Create a training preset

A training preset is a saved configuration that combines a runtime, compute sizing, and optional defaults. Presets let you standardize training job configurations across your team and speed up job submission.

Prerequisites

• At least one training runtime configured
• Admin permissions on the workspace

View presets

1. Click Workbench in the left sidebar.
2. Navigate to Training Jobs > Presets.
3. The page shows three sections:
   • Named presets -- full configurations (runtime + sizing + defaults)
   • Sizing presets -- compute-only configurations (nodes, CPU, memory, GPU)
   • TTL and defaults -- default time-to-live and resource limits

Create a named preset

1. On the Presets page, click Create Preset.
2. Configure the preset:
   • Key -- a unique identifier (for example, pytorch-single-gpu or deepspeed-multi-node)
   • Runtime -- select a training runtime
   • Sizing -- select a sizing preset or configure custom compute resources
   • Description -- a short summary of the intended use case
3. Click Create.

Create a sizing preset

Sizing presets define compute resources independently of a runtime:

1. Navigate to the Sizing presets section.
2. Click Create Sizing Preset.
3. Configure:
   • Key -- a unique identifier (for example, small-1x-gpu)
   • Display name -- a human-readable label
   • Nodes -- number of nodes
   • CPU / Memory -- per-node resources
   • GPU -- GPU type and count per node
4. Click Create.

Use a preset when submitting a job

1. Navigate to Training Jobs and click Submit Training Job.
2. In the configuration form, click Use Preset and select a named preset.
3. The runtime, sizing, and defaults auto-fill from the preset.
4. Override any fields as needed, then submit.

note

Presets are managed via ConfigMap by workspace administrators. Users can select presets but cannot create or modify them unless they have admin permissions.

Related

• Configure a training runtime
• Submit a training job
• Presets reference