Repository navigation
This branch now contains only v3 notebooks
Every example on default targets SDK v3, so the notebooks you land on match the SDK you get from a fresh install.
The SDK 2.x examples are preserved on the v2-archive branch. That branch pins sagemaker<3.0, and each notebook explains how to run it.
Notebooks are now organized by ML capability
The previous category folders have been replaced by five folders in ML lifecycle order:
| Folder | What's in it |
|---|---|
training/ |
Training with ModelTrainer — distributed training, managed spot, heterogeneous clusters, bring-your-own-container, @remote |
model_customization/ |
Fine-tuning foundation models — SFT, DPO, RLVR, RLAIF, MTRL, CPT, recipe overrides, data mixing |
evaluation/ |
Benchmark evaluation, custom scorers, LLM-as-a-judge, Inspect AI |
inference/ |
Deployment with ModelBuilder — real-time, serverless and async endpoints, local modes, A/B testing, autoscaling, model optimization, Bedrock |
mlops/ |
Pipelines and Model Registry, lineage, experiments, Feature Store, processing jobs, Clarify, MLflow |
Each folder has a README.md listing its notebooks, and every notebook is self-contained — it ships with the scripts, data and images it needs.