Introduction to GenericModel

Overview:

The GenericModel module in ADS provides different ways of serializing a trained model. This example demonstrates how to utilize the GenericModel module to prepare model artifacts, and save models to the model catalog then deploy any unsupported model framework.

A model artifact includes the model, metadata about the model, input and output schema, and a script to load the model and make predictions. You can share model artifacts among data scientists, tracked for provenance, reproduced, and deployed.

Initialize

GenericModel() initiates a generic model instance. It accepts the following variables:

estimator: (Callable). Trained model.
artifact_dir: str. Artifact directory to store the files needed for deployment.
properties: (ModelProperties, optional). Defaults to None. ModelProperties object required to save and deploy model.
auth :(Dict, optional). Defaults to None. The default authentication is set using the ads.set_auth API. If you need to override the default, use ads.common.auth.api_keys or ads.common.auth.resource_principal to create appropriate authentication signer and kwargs required to instantiate the IdentityClient object.
serialize: (bool, optional). Defaults to True. Whether to serialize the model to a pkl file by default. If False, you must set the model_file_name in .prepare function, serialize the model manually, save it under artifact_dir, and update the score.py file manually.

If you would like to serialize a model your way, you can set serialize=False, and then save the serialized model into the artifact_dir. However, you are also responsible for setting the model_file_name in .prepare function, making changes to load_model and other functions in score.py. Also, in score.py, pre_inference and post_inference are currently not used.

properties is an instance of ModelProperties and has the following predefined fields:

inference_conda_env: str
inference_python_version: str
training_conda_env: str
training_python_version: str
training_resource_id: str
training_script_path: str
training_id: str
compartment_id: str
project_id: str
deployment_instance_shape: str
deployment_instance_count: int
deployment_bandwidth_mbps: int
deployment_log_group_id: str
deployment_access_log_id: str
deployment_predict_log_id: str

By default, properties is populated from environment variables if it’s not specified. For example, in the notebook session the environment variables for project id and compartment id are preset and stored in PROJECT_OCID and NB_SESSION_COMPARTMENT_OCID``by default. And ``properties populates these variables from the environment variables, and uses the values in functions such as .save(), .deploy() by default. However, if these aren’t the values you want, you can always explicitly pass the variables into functions to overwrite those values. For the fields that properties has, it records the values that you pass into the functions. For example, when you pass inference_conda_env into .prepare(), then properties records this value. Later you can export it using .to_yaml(), and reload it using .from_yaml() from any machine. This allows you to reuse the properties in different places.

Summary_status

You can call the summary_status() function any time after the GenericModel instance is created. This applies to other model frameworks such as SklearnModel and so on. It returns a Pandas dataframe that guides you though the whole workflow. It shows which function is available to call, which ones aren’t and what each function is doing. If extra actions are required, it also shows those.

An example of a summary status table looks similar to the following after you initiate the model instance. The step column shows all the functions. It shows that init step is completed where the Details column explained that what the init step did and now prepare() is available. The next step is to call prepare().

Prepare

.prepare() takes the following parameters:

inference_conda_env: (str, optional). Defaults to None. Can be either a slug or an object storage path of the conda environment. You can only pass in slugs if the conda pack is a service environment.
inference_python_version: (str, optional). Defaults to None. Python version to use in deployment.
training_conda_env: (str, optional). Defaults to None. Can be either a slug or an object storage path of the conda environment. You can only pass in slugs if the conda environment is a service environment.
training_python_version: (str, optional). Defaults to None. Python version used during training.
model_file_name: (str). Name of the serialized model.
as_onnx: (bool, optional). Defaults to False. Whether to serialize as Onnx model. This will only enforce the model_file_name to have a suffix of .onnx. The actually onnx serialization is not implemented here.
force_overwrite: (bool, optional). Defaults to False. Whether to overwrite existing files.
namespace: (str, optional). Namespace of region. This is used for identifying which region the service environment is from when you pass a slug to inference_conda_env and training_conda_env.
use_case_type: str. The use case type of the model. Use it through UserCaseType class or string provided in UseCaseType. For example, use_case_type=UseCaseType.BINARY_CLASSIFICATION or use_case_type="binary_classification". Check using UseCaseType class to see all of the supported types.
X_sample: Union[list, pd.Series, np.ndarray, pd.DataFrame]. Defaults to None. A sample of input data to use to generate input schema.
y_sample: Union[list, pd.Series, np.ndarray, pd.DataFrame]. Defaults to None. A sample of output data to use to generate output schema.
training_script_path: str. Defaults to None. Training script path.
training_id: (str, optional). Defaults to the value from environment variables. The training OCID for model. Can be a notebook session or job OCID.
ignore_pending_changes: bool. Defaults to False. Whether to ignore the pending changes in Git.
max_col_num: (int, optional). Defaults to utils.DATA_SCHEMA_MAX_COL_NUM. The maximum column size of the data that allows you to automatically a generate schema.

kwargs: impute_values: (dict, optional). The dictionary where the key is the column index (or names is accepted for pandas dataframe), and the value is the impute value for the corresponding column.

It automatically generates the following files.

runtime.yaml
score.py. load_model function is only implemented when serialize=True, otherwise, you need to implement it on your own.
model.pkl by default model is serialized in pkl format when serialize=True in the init function. If serialize=False, the model isn’t automatically serialized.
input_schema.json when X_sample is passed in and the schema is more than 32 KB.
output_schema.json when y_sample is passed in and the schema is more than 32 KB.
hyperparameters.json if the extracted hyperparameters are more than 32KB.

Verify

The .verify() function takes one parameter:

data (Union[dict, str, tuple, list]). Data used to test if deployment works in a local environment.

You use it to test if deployment would work in the local environment. Before saving and deploying the model, we recommended that you call this function to check if the load_model and predict functions in score.py work. It takes and returns the same data as model deployment predict takes and returns.

In GenericModel data serialization is not supported, which means that you still must ensure that you pass in JSON serializable data to the verify and predict functions unless you implement extra data serialization and deserialization. Also, in the generated score.py template,

Save

The Save function takes the following parameters:

display_name: (str, optional). Defaults to None. The name of the model.
description: (str, optional). Defaults to None. The description of the model.
freeform_tags : Dict(str, str). Defaults to None. Free form tags for the model.
defined_tags : (Dict(str, dict(str, object)), optional). Defaults to None. Defined tags for the model.
ignore_introspection: (bool, optional). Defaults to None. Determines whether to ignore the result of model introspection or not. If set to True, the save ignores all model introspection errors.

kwargs: - project_id: (str, optional). Project OCID. If not specified, the value is taken either from the environment variables or model properties. - compartment_id : (str, optional). Compartment OCID. If not specified, the value is taken either from the environment variables or model properties. - timeout: (int, optional). Defaults to 10 seconds. The connection timeout in seconds for the client.

It first reloads the score.py and runtime.yaml files from disk so that any changes made to those files are used. Then it conducts an introspection test by default. However, you can set ignore_introspection=False to avoid it. Introspection tests check if .deployment() could have some issues and suggests necessary actions about how to fix them. Lastly, it uploads the artifacts to the model catalog, and returns a model_id for the saved model. You can also call .instrospect() to conduct the test any time after you call .prepare().

Deploy

.deploy() takes the following parameters:

wait_for_completion : (bool, optional). Defaults to True. Set to wait for the deployment to complete before proceeding.
display_name: (str, optional). Defaults to None. The name of the model.
description: (str, optional). Defaults to None. The description of the model.
deployment_instance_shape: (str, optional). Default to VM.Standard2.1. The shape of the instance used for deployment.
deployment_instance_count: (int, optional). Defaults to 1. The number of instances used for deployment.
deployment_bandwidth_mbps: (int, optional). Defaults to 10. The bandwidth limit on the load balancer in Mbps.
deployment_log_group_id: (str, optional). Defaults to None. The oci logging group id. The access log and predict log share the same log group.
deployment_access_log_id: (str, optional). Defaults to None. The access log OCID for the access logs, see link
deployment_predict_log_id: (str, optional). Defaults to None. The predict log OCID for the predict logs, see link

kwargs: - project_id: (str, optional). Project OCID. If not specified, the value is taken from the environment variables. - compartment_id : (str, optional). Compartment OCID. If not specified, the value is taken from the environment variables. - max_wait_time : (int, optional). Defaults to 1200 seconds. Maximum amount of time to wait in seconds. Negative implies an infinite wait time. - poll_interval : (int, optional). Defaults to 60 seconds. Poll interval in seconds.

It deploys the model. In order to make deployment more smooth, we suggest using exactly the same conda environments for both local development and deployment. Discrepancy between the two could cause problems.

You can pass in deployment_log_group_id, deployment_access_log_id and deployment_predict_log_id to enable the logging. To create a log group, you can reference Logging.

Logging Example

import tempfile
from ads.model.generic_model import GenericModel

log_group_ocid = 'ocid1.loggroup.oc1.iad.amaaaaaav...'
log_ocid = 'ocid1.log.oc1.iad.amaaaaaav66vvniagrvj...'

class Toy:
    def predict(self, x):
        return x ** 2
estimator = Toy()

model = GenericModel(estimator=estimator, artifact_dir=tempfile.mkdtemp())
model.summary_status()
model.prepare(inference_conda_env="dataexpl_p37_cpu_v3")
model.verify(2)
model.save()
model.deploy(
    deployment_log_group_id=log_group_ocid,
    deployment_access_log_id=log_ocid,
    deployment_predict_log_id =log_ocid,
)

model.predict(2)
model.model_deployment.show_logs(log_type="predict")
model.model_deployment.show_logs(log_type="access")
model.model_deployment.access_log.tail()
model.model_deployment.predict_log.tail()

Predict

.predict() takes one parameter: - data: Union[dict, str, tuple, list]. JSON serializable data for the prediction for Onnx models. For the local serialization method, the data can be the data types that each framework support.

.predict()``takes the same data that ``.verify() takes. You must ensure that the data passed and returned by predict in the score.py is JSON serializable. It passes the data to the model deployment endpoint and calls the predict function in score.py.

Delete_deployment

.delete_deployment() takes one parameter:

wait_for_completion: (bool, optional). Defaults to False. Whether to wait till completion.

When you don’t need the deployment any more, call delete_deployment to delete the current deployment that is attached to the model. Note that each time you call deploy, it creates a new deployment and only the new deployment is attached to the model.

from_model_artifact

.from_model_artifact() allows to load a model from a folder, zip or tar achive files, where the folder/zip/tar files should contain the files such as runtime.yaml, score.py, the serialized model file needed for deployments. It takes the following parameters:

uri: str: The folder path, ZIP file path, or TAR file path. It could contain a seriliazed model(required) as well as any files needed for deployment including: serialized model, runtime.yaml, score.py and etc. The content of the folder will be copied to the artifact_dir folder.
model_file_name: str: The serialized model file name.
artifact_dir: str: The artifact directory to store the files needed for deployment.
auth: (Dict, optional): Defaults to None. The default authetication is set using ads.set_auth API. If you need to override the default, use the ads.common.auth.api_keys or ads.common.auth.resource_principal to create appropriate authentication signer and kwargs required to instantiate IdentityClient object.
force_overwrite: (bool, optional): Defaults to False. Whether to overwrite existing files or not.
properties: (ModelProperties, optional): Defaults to None. ModelProperties object required to save and deploy model.

After this is called, you can call .verify(), .save() and etc.

from_model_catalog

from_model_catalog allows to load a remote model from model catalog using a model id , which should contain the files such as runtime.yaml, score.py, the serialized model file needed for deployments. It takes the following parameters:

model_id: str. The model OCID.
model_file_name: (str). The name of the serialized model.
artifact_dir: str. The artifact directory to store the files needed for deployment. Will be created if not exists.
auth: (Dict, optional). Defaults to None. The default authetication is set using ads.set_auth API. If you need to override the default, use the ads.common.auth.api_keys or ads.common.auth.resource_principal to create appropriate authentication signer and kwargs required to instantiate IdentityClient object.
force_overwrite: (bool, optional). Defaults to False. Whether to overwrite existing files or not.
properties: (ModelProperties, optional). Defaults to None. ModelProperties object required to save and deploy model.

kwargs:

compartment_id : (str, optional). Compartment OCID. If not specified, the value will be taken from the environment variables.
timeout : (int, optional). Defaults to 10 seconds. The connection timeout in seconds for the client.

Examples

Serialize the Model to pkl Format by Default

import tempfile
from ads.model.generic_model import GenericModel

class Toy:
    def predict(self, x):
        return x ** 2
estimator = Toy()

model = GenericModel(estimator=estimator, artifact_dir=tempfile.mkdtemp())
model.summary_status()
model.prepare(inference_conda_env="oci://service-conda-packs@id19sfcrra6z/service_pack/cpu/Data Exploration and Manipulation for CPU Python 3.7/3.0/dataexpl_p37_cpu_v3",
              inference_python_version="3.7",
              model_file_name="toy_model.pkl",
              training_id=None,
              force_overwrite=True
            )
model.verify(2)
model.save()
model.deploy()
model.predict(2)
model.delete_deployment()

Serialize the Model In Your Way

import tempfile
from ads.model.generic_model import GenericModel

class Toy:
    def predict(self, x):
        return x ** 2
estimator = Toy()
model = GenericModel(estimator=estimator,
                     artifact_dir=tempfile.mkdtemp(),
                     serialize=False)
model.summary_status()
model.prepare(inference_conda_env="oci://service-conda-packs@id19sfcrra6z/service_pack/cpu/Data Exploration and Manipulation for CPU Python 3.7/3.0/dataexpl_p37_cpu_v3",
              inference_python_version="3.7",
              model_file_name="toy_model.pkl",
              training_id=None,
              force_overwrite=True
            )
model.summary_status()

# Serialize the model
import cloudpickle
import os

with open(os.path.join(model.artifact_dir, "toy_model.pkl"), "wb") as outfile:
    cloudpickle.dump(estimator, outfile)
# Finish implementing the score.py
score = '''
import json
import os
from cloudpickle import cloudpickle


model_name = 'toy_model.pkl'

def load_model(model_file_name=model_name):
    """
    Loads model from the serialized format

    Returns
    -------
    model:  a model instance on which predict API can be invoked
    """
    model_dir = os.path.dirname(os.path.realpath(__file__))
    contents = os.listdir(model_dir)
    if model_file_name in contents:
        with open(os.path.join(os.path.dirname(os.path.realpath(__file__)), model_file_name), "rb") as file:
            return cloudpickle.load(file)
    else:
        raise Exception('{0} is not found in model directory {1}'.format(model_file_name, model_dir))


def predict(data, model=load_model()):
    """
    Returns prediction given the model and data to predict

    Parameters
    ----------
    model: Model instance returned by load_model API
    data: Data format as expected by the predict API of the core estimator. For eg. in case of sckit models it could be numpy array/List of list/Panda DataFrame

    Returns
    -------
    predictions: Output from scoring server
        Format: {'prediction':output from model.predict method}

    """

    return {'prediction':model.predict(data)}
'''

with open(os.path.join(model.artifact_dir, "score.py"), 'w') as f:
    print(f.write(score))
model.verify(2)
model.save()
model.deploy(deployment_log_group_id="ocid1.loggroup.oc1.iad.amaaaa...",
            deployment_access_log_id="ocid1.log.oc1.iad.amaaaaaav...",
            deployment_predict_log_id="ocid1.log.oc1.iad.amaaaaaav...")
model.delete_deployment()

Loading Model From a Zip Archive

model = GenericModel.from_model_artifact("/folder_to_your/artifact.zip",
                                      model_file_name="your_model_file_name",
                                      artifact_dir=tempfile.mkdtemp())

model.verify(your_data)

Loading Model From Model Catalog

model = GenericModel.from_model_catalog(model_id="ocid1.datasciencemodel.oc1.iad.amaaaa....",
                                      model_file_name="your_model_file_name",
                                      artifact_dir=tempfile.mkdtemp())
model.verify(your_data)