records execution details for Designer experiments and custom jobs for comparison and analysis. |
|
|
⑤ |
Centrally manage trained models and deploy them directly through EAS. |
|
|
⑥ |
Deploy trained models as online inference services through EAS. |
FAQ
Activation permission issue
A: Do one of the following:
-
Use an Alibaba Cloud account to activate the service.
-
If you are a RAM user, you must attach the AliyunPAIFullAccess system policy to the RAM user. Note: This policy grants extensive permissions. The owner of the Alibaba Cloud account must assess the security risks.
Q: Why do I receive the error message Create order error: message is Your account's available credit is less than 0. Please top up your account before trying to purchase. productRequestId is *** when I enable PAI?
A: Go to User Center to view your bills and pay any outstanding fees. Then, reactivate PAI.
Q: An error occurred: Create order error: message is Pre-order validation failed productRequestId is 8F5?
A: Insufficient permissions. Use an Alibaba Cloud account or a RAM user with the AliyunPAIFullAccess permission.