Data Sources

Data Sources connect external storage to your Run:AI workloads. They appear as mounted directories inside your containers.

Pre-configured Data Sources

Your lab environment includes these data sources:

Name	Type	Mount Path	Contents
lab-datasets	PVC	/mnt/datasets	Pre-loaded training datasets
shared-models	PVC	/mnt/models	Pre-trained model checkpoints

Creating a Custom Data Source

1. Go to Data Sources in the Run:AI UI
2. Click New Data Source
3. Choose your storage type:
4. PVC - Kubernetes Persistent Volume Claim
5. S3 - Amazon S3 bucket
6. NFS - Network File System
7. Git - Clone a Git repository

Attaching Data Sources to Workloads

When creating a workspace or training job:

1. Scroll to the Data Sources section
2. Click Add Data Source
3. Select the data source and specify the mount path

Data sources are mounted read-only by default. To write results, use your project's home directory at /home/user/.