KubeDL enables deep learning workloads to run on Kubernetes more easily and efficiently.
Its core functionalities include:
- Support running training and inference workloads on Kubernetes (Tensorflow, Pytorch. Mars etc.)in a single unified controller.
- Automatically tunes the best container-level configurations before an ML model is deployed as inference services. - Morphling
- Model lineage and versioning to track the history of a model natively in CRD: when the model is trained using which data and which image, each version of the model, which version is running etc.
- Enables storing and versioning a model leveraging container images. Each model version is stored as its own image and can later be served with Serving framework.
KubeDL is a CNCF sandbox project.