Documentation Overview
To help you fully understand the core concepts, usage methods, and best practices of Neutree, please refer to the following sections.
-
For a quick hands-on experience with platform features, the Quick Start section will guide you through deploying a minimal version of the AI platform on a single node and creating your first AI workload.
-
To understand the complete deployment details, including production-grade deployment and more deployment options, please refer to Deploy AI Platform to complete a full Neutree deployment.
-
After completing the installation and deployment, set up the infrastructure:
-
Create a container image registry to provide the required container images for cluster deployment.
-
If cluster nodes cannot access Docker Hub, you can manually import images.
-
-
After completing the infrastructure setup, manage model and inference-related resources:
-
Create a model registry to store models.
-
Use command-line tools to manage models in the model registry.
-
The platform provides a model catalog, which you can use when creating endpoints to select appropriate models and simplify the configuration process.
-
Create an inference endpoint. The platform will load the specific model from the selected model registry for use by the inference engine.
-
-
In addition to running AI workloads within the platform, you can also integrate Neutree from external systems using API keys.
-
Administrators can read the User and Permission Management section to learn about user management, permission control, and how to implement multi-tenant isolation through workspace policies.
-
The platform also provides a series of auxiliary tools and configuration features, such as customizing platform appearance, importing YAML, and exporting YAML, to help you simplify management and operations tasks and achieve personalized configurations.