Branch | Status |
---|---|
master | |
release-0.1 |
Seldon Core is an open source platform for deploying machine learning models on Kubernetes.
Machine learning deployment has many challenges. Seldon Core intends to help with these challenges. Its high level goals are:
- Allow data scientists to create models using any machine learning toolkit or programming language. We plan to initially cover the tools/languages below:
- Python based models including
- Tensorflow models
- Sklearn models
- Spark models
- H2O models
- R models
- Python based models including
- Expose machine learning models via REST and gRPC automatically when deployed for easy integration into business apps that need predictions.
- Allow complex runtime inference graphs to be deployed as microservices. These graphs can be composed of:
- Models - runtime inference executable for machine learning models
- Routers - route API requests to sub-graphs. Examples: AB Tests, Multi-Armed Bandits.
- Combiners - combine the responses from sub-graphs. Examples: ensembles of models
- Transformers - transform request or responses. Example: transform feature requests.
- Handle full lifecycle management of the deployed model:
- Updating the runtime graph with no downtime
- Scaling
- Monitoring
- Security
A Kubernetes Cluster.
Kubernetes can be deployed into many environments, both in cloud and on-premise.
- Quick Start tutorial using Minikube
- Jupyter notebooks showing worked examples:
- Minikube:
- GCP:
- Advanced graphs showing the various types of runtime prediction graphs that can be built.
Seldon-core allows various types of components to be built and plugged into the runtime prediction graph. These include routers, transformers and combiners. Some components that are available as part of the project are:
- Multi-Armed bandits
- Epsilon-greedy multi-armed bandits for real time optimization of models
- Contextual bandits. (coming soon).
- Outlier Detection
- Mahalanobis distance outlier detection. Example usage can be found in the Advanced graphs notebook
- Model explainers (coming soon).
- seldon-core can be installed as part of the kubeflow project. A detailed end-to-end example provides a complete workflow for training various models and deploying them using seldon-core.
Official releases can be installed via helm from the repository https://storage.googleapis.com/seldon-charts.
To install seldon-core:
helm install seldon-core-crd --name seldon-core-crd --repo https://storage.googleapis.com/seldon-charts
helm install seldon-core --name seldon-core --repo https://storage.googleapis.com/seldon-charts
To install the optional analytics components including Prometheus and Grafana with a built-in dashboard for monitoring the running ML deployments run:
helm install seldon-core --name seldon-core \
--set grafana_prom_admin_password=password \
--set persistence.enabled=false \
--repo https://storage.googleapis.com/seldon-charts
Three steps: