Using Ray to perform hyperparameter optimization for an ML model all within a Jupyter Notebook. This repo utilises the nersc_cluster_deploy python library to create the Ray clusters easily via the SF API all within a Jupyter Notebook.
These example notebooks will cover how different machine learning frameworks and codes.
| Notebook | Description | |
|---|---|---|
| 1 | Tuning Hyperparameters of a Distributed PyTorch Model with PBT using Ray Train & Tune | Deploying a Ray cluster via Login Node in order to do Distributed Tunning of Hyperparameters with PyTorch. |
| 2 | Tuning Hyperparameters of a Distributed TensorFlow Model using Ray Train & Tune | Deploying a Ray cluster via Jupyter Compute in order to do Distributed Tunning of Hyperparameters with TensorFlow. |
Note To setup the environment for each notebook, execute on command line:
./setup.sh <exercise-number>(e.g./setup.sh 1).