gravitino.api.model.model_version.ModelVersion

class gravitino.api.model.model_version.ModelVersion

Bases: Auditable

An interface representing a single model checkpoint under a model Model. A model version is a snapshot at a point of time of a model artifact in ML. Users can link a model version to a registered model.

__init__()

Methods

__init__()
aliases()	The aliases of this model version.
audit_info()
comment()	The comment of this model version.
properties()	The properties of the model version.
uri()	The URI of the model artifact.
version()	The version of this model object.

abstract aliases() → List[str]

The aliases of this model version. The aliases are the alternative names of the model version. The aliases are optional. The aliases are unique for a model version. If the alias is already set to one model version, it cannot be set to another model version.

Returns:

The aliases of the model version.

abstract comment() → str | None

The comment of this model version. This comment can be different from the comment of the model to provide more detailed information about this version.

Returns:

The comment of the model version. None is returned if no comment is set.

properties() → Dict[str, str]

The properties of the model version. The properties are key-value pairs that can be used to store additional information of the model version. The properties are optional.

Returns:

The properties of the model version. An empty dictionary is returned if no properties are set.

abstract uri() → str

The URI of the model artifact. The URI is the location of the model artifact. The URI can be a file path or a remote URI.

Returns:

The URI of the model artifact.

abstract version() → int

The version of this model object. The version number is an integer number starts from 0. Each time the model checkpoint / snapshot is linked to the registered, the version number will be increased by 1.

Returns:

The version of the model object.