Train Models

For an overview on training models with Amazon SageMaker, see Train a Model with Amazon SageMaker.

SageMaker provides features to monitor and manage the training and validation of machine learning models. For guidance on metrics available, incremental training, automatic model tuning, and the use of augmented manifest files to label training data, see the following topics.

• For guidance on choosing a machine learning algorithm and its implementation for your task or problem, see Choose an Algorithm.

• For guidance on debugging the training of machine learning models, see Amazon SageMaker Debugger.

• For guidance on metrics used to monitor and train models, see Monitor and Analyze Training Jobs Using Metrics.

• For guidance on metrics used to detect model post-processing bias, see Detect Posttraining Data and Model Bias.

• For guidance on model explainability, see Model Explainability.

• For guidance on incremental training in SageMaker, see Incremental Training in Amazon SageMaker.

• For guidance on using managed spot training in SageMaker, see Managed Spot Training in Amazon SageMaker.

• For guidance on using training checkpoints in SageMaker, see Use Checkpoints in Amazon SageMaker.

• For guidance on automatic model tuning, also known as hyperparameter tuning, see Perform Automatic Model Tuning.

• For guidance on using an augmented manifest file to label training data, see Provide Dataset Metadata to Training Jobs with an Augmented Manifest File.

Topics

• Choose an Algorithm
• Manage Machine Learning with Amazon SageMaker Experiments
• Amazon SageMaker Debugger
• Perform Automatic Model Tuning
• Distributed Training
• Detect Posttraining Data and Model Bias
• Model Explainability
• Incremental Training in Amazon SageMaker
• Managed Spot Training in Amazon SageMaker
• Use Checkpoints in Amazon SageMaker
• Provide Dataset Metadata to Training Jobs with an Augmented Manifest File
• Monitor and Analyze Training Jobs Using Metrics