Create per-device Lab

Create a Jupyter environment with a GPU or NPU device attached. Use a GPU Lab for downloading and experimenting with models, and an NPU Lab for compilation and NPU inference work.

---

GPU Lab

Use this in a GPU environment when downloading or experimenting with models.

1. Go to the Lab menu

In the left sidebar, click Development > Lab and press the Create button.

2. Enter the Lab settings

Field	Example value
Name	tutorial-lab
Server Type	Jupyter
Accelerator Type	Nvidia GPU
Count	1
Data Volumes	tutorial-volume → mount path /data

3. Connect to the Lab

When the status in the Lab list becomes Running, click the Connect button to enter the Jupyter environment.

---

NPU Lab

Create a Jupyter environment equipped with an RNGD NPU to perform compilation and NPU work.

1. Go to the Lab menu

In the left sidebar, click Development > Lab and press the Create button.

2. Enter the Lab settings

Field	Example value
Name	tutorial-lab-npu
Server Type	Jupyter
Image	Jupyter Notebook with FuriosaAI RNGD
Accelerator Type	RNGD
Count	1
Data Volumes	tutorial-volume → mount path /data

Image selection

Any Jupyter image works to bring the Lab up. However, to use the furiosa-smi command, you must select the Jupyter Notebook with FuriosaAI RNGD image. Other images will produce furiosa-smi: command not found.

3. Connect to the Lab

When the status in the Lab list becomes Running, click the Connect button to enter the Jupyter environment.

---

Check NPU Recognition

Clicking Connect opens the Jupyter environment. Open a terminal from the top menu File > New > Terminal and run the command below.

Run in the NPU Lab terminal

• Open a terminal via the Jupyter top menu File > New > Terminal and run there.
• If you enter it into a notebook cell (.ipynb) Python interprets it and you'll get a SyntaxError.
• furiosa-smi is available only in the NPU Lab. Running it in a GPU Lab produces command not found.

furiosa-smi info

If the NPU is healthy, you should see output similar to:

+------+--------+---------+-------+--------+---------+
| Arch | Dev    | Temp(C) | Power | PCI-BDF | UUID    |
+------+--------+---------+-------+--------+---------+
| RNGD | npu0pe0|   35.0  | 70 W  | ...    | ...     |
+------+--------+---------+-------+--------+---------+

If the NPU is not recognized

Ask your administrator to check the NPU device status.

---

Next Step

→ 03. Download the model — Save a model from Hugging Face to a Volume