Tune your ML models to the highest accuracy with automatic model tuning
summary
Amazon SageMaker automatically tunes hyperparameters using Bayesian optimization by launching many child training jobs from a provided estimator, evaluating an objective metric parsed from training logs via a regex, and iterating (example: run up to max_jobs with N jobs in parallel — often 3). It supports XGBoost, PyTorch, TensorFlow, warm-starting, transfer learning, early stopping, checkpointing to S3 for Spot interruptions, retrain triggers via Lambda/S3 events, and can save up to 90% with Spot.
AI-written summary · about
chapters (7)
- 0:00Introduction to tuning
- 1:07Evaluating ML models
- 9:02Training models on SageMaker
- 19:30Spot instances savings
- 21:58Automatic model tuning steps
- 30:22Retrain and tune pipelines
- 32:31Closing and resources
related talks
31mTrain and tune ML models to the highest accuracy using Amazon SageMaker
39mTrain ML models at scale with Amazon SageMaker Training
27mScaling ML to the next level: Hosting thousands of models on SageMaker
32mBuild quality ML models easily & quickly with Amazon SageMaker Autopilot
31mUse AutoML to create high-quality models with just a few clicks
31mUse AutoML to create high-quality models with just a few clicks
43mDesigning and Building Complex Machine Learning Projects
19mDeveloping large-scale machine learning algorithms on Amazon SageMaker