GPU-accelerated notebook with VESSL AI

Under Project → Runs, create a **New run**.

Define the **Name** and **Description** for your interactive run.

Select **Cluster** as [[vessl-gcp-oregon]] **Resource** as [[gpu-l4-small]] .

Define **Container image** as [quay.io/vessl-ai/torch:2.1.0-cuda12.2-r3](http://quay.io/vessl-ai/torch:2.1.0-cuda12.2-r3).

Enable **Interactive** mode.

Set the **Max runtime** and **Jupyter idel timeout** as [[24h]] and [[120m]].

Click **Create** to launch the workload.