SageMaker Flashcards
SageMaker
Fully managed service for developers and data scientists to build ML models
Collect and prepare data, Build and train ML models, then Deploy them and monitor prediction performance
Deploy with one click, automatic scaling, no servers to manage
Real-Time Inference
SageMaker model deployment method for real-time endpoints that make one prediction at a time
Serverless Inference
SageMaker model deployment method for workloads that have idle periods between traffic spikes and can tolerate cold starts
Asynchronous Inference
SageMaker model deployment method for requests with large payload sizes up to 1GB, long processing times, and near real-time latency requirements
Batch Transform
SageMaker model deployment method for getting predictions for an entire dataset
Automatic Model Tuning
SageMaker feature that automatically chooses hyperparameter ranges, search
strategy, maximum runtime of a tuning job, and early stop condition
Define the Objective Metric to determine how this feature actually tunes your model
Saves you time and money; Helps you avoid wasting money on suboptimal configurations
SageMaker Studio
SageMaker feature that offers end-to-end ML development from a unified interface
Team collaboration; deploy, tune, and debug ML models; automated workflows
SageMaker Data Wrangler
SageMaker feature for preparing tabular and image data for ML that is imported from S3
Single interface for data selection,
cleansing, exploration, visualization,
and processing
Data preparation, transformation and
feature engineering; SQL supported
Data Quality tool for ensuring ML training data does not negatively impact inference accuracy
ML Features
Inputs to ML models used during training and used for inference
For example, converting a rigid BirthDate value into a simple Age number
Important to have high quality features across your datasets in your company for re-use
SageMaker Feature Store
SageMaker feature that ingests features from a variety of sources, which become available in SageMaker Studio
Ability to define the transformation of data into feature from within the store
Can publish directly from SageMaker Data Wrangler into this feature
SageMaker Clarify
SageMaker feature for evaluating FMs on human-like factors, such as friendliness or humor
Leverage an AWS-managed team or bring your own employees for evaluation
Use built-in datasets or bring your
own dataset; offers built-in metrics and algorithms
Integrated with SageMaker Studio
Model Explainability
SageMaker Clarify feature that offers set of tools to help explain how MLs make predictions
Understand model characteristics as a whole prior to deployment; debug predictions provided by the model after it’s deployed
Helps increase the trust and understanding of the model
Bias Detection
SageMaker Clarify feature that offers ability to detect and explain biases in your
datasets and models
Measure bias using statistical metrics; specify input features and bias will be automatically detected
SageMaker Ground Truth
SageMaker feature for Reinforcement Learning from Human Feedback
RLHF: Model review, customization and evaluation; Align model to human preferences; human feedback included in “reward” function
Human feedback for ML includes creating or evaluating your models, and data generation or annotation like data labels
Reviewers include Amazon Mechanical Turk workers, your employees, or third-party vendors
SageMaker Model Cards
SageMaker feature for ML governance that allows you to document essential model information
Examples include intended uses, risk ratings, and training details
SageMaker Model Dashboard
SageMaker feature that provides a centralized repository for information and insights for all models
Aggregates model-related information from several SageMaker features
View model cards, visualize workflow lineage, and track your endpoint performance
SageMaker Role Manager
SageMaker feature for building and managing persona-based IAM roles for common ML needs
Implement least-privilege access for models are accessed securely by your roles
SageMaker Model Monitor
SageMaker feature for monitoring the quality of your model in production, continuous or on-schedule
Alerts for deviations in the model quality: fix data & retrain model
For example, model starts giving loans to people who don’t have the correct credit score
SageMaker Model Registry
SageMaker feature that provides centralized repository for tracking, managing, and version ML models
Catalog models, manage model versions, associate metadata with a model
Manage approval status of a model, automate model deployment, share models. and more
SageMaker Pipeline
SageMaker feature that provides a workflow that automates the process of building, training, and deploying a ML model
Essentially a CI/CD service for ML; helps you easily build, train, test, and deploy 100s of models automatically
Iterate faster, reduce errors with no manual steps, repeatable mechanisms
Steps: Processing, Training, Tuning, AutoML, Model, ClarityCheck, QualityCheck
SageMaker JumpStart
SageMaker feature that provides ML Hub to find pre-trained FMs, computer vision models, or NLP models
Large collection of models from Hugging
Face, Databricks, Meta, Stability AI, and more
Models can be fully customized for your data and use-case, and are deployed on SageMaker Directly
Pre-built ML solutions for demand
forecasting, credit rate prediction, fraud
detection and computer vision
SageMaker Canvas
SageMaker feature for building ML models using a visual interface, no coding required
Access to ready-to-use models from
Bedrock or JumpStart; part of SageMaker Studio
Build your own custom model using AutoML powered by SageMaker Autopilot
Leverage Data Wrangler for data
preparation
MLFlow
Open-source tool available on SageMaker which helps ML teams manage the entire ML lifecycle
Tracking Servers used to track runs and experiments; launch on SageMaker with a few clicks
Fully integrated with SageMaker and part of SageMaker Studio
