Skip to content

Latest commit

 

History

History
6 lines (4 loc) · 742 Bytes

File metadata and controls

6 lines (4 loc) · 742 Bytes

Distributed Optimization on Kubernetes

This folder contains two kinds of examples with Kubernetes: one is based on sklearn_simple.py and the other is based on pytorch_lightning_simple.py with MLflow.

Currently, both simple/sklearn_distributed.py and mlflow/pytorch_lightning_distributed.py use POSTGRESQL for their backend of optuna.Study.optimize to be parallelized.
Though we do not use it for MLflow records. Of course, you can use POSTGRESQL as backend store of MLflow (https://mlflow.org/docs/latest/tracking.html#where-runs-are-recorded), current example uses HTTP server.