Configure a training runtime

A training runtime is a pre-built environment that combines a deep learning framework, a container image, and a parallelism strategy. Runtimes are prerequisites for submitting training jobs.

Prerequisites

• A Kubernetes cluster in ready status
• Admin permissions on the workspace

Available runtimes

Vantage ships with several built-in runtimes:

Runtime	Framework	Use case
torch-distributed	PyTorch + torchrun	General multi-GPU training
deepspeed-zero3	DeepSpeed	Memory-bound LLMs with ZeRO-3 sharding
mlx-distributed	MLX	Apple-silicon training
huggingface-trainer	Transformers	Standard HF Trainer workflows

View runtimes

1. Click Workbench in the left sidebar.
2. Navigate to Training Jobs > Runtimes (or find runtimes in the Training section).
3. The list shows each runtime with its framework, image, and status.

Create a custom runtime

1. On the Runtimes page, click Create Runtime.
2. Configure the runtime:
   • Name -- a short identifier (for example, jax-tpu or custom-pytorch)
   • Framework -- the ML framework (PyTorch, DeepSpeed, JAX, or custom)
   • Container image -- the Docker image URI (for example, pytorch/pytorch:2.1.0-cuda12.1-cudnn8-runtime)
   • Default entrypoint -- the command or script that runs when a job starts (for example, torchrun)
3. Click Create.

Edit a runtime

1. Click the runtime name to open the detail page.
2. Click Edit and modify the image, entrypoint, or other settings.
3. Click Save.

Set a default runtime

Click the Set as default action on a runtime row. The default runtime is pre-selected when creating new training jobs.

note

Runtimes are workspace-scoped. Admins can publish runtimes that are available to all users in the workspace.

Related

• Create a training preset
• Submit a training job
• Runtimes reference