# User guide

Build a sleepKIT workflow from datasets, feature extraction and model architectures. Use the guides below to configure each step, then follow a tutorial for a complete experiment.

## Run an experiment

Start with a task configuration and work through the modes you need.

Configure a run
Download data
Train
Evaluate
Export
Run a demo

## Choose your components

- **Datasets** provide recordings and labels for your task. Review each dataset's access requirements and terms.
- **Feature sets** transform recordings into inputs for training and evaluation.
- **Model architectures** define the network and its configurable parameters.

Browse datasets
Explore feature sets
Choose an architecture

## Tutorials and experiments

Train a sleep-detection model from wrist IMU data in the saved notebook, or explore how feature and model choices affect sleep staging.

Detection training notebook
Staging ablation study