Configuring Deployment
Cluster Deployment is the process of integrating a trained model into the cluster's environment for inference or other applications. Once the model is trained, deployment involves configuring it to make predictions or provide results by utilizing the computational resources within the cluster.
Deployment Types
There are several deployment types you can choose from based on your use case and technology stack. Below are the supported deployment types:
Create Deployment
- To create a new Deployment, click the Create Deployment button.
Select the desired Framework by choosing the appropriate Cluster with the required CPU, GPU, and an available plan. Click on Next.
Select the desired Image according to your requirement.
Select the appropriate SSH Keys or create new one. Click on Next.
Select the Shared File-System and Dataset according to your requirement or create new one. Click on Finish.
Review the details you have selected and make any necessary changes by clicking on Edit. Click on Create.
Manage Deployment
Overview
You can view the details of the selected Deployment, including the Deployment Name, Number of Workers, SSH Keys, and the Deployment Image. By clicking Overview.
Worker Management
You can view the details of Workers created for your Deployment and view logs and connect to Master Node.
Connect
You can view connect details by clicking connect icon.
Logs
You can view Logs in details by clicking Logs. And select the worker you want view logs.
Volume
You can view Volume in details by clicking volume.
Volume
You can view added ssh keys in details by clicking User Management.
Monitoring
You can view the Memory Usage for the Workers within the Deployment. Additionally, the following metrics are also available: GPU Utilization, GPU Temperature, CPU Utilization, Memory Utilization.