ads.model.deployment.common package
Submodules
ads.model.deployment.common.progress_bar module
- class ads.model.deployment.common.progress_bar.DummyProgressBar(*args, **kwargs)
Bases:
ProgressBar
DummyProgressBar is represents a progress bar for non-notebook environments. It inherits from the abstract class ProgressBar. It allows use of the same contextlib enter and exit methods
- update(*args, **kwargs)
- class ads.model.deployment.common.progress_bar.ProgressBar
Bases:
object
ProgressBar is an abstract class for creating progress bars.
- __enter__()
runtime context entry method
- update(description)
abstract method for update
- __exit__(exc_type, exc_val, exc_tb)
runtime context exit method
- abstract update(description)
- class ads.model.deployment.common.progress_bar.TqdmProgressBar(max_progress=100, description='Running', verbose=False)
Bases:
ProgressBar
TqdmProgressBar represents a progress bar for notebooks. It inherits from the abstract class ProgressBar
- max_progress(int)
The maximum value for the progress bar
- decription(str)
The progres bar’s description
- progress_bar(tqdm_notebook)
Notebook widget
- verbose(bool)
verbosity flag
- __enter__()
runtime context entry method
- __init__(max_progress, description, verbose)
init method for class
- update(description=None)
updates progress bar
- __exit__(exc_type, exc_val, exc_tb)
runtime context exit method
Class initialization function
- Parameters:
max_progress (int, optional) – the maximum progress bar value (defaults to 100)
description (str, optional) – the progress bar description (defaults to “Running”)
verbose (bool, optional) – verbosity flag (defaults to False)
- update(description=None)
update updates the progress bar
- Parameters:
description (str, optional) – progress bar description
- Returns:
Nothing
ads.model.deployment.common.utils module
Utilities used by the model deployment package
- class ads.model.deployment.common.utils.OCIClientManager(config=None)
Bases:
object
OCIClientManager is a helper class used for accessing DataScienceClient and DataScienceCompositeClient objects
- ds_client - class attribute for data science client
- ds_composite_client - class attribute for data science composite client
- default_compartment_id()
Determines the default compartment OCID This method finds the compartment OCID from (in priority order): an environment variable, an API key config or a resource principal signer.
- Parameters:
config (dict, optional) –
The model deployment config, which contains the following keys: auth: Authentication method, must be either “resource_principal” or “api_key”. If auth is not specified:
api_key will be used if available.
If api_key is not available, resource_principal will be used.
oci_config_file: OCI API key config file location. Defaults to “~/.oci/config” oci_config_profile: OCI API key config profile name. Defaults to “DEFAULT”
- Returns:
The compartment OCID if found. Otherwise None.
- Return type:
str or None
- prepare_artifact(model_uri: str, properties: Dict) str
Prepare model artifact. Returns model ocid.
- Parameters:
model_uri (str) – uri to model files, can be local or in cloud storage
properties (dict) – dictionary of properties that are needed for creating a model.
ds_client (DataScienceClient) – OCI DataScienceClient
- Returns:
model ocid
- Return type:
str
- class ads.model.deployment.common.utils.State(value)
Bases:
Enum
An enumeration.
- ACTIVE = 1
- CREATING = 2
- DELETED = 3
- DELETING = 4
- FAILED = 5
- INACTIVE = 6
- UNKNOWN = 8
- UPDATING = 7
- ads.model.deployment.common.utils.get_logger()
- ads.model.deployment.common.utils.get_progress_bar(max_progress, description='Initializing')
get_progress_bar return an instance of ProgressBar, sensitive to the runtime environment
- Parameters:
max_progress (int) – Progress bar max
description (str, optional) – Progress bar description (defaults to “Initializing”)
- Returns:
An instance of ProgressBar. Either a DummyProgressBar (non-notebook) or TqdmProgressBar (notebook environement)
- ads.model.deployment.common.utils.is_notebook()
is_notebook returns True if the environment is a Jupyter notebook and False otherwise
- Parameters:
None –
- Returns:
True if Jupyter notebook; False otherwise
- Return type:
bool
- Raises:
NameError – If retrieving the shell name from get_ipython() throws an error
- ads.model.deployment.common.utils.seconds_since(t)
seconds_since returns the seconds since t. t is assumed to be a time in epoch seconds since time.time() returns the current time in epoch seconds.
- Parameters:
t (int) –
- Returns
int: the number of seconds since t
- ads.model.deployment.common.utils.send_request(data, endpoint: str, dry_run: bool = False, is_json_payload: bool = False, header: dict = {})
Sends the data to the predict endpoint.
- Parameters:
data (bytes or Json serializable) – data need to be sent to the endpoint.
endpoint (str) – The model HTTP endpoint.
dry_run (bool, optional) – Defaults to False.
is_json_payload (bool, optional) – Indicate whether to send data with a application/json MIME TYPE. Defaults to False.
header (dict, optional) – A dictionary of HTTP headers to send to the specified url. Defaults to {}.
- Returns:
A JSON representive of a requests.Response object.
- ads.model.deployment.common.utils.set_log_level(level='INFO')
set_log_level sets the logger level
- Parameters:
level (str, optional) – The logger level. Defaults to “INFO”
- Returns:
Nothing