This is a quick guide to starting Part 1 of the fast.ai course Practical Deep Learning for Coders using Gradient Notebooks. With Gradient, you get quick access to a Jupyter instance (Lab or Notebook) without having to set up a virtual machine. Paperspace is a GPU-accelerated cloud platform
Summary of Charges
There are no storage fees associated with using Gradient Notebooks. Users pay hourly usage fees for the time that their Notebooks are Running. The hourly rate is dependent on the Compute Type selected, see all available types here. Notebooks must be stopped to end billing. Learn more about Gradient billing here.
Getting Set Up
Step 1: Create a Paperspace Account
Step 2: Access Gradient & Create Notebook
I. On the left-hand side of your Console under Gradient, select Notebooks.
II. Select the Paperspace + Fast.ai or Fast.ai JupyterLab BETA base container.
III. Select your Compute Type.
IV. Name your Notebook.
V. Enter your payment details (if you're new to Paperspace). Even if you have a promo or referral code, all active Paperspace accounts must have a valid credit card on file. You'll be able to enter your promo code later.
VI. Click Create Notebook
When you click Create Notebook, that will start your Notebook and your billing for utilization will begin. To stop billing, you must stop your Notebook. Notebooks will automatically shut down after 12 hours.
You'll be taken to your Notebook details page in your Console, where your Notebook will be created. It will go straight from Pending to Running, and will be ready to use.
Step 3: Stopping your Notebook
Under Action, just click stop. This will end the billing session.
Step 4: Restart your Notebook
Click Clone to pick up where you left off.
The Compute Type of your Notebook can be changed when you clone it. What this means for you is that you can prepare your work inexpensively and add a GPU when you're ready to run your model/algorithm.
Fast.ai data files (dogscats) can be found in the 'datasets' folder. Files in this directory are hosted by Paperspace and are read-only. See Public Datasets for more info.
The 'storage' folder is your Persistent Storage. Files placed here are available across runs of Paperspace Jobs, Notebooks, and Linux machines. Empty by default, this repository is meant to store training datasets.
The rest of the files in the notebooks directories will be available as Artifacts when your notebook is stopped.
Paperspace provides $10 of Gradient credit to start you off on your course. This code is to be used for fast.ai students only. In your console, click on Billing in the left-hand menu and enter the promo code at the bottom right. The promo code for this course is: FASTAIG5H7.
Note: If you opt for a Gradient 1 Subscription, promotional credit does not apply. Learn more about Gradient Subscription levels here.
Questions or issues related to course content, we recommend posting in the fast.ai forum.