Go SDK to manage artifacts stored by Kubeflow via MLMD.
go get github.com/Vernacular-ai/artifact-registry
Check documentation for sample code.
https://pkg.go.dev/github.com/Vernacular-ai/artifact-registry/registry
Follow the doc here: https://github.com/google/ml-metadata/ to setup local MLMD gRPC server.
To access the MLMD gRPC server bundled with Kubeflow:
kubectl port-forward -n kubeflow $(kubectl get pods -nkubeflow | grep metadata-grpc-deployment | head -n 1 | cut -d' ' -f1) 8080:8080
.
├── LICENSE
├── README.md
├── go.mod
├── go.sum
├── protos
│ ├── artifact-registry.pb.go
│ ├── artifact-registry.proto
│ ├── metadata_source.pb.go
│ ├── metadata_source.proto
│ ├── metadata_store.pb.go
│ ├── metadata_store.proto
│ ├── metadata_store_service.pb.go
│ └── metadata_store_service.proto
└── registry
├── artifact_registry.go
├── registry_test.go
└── utils.go
protos/
has protobufs and generated code for MLMD data store, MLMD gRPC service and the artifact registry SDK's data definition.registry/artifact_registry.go
has all the code to manage artifacts.