Jupyter Notebook Utilities

Deploy only

cognite.model_hosting.notebook.notebook.deploy_model_version(name: str, model_id: int, runtime_version: str, artifacts_directory: Optional[str] = None, description: Optional[str] = None, metadata: Optional[Dict[str, str]] = None, notebook_path: Optional[str] = None, cognite_client: Optional[cognite.client.experimental.CogniteClient] = None) → int

Deploy the model version in the current notebook to the model hosting environment.

Parameters:
  • name (str) – The name of the model version.
  • model_id (int) – Id of the model to deploy the version to.
  • runtime_version (str) – The model hosting runtime version to deploy the model to.
  • artifacts_directory (str, optional) – Path of the directory containing any artifacts you want to include with your deployment.
  • description (str, optional) – Description of this model version.
  • metadata (Dict[str,str], optional) – Any metadata to include about this model verison.
  • notebook_path (str, optional) – The path to the notebook. If omitted, the notebook you’re in will be used.
  • cognite_client (CogniteClient, optional) – The CogniteClient instance to use for uploading the model. If omitted, a new instance will be created using the API key in the COGNITE_API_KEY environment variable.
Returns:

The model version id

Return type:

int

Local Artifacts

cognite.model_hosting.notebook.notebook.local_artifacts(model_version_name: str, root_dir: Optional[str] = None) → Callable

Local artifacts storage.

Returns a function which works like the builtin open(), but pointing to a directory specific to the provided model version name. The directory will be artifacts/<model_version_name>. By default the artifacts directory will reside in the current working directory.

Parameters:
  • model_version_name (str) – The name of the model version which the artifacts belong to.
  • root_dir (str, optional) – The root directory where the artifacts directory reside. Defaults to the current working directory.
Returns:

A function/context manager which works like the builtin open. Let’s your read and write to the

local artifacts directory.

Return type:

Callable

Examples

Using local_artifacts():

open_artifact = local_artifacts('my-model')
with open_artifact('my_file.txt', 'w') as f:
    f.write('A cool file\n') # will be stored in artifacts/my-model/my_file.txt

Exceptions

exception cognite.model_hosting.notebook.notebook.UnsupportedNotebookVersion

Raised if the current version of Jupyter Notebook is not supported by this integration.

exception cognite.model_hosting.notebook._model_file.InvalidCodeFormat

Raised if there is a mistake in how you have defined your model within the notebook.

exception cognite.model_hosting.notebook._setup_file.InvalidRequirements

Raised if your cell defining requirements is not correctly formatted.