Cloud-based AI development IDE with Jupyter Notebook, VS Code, and Terminal. Ships with pre-configured PyTorch and TensorFlow images, supports heterogeneous compute, and mounts OSS, NAS, and CPFS datasets.
Product overview
DSW development environment:
New UI
Classic UI
Benefits
-
Flexible development: Integrates Notebook, VS Code, and Terminal. Supports PyTorch, TensorFlow, and other framework images. Offers public and dedicated compute resources (general-purpose or Lingjun resources).
-
End-to-end workflow: Connects to PAI-DLC for distributed training and PAI-EAS for deployment, covering data processing, debugging, training, and deployment.
-
Cost management: Lifecycle management with scheduled and idle shutdown. Workspace enables global resource allocation and reclamation.
-
Scenario-based examples: Notebook Gallery provides LLM and AIGC tutorials you can use as starting points for your projects.
Core features
Create and manage
-
Create a DSW instance: Select resource type, mount datasets, and specify custom container images.
-
Access and manage DSW from the console: Stop, delete, or reconfigure instances.
-
Use an instance RAM role: Associate a RAM role with an instance to access cloud resources through temporary STS credentials instead of long-term AccessKeys.
Model development environment
-
Manage third-party libraries: Install and manage Python libraries and software.
-
Visualize training with TensorBoard: Visualize metrics and training information with the TensorBoard extension.
-
Deploy a model as an online service: Deploy models as online services via PAI-EAS with auto-scaling, version control, and monitoring.
Data access and mounting
-
Mount a dataset, OSS, NAS, or CPFS: Mount OSS, NAS, or CPFS storage to expand capacity and persist data.
-
Read and write data in OSS: Access OSS data from a DSW instance using APIs or SDKs.
-
Upload and download files: Transfer data and models between your local machine and DSW instances.
Network configuration
-
Connect remotely via SSH: Use DSW compute resources from your local environment through SSH.
-
Improve public network access speed with an Internet NAT Gateway: Bind an EIP via NAT Gateway to accelerate public network access for your instance.
-
Access services in an instance from the internet: Expose instance services to the VPC or public internet for model testing.
-
Pull models or container images from overseas: Configure Global Accelerator to speed up pulls from
docker.ioandhuggingface.co.
Billing
Compute instances
DSW offers public resources and dedicated resources (general-purpose compute or Lingjun), each with different billing methods.
|
Instance type |
Billing method |
Billable item |
Billing rules |
Stop billing |
|
public resource |
Pay-as-you-go |
Runtime of instance using public resources. |
Billed for instance runtime. Important
Billing details: Billed by the minute, bills generated hourly. Due to data aggregation, bills may be delayed 2 to 3 hours. Final bill prevails. |
Stop or delete instance. Important
Stop instances manually or configure scheduled shutdown. For more information, see Manage DSW instances. |
|
dedicated resource (general-purpose compute resource or Lingjun resource) |
Subscription |
Node count and subscription duration. |
Charged based on node count and subscription duration. For more information, see Billing of AI computing resources. |
Unsubscribe from the resource. |
System disk
|
Billing method |
Billable item |
Billing rules |
Stop billing |
|
Pay-as-you-go |
System disk capacity and usage duration. |
Each instance type includes free quota. Capacity expansion is charged. |
Delete instance. |
For more information, see Billing of DSW. To view bills, see View your bill details.
Quick start
The Quick Start for DSW tutorial uses MNIST handwritten digit recognition to walk you through the basics.
Get help
Find solutions to instance startup failures, billing, and network issues in the FAQ about DSW.