Run Jupyter Notebook

​

Intro

You can quickly set up a Jupyter Notebook instance running on a Shadeform GPU using Shadeform’s docker launch configuration. All Shadeform instances come pre-installed with Python, Cuda, and Docker so running containers is easy.

​

Example

For this example, we will run the container image quay.io/jupyter/pytorch-notebook:cuda12-python-3.11.8. See the Quickstart tutorial to see how to get an API key and select a different GPU. We will be running the instance on an A6000 GPU at $0.57/hr. Make sure to replace <api-key> with your account’s API Key.

After making the curl request, you should see a response like this:

{"id":"26bedfa2-1e98-4ff5-9342-a5b0987a2f0f","cloud_assigned_id":"b9ecaf4e-d6ac-42eb-a89a-60010065388a"}

After you’ve launched the instance, you can track the instance progress by visiting the Running Instances page. When you’re on the page, select the instance named jupyter-notebook and click on the Logs tab.

You can follow the progress of the container download and spin up in the log feed.

When the container is full spun up and your Jupyter Notebook instance is running, you will see an entry that looks like this:

http://shadecloud:8888/lab?token=736db552017247000023364f8e215cebebaa2f4431b4b015

Make sure to save the token from that url. In this example, the token is 736db552017247000023364f8e215cebebaa2f4431b4b015 but your token will be different.

After the container is spun up, you can take the IP of the instance and go to the following url:

http://<ip_address>:8888

On this page, you will need to paste in the token you saved earlier and click the Login button. After logging in with your token, your Jupyter Notebook instance is up and running! You can use the following code to check for the GPU on your machine.

import torch

# Check if CUDA is available
if torch.cuda.is_available():
  print("CUDA is available. GPU Details:")
  # Number of GPUs available
  print(f"Number of GPUs: {torch.cuda.device_count()}")
  for i in range(torch.cuda.device_count()):
  # Get the name of the current GPU
  print(f"GPU {i}: {torch.cuda.get_device_name(i)}")
else:
  print("CUDA is not available. Running on CPU.")