SageMaker Studio Tutorial
What is SageMaker Studio?
From data preparation to the creation, training, deployment, and management of your machine learning (ML) models, Amazon SageMaker Studio provides a broad range of tools designed specifically to handle all ML development tasks. You can use your favourite integrated development environment (IDE) to design models and submit data easily. Using a single, cohesive web-based interface, streamline ML team collaboration, code effectively with the AI-powered coding companion, tune and debug models, deploy and manage models in production, and automate workflows.
Benefits of SageMaker Studio
Choice of fully managed IDEs
For ML development, Amazon SageMaker Studio provides a wide range of fully managed IDEs, such as RStudio, Code Editor based on Code – OSS (Visual Studio Code Open Source), and JupyterLab. Launch your favourite IDE quickly, then adjust the underlying computational resources as needed.
Purpose-built tools for ML development with generative AI support
Get access to the most complete collection of tools for every stage of machine learning development, including data preparation, model construction, training, deployment, and management. Easily switch between phases to optimise your models, rerun training trials, and scale to distributed training straight from SageMaker AI’s JupyterLab, Code Editor, or RStudio. With Amazon Q Developer, a generative AI-powered assistant that is integrated into your workflow, you can build more quickly. Within the confines of your notebook, Amazon Q Developer provides real-time support throughout your whole machine learning development journey.
Browse, tune, evaluate, and deploy FMs
With Amazon SageMaker JumpStart, you can create generative AI applications using more than 15 prebuilt solutions and hundreds of well-known publically available models. Top model suppliers including AI21 Labs, LightOn, Stability AI, Hugging Face, Alexa, and Meta AI are available for you to use. The best foundation models (FMs) for your use case may then be swiftly assessed, contrasted, and chosen using predetermined parameters like toxicity, accuracy, and robustness. More subjective aspects like creativity and style can be assessed by humans.
Securely run anywhere
SageMaker Studio is accessible through a web browser on any device. There is no need to download critical machine learning artefacts to your local computer because both code and data are stored in your secure cloud environment.
Specialized generative AI and ML development apps from leading partners
Use well-liked apps from AWS partners to speed up the creation of your models. Within the safety and privacy of your SageMaker environment, take advantage of a smooth, completely managed experience that requires no infrastructure to provision or run.
Use cases
Accelerate generative AI development
With a variety of publicly available FMs, model evaluation tools, IDEs supported by high-performance accelerated computation, and the ability to immediately refine and scale FMs from SageMaker Studio, you can create generative AI applications more quickly.
Scale and standardize ML workflows
Use the most complete ML tools in one location to streamline your end-to-end ML development in SageMaker Studio. SageMaker AI provides governance solutions to enhance transparency and auditability throughout your company, as well as high-performing MLOps tools to assist you in automating and standardising ML workflows.
Combine big data analytics and ML
SageMaker Studio provides a single experience for all ML and data analytics tasks. Construct, explore, and establish a connection to Amazon EMR clusters. Use AWS Glue interactive sessions to create, test, and execute interactive analytics and data preparation apps. Using well-known tools like Spark UI, you can monitor and debug Spark operations directly from SageMaker Studio.
Features of Amazon SageMaker Studio
JupyterLab
In only a few seconds, launch JupyterLab completely managed. Make use of the newest interactive web-based development environment for data, code, and notes. Its elastic and adaptable interface makes it simple to set up machine learning (ML) processes. In your notebook environment, get AI-powered help with code generation, troubleshooting, and professional advice to speed up your machine learning progress.
Code Editor, based on Code-OSS
Utilise the code editor’s well-known shortcuts, terminal, debugger, and refactoring tools to increase productivity. It is both lightweight and powerful. To improve your programming experience, select from thousands of Visual Studio Code-compatible extensions in the Open VSX extension gallery. Use GitHub repositories to facilitate cross-team collaboration and version control. With the preconfigured SageMaker AI distribution, you can use the most widely used machine learning frameworks right out of the box. Increase coding efficiency with chat-based and inline code suggestions powered by Amazon Q Developer, and seamlessly integrate with AWS services with the AWS Toolkit for Visual Studio Code, which includes built-in access to AWS data sources like Amazon Simple Storage Service (Amazon S3) and Amazon Redshift.
RStudio
Plotting, history, debugging, and workspace management tools are all included in the fully controlled integrated development environment (IDE) for R, which also includes a console and a syntax-highlighting editor that allows direct code execution. Create insights with preconfigured R packages like tidyverse, rmarkdown, devtools, and shine, then publish them with RStudio Connect. For R and Python work, you can easily switch between the IDEs of RStudio, JupyterLab, and Code Editor.
Access and evaluate FMs
Start developing generative AI quickly with Amazon SageMaker JumpStart’s hundreds of publicly available FMs and prebuilt solutions that can be deployed in a matter of steps. Use Amazon SageMaker Clarify to rapidly assess, compare, and choose the best FMs for your use case based on a range of characteristics, including accuracy, robustness, toxicity, and bias, in just a few minutes. Use carefully chosen prompt datasets to begin FM assessments, or expand the evaluation using your own unique prompt datasets. More subjective aspects like creativity and style can be assessed by humans.
Prepare data at scale
Utilise a single platform for data engineering, analytics, and machine learning to streamline your data workflows. Use AWS Glue serverless Spark environments and Amazon EMR to run Spark jobs interactively, and use Spark UI to keep an eye on them. Utilise the integrated data preparation feature to visualise data, spot problems with data quality, and implement suggested fixes to enhance data quality. In just a few simple steps, schedule your notebook as a job to automate your data preparation activities. ML model features can be managed, shared, and stored in a central feature store.
