Getting Started / Hands-on / ...
Learn and Experiment with Machine Learning Using a No-Setup, Free Development Environment
GETTING STARTED GUIDE
Overview
In this tutorial, you learn and experiment with machine learning using Amazon SageMaker Studio Lab, a no-setup, free development environment.
Amazon SageMaker Studio Lab is a free service that enables data scientists to quickly experiment with machine learning (ML) without the need for an AWS account, credit card or cloud configuration skills. With just an email address and a mobile phone number, users can register for an account, that gives the access to timed user sessions with CPU or GPU compute power to experiment with Jupyter notebooks. When the users’ sessions end, their work is automatically saved in dedicated storage, enabling them to pick up where they left off. Users that want to move to production and need capabilities such as continuous integration and continuous delivery (CI/CD), and real-time predictions, can easily migrate their experiments to Amazon SageMaker Studio.
What you will accomplish
In this tutorial, you will:
- Create an Amazon SageMaker Studio Lab account
- Launch your own JupyterLab environment
- Clone the Machine Learning University course repository
- Run the notebook
- Stop and restart the runtime
Prerequisites
Before starting this guide, you will need:
- An email address and valid mobile phone number.
AWS experience
Beginner
Time to complete
10 minutes
Cost to complete
Free
Requires
A valid email address and phone number.
Services used
Last updated
May 23, 2023
Implementation
In this tutorial, you will create the Studio Lab account and clone the Machine Learning University: Accelerated Natural Language Processing Class materials in your Studio Lab environment. Finally, you run the notebooks in it.
Step 1: Create Amazon SageMaker Studio Lab account
Go to the SageMaker Studio Lab landing page to request an account.
If you have a referral code then you can get immediate access to Studio Lab, otherwise you will have to wait 1-3 days for approval. Note: multiple requests will be discarded.
After you request the account, you will receive 4 emails.
- Account request confirmed: This email notifies, we have received your account request correctly.
- Account request approved: This email notifies that your account request is approved. The approval is valid for 7 days, please create the account from the link in the email within this period.
- Verify your email: Please authorize your mail address from the link in the mail.
- Your account is ready: This email notifies that we prepared your Studio Lab account.
If you do not receive our emails, please confirm the email filter setting and check your spam folder. It will take maximum of 5 business days from the account request confirmation to account approval.
Step 2: Launch your own JupyterLab environment
You will be presented with the CAPTCHA option to proceed.
Note to staging team: Please see correct image for CAPTCHA (solve the puzzle)
After signing in, you will see the project page. By clicking the Start runtime button, you can launch your own JupyterLab environment. Studio Lab provides free CPU and GPU computing instance.
You can confirm the remaining hours on this page. Please review the latest computing resource limits by visiting the FAQ page.
After your runtime started, please click the Open Project button.
Step 3: Clone the Machine Learning University course repository
Let’s start the journey by using the materials from the Machine Learning University: Accelerated Natural Language Processing Class. This free machine learning courses are published on GitHub.
By scrolling down the page, you will find the Open in Studio Lab button. This button enables you to quickly clone and prepare the environment.
Select the Clone Entire Repo.
The project directory to clone into should be blank to copy the contents in the root directory.
Note:
- By selecting “Open README files,” Studio Lab will automatically open the readme file of the repository when it is done cloning.
- By selecting “Search for environment.yml and build Conda environment,” Studio Lab will automatically build a Conda environment from the environment.yml file. If this option is selected, but the file cannot be found, you will get a message.
Next, select “Clone.”
The terminal is launched automatically and the contents of the repository are copied by a Git command. Since this repository does not have the environment.yml file, the following prompt appears. You can click Dismiss on the dialog box.
Right click mlu-nlp.yml and select Build Conda Environment. In this repository, the file name that defines required libraries is not environment.yml but mlu-nlp.yml. In that case, we follow the process. If the file name is environment.yml, this process is automatically executed.
Step 4: Run the notebook
Let’s run the notebook by using prepared environment. Select the notebooks folder in the left file browser pane and open the MLA-NLP-Lecture1-Text-Process.ipynb file that is the first notebook for Machine Learning University.
After opening the notebook, select the kernel that you prepared, mlu-nlp.
Step 5: Stop the runtime
After you finish your learning, Select the Stop runtime to save the free hours. All your work is stored in the static storage, and you can continue your work by clicking Start runtime.
Note: Studio Lab is a free service and you will not get charged for the compute sessions. However, it is good to get into the habit of conserving cloud compute when you can, so please stop sessions if you no longer need them.
Conclusion
Congratulations! You have finished the Learn and Experiment with Machine Learning Using a No-Setup, Free Development Environment tutorial.
You have successfully set up the environment and run the notebooks using Amazon SageMaker Studio Lab.. You can continue your journey to become a machine learning practitioner by reviewing the following resources: