The 2020.09 release contains klever-model-registry and ormb.
klever-model-registry contains one long-running backend server, one Kubernetes CRD operator, one dashboard.
klever-model-registry features:
- Manage ML models
- Keep track of ML models' hyperparameters and so on to help decision makers
- Convert models between different formats
- Extract signatures for the given models
- Serve the simple models ( single model, without inference graph support ) using Triton Inference Server
- Support uploading models from the dashboard
- Get model conversion/extraction job logs and events
ormb contains two executive programs.
ormb features:
- Save, export models from/to local filesystem cache
- Push, pull models from/to remote registry
The 2020.11 release contains klever-model-registry and ormb.
klever-model-registry contains one long-running backend server, one Kubernetes CRD operator, one dashboard.
klever-model-registry new features:
- Support multi-users
- Provide offline unified batch inference interface
- Support traffic management for model inference
ormb contains two executive programs.
ormb new features:
- Support TLS verification
- Validate model directory layout for the given format
The 2020.01 release contains klever-model-registry and ormb.
klever-model-registry contains one long-running backend server, one Kubernetes CRD operator, one dashboard.
klever-model-registry new features:
- Support model compression
- Support edge deployment
ormb contains two executive programs.
ormb new features:
- Support README addition layer
We may start working on training in this release.