API reference

Orbit

The mlflavors.orbit module provides an API for logging and loading orbit models. This module exports orbit models with the following flavors:

orbit (native) format

This is the main flavor that can be loaded back into orbit, which relies on pickle internally to serialize a model.

Note that pickle serialization requires using the same python environment (version) in whatever environment you’re going to use this model for inference to ensure that the model will load with appropriate version of pickle.

mlflow.pyfunc format

Produced for use by generic pyfunc-based deployment tools and batch inference.

The interface for utilizing an orbit model loaded as a pyfunc type for generating forecast predictions uses a single-row Pandas DataFrame configuration argument. The following columns in this configuration Pandas DataFrame are supported:

Column

Type

Description

X

numpy ndarray or list (required)
Exogenous regressor for future time period events.
For more information, read the underlying library explanation:
https://orbit-ml.readthedocs.io/en/latest/.

X_cols

list (required)
Column names of the exogenous regressor matrix
(Required to construct Pandas DataFrame inside model wrapper class).

X_dtypes (required)

list (required)
Data types of the exogenous regressor matrix
(Required to construct Pandas DataFrame inside model wrapper class).

decompose

bool (optional)
If True, returns each prediction component separately.
(Default: False)

store_prediction_array

bool (optional)
If True, prediction array is stored.
(Default: False)

seed

int (optional)
Seed in prediction is set to be random by default unless provided.
(Default: None)
mlflavors.orbit.get_default_conda_env(include_cloudpickle=False)[source]
Returns

The default Conda environment for MLflow Models produced by calls to save_model() and log_model().

mlflavors.orbit.get_default_pip_requirements(include_cloudpickle=False)[source]
Returns

A list of default pip requirements for MLflow Models produced by this flavor. Calls to save_model() and log_model() produce a pip environment that, at minimum, contains these requirements.

mlflavors.orbit.load_model(model_uri, dst_path=None)[source]

Load an orbit model from a local file or a run.

Parameters

  • model_uri –

    The location, in URI format, of the MLflow model, for example:

    • /Users/me/path/to/local/model

    • relative/path/to/local/model

    • s3://my_bucket/path/to/model

    • runs:/<mlflow_run_id>/run-relative/path/to/model

    • models:/<model_name>/<model_version>

    • models:/<model_name>/<stage>

    For more information about supported URI schemes, see Referencing Artifacts.

  • dst_path – The local filesystem path to which to download the model artifact. This directory must already exist. If unspecified, a local output path will be created.

Returns

An orbit model.

mlflavors.orbit.log_model(orbit_model, artifact_path, conda_env=None, code_paths=None, registered_model_name=None, signature=None, input_example=None, await_registration_for=300, pip_requirements=None, extra_pip_requirements=None, serialization_format='pickle', **kwargs)[source]

Log an orbit model as an MLflow artifact for the current run. Produces an MLflow Model containing the following flavors:

Parameters

  • orbit_model – Fitted orbit model object.

  • artifact_path – Run-relative artifact path to save the model instance to.

  • conda_env –

    Either a dictionary representation of a Conda environment or the path to a conda environment yaml file. If provided, this describes the environment this model should be run in. At minimum, it should specify the dependencies contained in get_default_conda_env(). If None, a conda environment with pip requirements inferred by mlflow.models.infer_pip_requirements() is added to the model. If the requirement inference fails, it falls back to using get_default_pip_requirements(). pip requirements from conda_env are written to a pip requirements.txt file and the full conda environment is written to conda.yaml. The following is an example dictionary representation of a conda environment:

    {
    "name": "mlflow-env",
    "channels": ["conda-forge"],
    "dependencies": [
        "python=3.8.15",
        {
            "pip": [
                "orbit==x.y.z"
            ],
        },
    ],
}

  • code_paths – A list of local filesystem paths to Python file dependencies (or directories containing file dependencies). These files are prepended to the system path when the model is loaded.

  • registered_model_name – This argument may change or be removed in a future release without warning. If given, create a model version under registered_model_name, also creating a registered model if one with the given name does not exist.

  • signature –

    Model Signature mlflow.models.ModelSignature describes model input and output Schema. The model signature can be inferred from datasets with valid model input (e.g. the training dataset with target column omitted) and valid model output (e.g. model predictions generated on the training dataset), for example:

    from mlflow.models.signature import infer_signature

train = df.drop_column("target_label")
predictions = ...  # compute model predictions
signature = infer_signature(train, predictions)

  • input_example – Input example provides one or several instances of valid model input. The example can be used as a hint of what data to feed the model. The given example will be converted to a Pandas DataFrame and then serialized to json using the Pandas split-oriented format. Bytes are base64-encoded.

  • await_registration_for – Number of seconds to wait for the model version to finish being created and is in READY status. By default, the function waits for five minutes. Specify 0 or None to skip waiting.

  • pip_requirements – Either an iterable of pip requirement strings (e.g. ["orbit", "-r requirements.txt", "-c constraints.txt"]) or the string path to a pip requirements file on the local filesystem (e.g. "requirements.txt"). If provided, this describes the environment this model should be run in. If None, a default list of requirements is inferred by mlflow.models.infer_pip_requirements() from the current software environment. If the requirement inference fails, it falls back to using get_default_pip_requirements(). Both requirements and constraints are automatically parsed and written to requirements.txt and constraints.txt files, respectively, and stored as part of the model. Requirements are also written to the pip section of the model’s conda environment (conda.yaml) file.

  • extra_pip_requirements –

    Either an iterable of pip requirement strings (e.g. ["pandas", "-r requirements.txt", "-c constraints.txt"]) or the string path to a pip requirements file on the local filesystem (e.g. "requirements.txt"). If provided, this describes additional pip requirements that are appended to a default set of pip requirements generated automatically based on the user’s current software environment. Both requirements and constraints are automatically parsed and written to requirements.txt and constraints.txt files, respectively, and stored as part of the model. Requirements are also written to the pip section of the model’s conda environment (conda.yaml) file.

    Warning

    The following arguments can’t be specified at the same time:

    • conda_env

    • pip_requirements

    • extra_pip_requirements

    This example demonstrates how to specify pip requirements using pip_requirements and extra_pip_requirements.

  • serialization_format – The format in which to serialize the model. This should be one of the formats “pickle” or “cloudpickle”

Returns

A ModelInfo instance that contains the metadata of the logged model.

mlflavors.orbit.save_model(orbit_model, path, conda_env=None, code_paths=None, mlflow_model=None, signature=None, input_example=None, pip_requirements=None, extra_pip_requirements=None, serialization_format='pickle')[source]

Save an orbit model to a path on the local file system. Produces an MLflow Model containing the following flavors:

Parameters

  • orbit_model – Fitted orbit model object.

  • path – Local path where the model is to be saved.

  • conda_env –

    Either a dictionary representation of a Conda environment or the path to a conda environment yaml file. If provided, this describes the environment this model should be run in. At minimum, it should specify the dependencies contained in get_default_conda_env(). If None, a conda environment with pip requirements inferred by mlflow.models.infer_pip_requirements() is added to the model. If the requirement inference fails, it falls back to using get_default_pip_requirements(). pip requirements from conda_env are written to a pip requirements.txt file and the full conda environment is written to conda.yaml. The following is an example dictionary representation of a conda environment:

    {
    "name": "mlflow-env",
    "channels": ["conda-forge"],
    "dependencies": [
        "python=3.8.15",
        {
            "pip": [
                "orbit==x.y.z"
            ],
        },
    ],
}

  • code_paths – A list of local filesystem paths to Python file dependencies (or directories containing file dependencies). These files are prepended to the system path when the model is loaded.

  • mlflow_model – mlflow.models.Model configuration to which to add the python_function flavor.

  • signature –

    Model Signature mlflow.models.ModelSignature describes model input and output Schema. The model signature can be inferred from datasets with valid model input (e.g. the training dataset with target column omitted) and valid model output (e.g. model predictions generated on the training dataset), for example:

    from mlflow.models.signature import infer_signature

train = df.drop_column("target_label")
predictions = ...  # compute model predictions
signature = infer_signature(train, predictions)

  • input_example – Input example provides one or several instances of valid model input.The example can be used as a hint of what data to feed the model. The given example will be converted to a Pandas DataFrame and then serialized to json using the Pandas split-oriented format. Bytes are base64-encoded.

  • pip_requirements – Either an iterable of pip requirement strings (e.g. ["orbit", "-r requirements.txt", "-c constraints.txt"]) or the string path to a pip requirements file on the local filesystem (e.g. "requirements.txt"). If provided, this describes the environment this model should be run in. If None, a default list of requirements is inferred by mlflow.models.infer_pip_requirements() from the current software environment. If the requirement inference fails, it falls back to using get_default_pip_requirements(). Both requirements and constraints are automatically parsed and written to requirements.txt and constraints.txt files, respectively, and stored as part of the model. Requirements are also written to the pip section of the model’s conda environment (conda.yaml) file.

  • extra_pip_requirements –

    Either an iterable of pip requirement strings (e.g. ["pandas", "-r requirements.txt", "-c constraints.txt"]) or the string path to a pip requirements file on the local filesystem (e.g. "requirements.txt"). If provided, this describes additional pip requirements that are appended to a default set of pip requirements generated automatically based on the user’s current software environment. Both requirements and constraints are automatically parsed and written to requirements.txt and constraints.txt files, respectively, and stored as part of the model. Requirements are also written to the pip section of the model’s conda environment (conda.yaml) file.

    Warning

    The following arguments can’t be specified at the same time:

    • conda_env

    • pip_requirements

    • extra_pip_requirements

    This example demonstrates how to specify pip requirements using pip_requirements and extra_pip_requirements.

  • serialization_format – The format in which to serialize the model. This should be one of the formats “pickle” or “cloudpickle”

Sktime

The mlflavors.sktime module provides an API for logging and loading sktime models. This module exports sktime models with the following flavors:

sktime (native) format

This is the main flavor that can be loaded back into sktime, which relies on pickle internally to serialize a model.

Note that pickle serialization requires using the same python environment (version) in whatever environment you’re going to use this model for inference to ensure that the model will load with appropriate version of pickle.

mlflow.pyfunc format

Produced for use by generic pyfunc-based deployment tools and batch inference.

The interface for utilizing a sktime model loaded as a pyfunc type for generating forecast predictions uses a single-row Pandas DataFrame configuration argument. The following columns in this configuration Pandas DataFrame are supported:

Column

Type

Description

predict_method

str (required)
Specifies the sktime predict method. The supported predict methods are
predict, predict_interval, predict_quantiles, and
predict_var.

fh

list (optional)
Specifies the number of future periods to generate starting from the last
datetime value of the training dataset, utilizing the frequency of the input
training series when the model was trained. (for example, if the training
data series elements represent one value per hour, in order to forecast 3
hours of future data, set the column fh to [1,2,3]. If the parameter
is not provided it must be passed during fit().
(Default: None)

X_dtypes

numpy ndarray or list (optional)
Exogenous regressor for future time period events.
For more information, read the underlying library explanation:
https://www.sktime.net/en/latest/examples/AA_datatypes_and_datasets.html#Section-1:-in-memory-data-containers.

coverage

float (optional)
The nominal coverage value for calculating prediction interval forecasts.
Can only be provided in combination with predict method
predict_interval.
(Default: 0.9)

alpha

float (optional)
The probability value for calculating prediction quantile forecasts.
Can only be provided in combination with predict method
predict_quantiles.
(Default: None)

cov

bool (optional)
If True, computes covariance matrix forecast.
Can only be provided in combination with predict method predict_var.
(Default: False)

An example configuration for the pyfunc predict of a sktime model is shown below, using an interval forecast with nominal coverage value [0.9,0.95], a future forecast horizon of 3 periods, and no exogenous regressor elements:

Index

predict_method

coverage

fh

0

predict_interval

[0.9,0.95]

[1,2,3]
mlflavors.sktime.get_default_conda_env(include_cloudpickle=False)[source]
Returns

The default Conda environment for MLflow Models produced by calls to save_model() and log_model().

mlflavors.sktime.get_default_pip_requirements(include_cloudpickle=False)[source]
Returns

A list of default pip requirements for MLflow Models produced by this flavor. Calls to save_model() and log_model() produce a pip environment that, at minimum, contains these requirements.

mlflavors.sktime.load_model(model_uri, dst_path=None)[source]

Load a sktime model from a local file or a run.

Parameters

  • model_uri –

    The location, in URI format, of the MLflow model, for example:

    • /Users/me/path/to/local/model

    • relative/path/to/local/model

    • s3://my_bucket/path/to/model

    • runs:/<mlflow_run_id>/run-relative/path/to/model

    • models:/<model_name>/<model_version>

    • models:/<model_name>/<stage>

    For more information about supported URI schemes, see Referencing Artifacts.

  • dst_path – The local filesystem path to which to download the model artifact. This directory must already exist. If unspecified, a local output path will be created.

Returns

A sktime model.

mlflavors.sktime.log_model(sktime_model, artifact_path, conda_env=None, code_paths=None, registered_model_name=None, signature=None, input_example=None, await_registration_for=300, pip_requirements=None, extra_pip_requirements=None, serialization_format='pickle', **kwargs)[source]

Log a sktime model as an MLflow artifact for the current run. Produces an MLflow Model containing the following flavors:

Parameters

  • sktime_model – Fitted sktime model object.

  • artifact_path – Run-relative artifact path to save the model instance to.

  • conda_env –

    Either a dictionary representation of a Conda environment or the path to a conda environment yaml file. If provided, this describes the environment this model should be run in. At minimum, it should specify the dependencies contained in get_default_conda_env(). If None, a conda environment with pip requirements inferred by mlflow.models.infer_pip_requirements() is added to the model. If the requirement inference fails, it falls back to using get_default_pip_requirements(). pip requirements from conda_env are written to a pip requirements.txt file and the full conda environment is written to conda.yaml. The following is an example dictionary representation of a conda environment:

    {
    "name": "mlflow-env",
    "channels": ["conda-forge"],
    "dependencies": [
        "python=3.8.15",
        {
            "pip": [
                "sktime==x.y.z"
            ],
        },
    ],
}

  • code_paths – A list of local filesystem paths to Python file dependencies (or directories containing file dependencies). These files are prepended to the system path when the model is loaded.

  • registered_model_name – This argument may change or be removed in a future release without warning. If given, create a model version under registered_model_name, also creating a registered model if one with the given name does not exist.

  • signature –

    Model Signature mlflow.models.ModelSignature describes model input and output Schema. The model signature can be inferred from datasets with valid model input (e.g. the training dataset with target column omitted) and valid model output (e.g. model predictions generated on the training dataset), for example:

    from mlflow.models.signature import infer_signature

train = df.drop_column("target_label")
predictions = ...  # compute model predictions
signature = infer_signature(train, predictions)

    Warning

    If performing probabilistic forecasts (predict_interval, predict_quantiles) with a sktime model, the signature on the returned prediction object will not be correctly inferred due to the Pandas MultiIndex column type when using the these methods. infer_schema will function correctly if using the pyfunc flavor of the model, though.

  • input_example – Input example provides one or several instances of valid model input. The example can be used as a hint of what data to feed the model. The given example will be converted to a Pandas DataFrame and then serialized to json using the Pandas split-oriented format. Bytes are base64-encoded.

  • await_registration_for – Number of seconds to wait for the model version to finish being created and is in READY status. By default, the function waits for five minutes. Specify 0 or None to skip waiting.

  • pip_requirements – Either an iterable of pip requirement strings (e.g. ["sktime", "-r requirements.txt", "-c constraints.txt"]) or the string path to a pip requirements file on the local filesystem (e.g. "requirements.txt"). If provided, this describes the environment this model should be run in. If None, a default list of requirements is inferred by mlflow.models.infer_pip_requirements() from the current software environment. If the requirement inference fails, it falls back to using get_default_pip_requirements(). Both requirements and constraints are automatically parsed and written to requirements.txt and constraints.txt files, respectively, and stored as part of the model. Requirements are also written to the pip section of the model’s conda environment (conda.yaml) file.

  • extra_pip_requirements –

    Either an iterable of pip requirement strings (e.g. ["pandas", "-r requirements.txt", "-c constraints.txt"]) or the string path to a pip requirements file on the local filesystem (e.g. "requirements.txt"). If provided, this describes additional pip requirements that are appended to a default set of pip requirements generated automatically based on the user’s current software environment. Both requirements and constraints are automatically parsed and written to requirements.txt and constraints.txt files, respectively, and stored as part of the model. Requirements are also written to the pip section of the model’s conda environment (conda.yaml) file.

    Warning

    The following arguments can’t be specified at the same time:

    • conda_env

    • pip_requirements

    • extra_pip_requirements

    This example demonstrates how to specify pip requirements using pip_requirements and extra_pip_requirements.

  • serialization_format – The format in which to serialize the model. This should be one of the formats “pickle” or “cloudpickle”

Returns

A ModelInfo instance that contains the metadata of the logged model.

mlflavors.sktime.save_model(sktime_model, path, conda_env=None, code_paths=None, mlflow_model=None, signature=None, input_example=None, pip_requirements=None, extra_pip_requirements=None, serialization_format='pickle')[source]

Save a sktime model to a path on the local file system. Produces an MLflow Model containing the following flavors:

Parameters

  • sktime_model – Fitted sktime model object.

  • path – Local path where the model is to be saved.

  • conda_env –

    Either a dictionary representation of a Conda environment or the path to a conda environment yaml file. If provided, this describes the environment this model should be run in. At minimum, it should specify the dependencies contained in get_default_conda_env(). If None, a conda environment with pip requirements inferred by mlflow.models.infer_pip_requirements() is added to the model. If the requirement inference fails, it falls back to using get_default_pip_requirements(). pip requirements from conda_env are written to a pip requirements.txt file and the full conda environment is written to conda.yaml. The following is an example dictionary representation of a conda environment:

    {
    "name": "mlflow-env",
    "channels": ["conda-forge"],
    "dependencies": [
        "python=3.8.15",
        {
            "pip": [
                "sktime==x.y.z"
            ],
        },
    ],
}

  • code_paths – A list of local filesystem paths to Python file dependencies (or directories containing file dependencies). These files are prepended to the system path when the model is loaded.

  • mlflow_model – mlflow.models.Model configuration to which to add the python_function flavor.

  • signature –

    Model Signature mlflow.models.ModelSignature describes model input and output Schema. The model signature can be inferred from datasets with valid model input (e.g. the training dataset with target column omitted) and valid model output (e.g. model predictions generated on the training dataset), for example:

    from mlflow.models.signature import infer_signature

train = df.drop_column("target_label")
predictions = ...  # compute model predictions
signature = infer_signature(train, predictions)

    Warning

    If performing probabilistic forecasts (predict_interval, predict_quantiles) with a sktime model, the signature on the returned prediction object will not be correctly inferred due to the Pandas MultiIndex column type when using the these methods. infer_schema will function correctly if using the pyfunc flavor of the model, though.

  • input_example – Input example provides one or several instances of valid model input. The example can be used as a hint of what data to feed the model. The given example will be converted to a Pandas DataFrame and then serialized to json using the Pandas split-oriented format. Bytes are base64-encoded.

  • pip_requirements – Either an iterable of pip requirement strings (e.g. ["sktime", "-r requirements.txt", "-c constraints.txt"]) or the string path to a pip requirements file on the local filesystem (e.g. "requirements.txt"). If provided, this describes the environment this model should be run in. If None, a default list of requirements is inferred by mlflow.models.infer_pip_requirements() from the current software environment. If the requirement inference fails, it falls back to using get_default_pip_requirements(). Both requirements and constraints are automatically parsed and written to requirements.txt and constraints.txt files, respectively, and stored as part of the model. Requirements are also written to the pip section of the model’s conda environment (conda.yaml) file.

  • extra_pip_requirements –

    Either an iterable of pip requirement strings (e.g. ["pandas", "-r requirements.txt", "-c constraints.txt"]) or the string path to a pip requirements file on the local filesystem (e.g. "requirements.txt"). If provided, this describes additional pip requirements that are appended to a default set of pip requirements generated automatically based on the user’s current software environment. Both requirements and constraints are automatically parsed and written to requirements.txt and constraints.txt files, respectively, and stored as part of the model. Requirements are also written to the pip section of the model’s conda environment (conda.yaml) file.

    Warning

    The following arguments can’t be specified at the same time:

    • conda_env

    • pip_requirements

    • extra_pip_requirements

    This example demonstrates how to specify pip requirements using pip_requirements and extra_pip_requirements.

  • serialization_format – The format in which to serialize the model. This should be one of the formats “pickle” or “cloudpickle”

StatsForecast

The mlflavors.statsforecast module provides an API for logging and loading statsforecast models. This module exports statsforecast models with the following flavors:

statsforecast (native) format

This is the main flavor that can be loaded back into statsforecast, which relies on pickle internally to serialize a model.

Note that pickle serialization requires using the same python environment (version) in whatever environment you’re going to use this model for inference to ensure that the model will load with appropriate version of pickle.

mlflow.pyfunc format

Produced for use by generic pyfunc-based deployment tools and batch inference.

The interface for utilizing an statsforecast model loaded as a pyfunc type for generating forecast predictions uses a single-row Pandas DataFrame configuration argument. The following columns in this configuration Pandas DataFrame are supported:

Column

Type

Description

X

numpy ndarray or list (optional)
Exogenous regressor for future time period events.
For more information, read the underlying library explanation:
https://nixtla.github.io/statsforecast/.
(Default: None)

X_cols

list (optional)
Column names of the exogenous regressor matrix
(Required to construct Pandas DataFrame inside model wrapper class).
(Default: None)

X_dtypes

list (optional)
Data types of the exogenous regressor matrix
(Required to construct Pandas DataFrame inside model wrapper class).
(Default: None)

h

int (required)
Specifies the number of future periods to generate starting from the last
datetime value of the training dataset, utilizing the frequency of the input
training series when the model was trained. (for example, if the training
data series elements represent one value per hour, in order to forecast 3
hours of future data, set the column fh to 3.

level

list (optional)
A list of floats with the confidence levels of the prediction intervals. For
example, level=[95] means that the range of values should include the
actual future value with probability 95%.
(Default: None)
mlflavors.statsforecast.get_default_conda_env(include_cloudpickle=False)[source]
Returns

The default Conda environment for MLflow Models produced by calls to save_model() and log_model().

mlflavors.statsforecast.get_default_pip_requirements(include_cloudpickle=False)[source]
Returns

A list of default pip requirements for MLflow Models produced by this flavor. Calls to save_model() and log_model() produce a pip environment that, at minimum, contains these requirements.

mlflavors.statsforecast.load_model(model_uri, dst_path=None)[source]

Load an statsforecast model from a local file or a run.

Parameters

  • model_uri –

    The location, in URI format, of the MLflow model, for example:

    • /Users/me/path/to/local/model

    • relative/path/to/local/model

    • s3://my_bucket/path/to/model

    • runs:/<mlflow_run_id>/run-relative/path/to/model

    • models:/<model_name>/<model_version>

    • models:/<model_name>/<stage>

    For more information about supported URI schemes, see Referencing Artifacts.

  • dst_path – The local filesystem path to which to download the model artifact. This directory must already exist. If unspecified, a local output path will be created.

Returns

An statsforecast model.

mlflavors.statsforecast.log_model(statsforecast_model, artifact_path, conda_env=None, code_paths=None, registered_model_name=None, signature=None, input_example=None, await_registration_for=300, pip_requirements=None, extra_pip_requirements=None, serialization_format='pickle', **kwargs)[source]

Log an statsforecast model as an MLflow artifact for the current run. Produces an MLflow Model containing the following flavors:

Parameters

  • statsforecast_model – Fitted statsforecast model object.

  • artifact_path – Run-relative artifact path to save the model instance to.

  • conda_env –

    Either a dictionary representation of a Conda environment or the path to a conda environment yaml file. If provided, this describes the environment this model should be run in. At minimum, it should specify the dependencies contained in get_default_conda_env(). If None, a conda environment with pip requirements inferred by mlflow.models.infer_pip_requirements() is added to the model. If the requirement inference fails, it falls back to using get_default_pip_requirements(). pip requirements from conda_env are written to a pip requirements.txt file and the full conda environment is written to conda.yaml. The following is an example dictionary representation of a conda environment:

    {
    "name": "mlflow-env",
    "channels": ["conda-forge"],
    "dependencies": [
        "python=3.8.15",
        {
            "pip": [
                "statsforecast==x.y.z"
            ],
        },
    ],
}

  • code_paths – A list of local filesystem paths to Python file dependencies (or directories containing file dependencies). These files are prepended to the system path when the model is loaded.

  • registered_model_name – This argument may change or be removed in a future release without warning. If given, create a model version under registered_model_name, also creating a registered model if one with the given name does not exist.

  • signature –

    Model Signature mlflow.models.ModelSignature describes model input and output Schema. The model signature can be inferred from datasets with valid model input (e.g. the training dataset with target column omitted) and valid model output (e.g. model predictions generated on the training dataset), for example:

    from mlflow.models.signature import infer_signature

train = df.drop_column("target_label")
predictions = ...  # compute model predictions
signature = infer_signature(train, predictions)

  • input_example – Input example provides one or several instances of valid model input. The example can be used as a hint of what data to feed the model. The given example will be converted to a Pandas DataFrame and then serialized to json using the Pandas split-oriented format. Bytes are base64-encoded.

  • await_registration_for – Number of seconds to wait for the model version to finish being created and is in READY status. By default, the function waits for five minutes. Specify 0 or None to skip waiting.

  • pip_requirements – Either an iterable of pip requirement strings (e.g. ["statsforecast", "-r requirements.txt", "-c constraints.txt"]) or the string path to a pip requirements file on the local filesystem (e.g. "requirements.txt"). If provided, this describes the environment this model should be run in. If None, a default list of requirements is inferred by mlflow.models.infer_pip_requirements() from the current software environment. If the requirement inference fails, it falls back to using get_default_pip_requirements(). Both requirements and constraints are automatically parsed and written to requirements.txt and constraints.txt files, respectively, and stored as part of the model. Requirements are also written to the pip section of the model’s conda environment (conda.yaml) file.

  • extra_pip_requirements –

    Either an iterable of pip requirement strings (e.g. ["pandas", "-r requirements.txt", "-c constraints.txt"]) or the string path to a pip requirements file on the local filesystem (e.g. "requirements.txt"). If provided, this describes additional pip requirements that are appended to a default set of pip requirements generated automatically based on the user’s current software environment. Both requirements and constraints are automatically parsed and written to requirements.txt and constraints.txt files, respectively, and stored as part of the model. Requirements are also written to the pip section of the model’s conda environment (conda.yaml) file.

    Warning

    The following arguments can’t be specified at the same time:

    • conda_env

    • pip_requirements

    • extra_pip_requirements

    This example demonstrates how to specify pip requirements using pip_requirements and extra_pip_requirements.

  • serialization_format – The format in which to serialize the model. This should be one of the formats “pickle” or “cloudpickle”

Returns

A ModelInfo instance that contains the metadata of the logged model.

mlflavors.statsforecast.save_model(statsforecast_model, path, conda_env=None, code_paths=None, mlflow_model=None, signature=None, input_example=None, pip_requirements=None, extra_pip_requirements=None, serialization_format='pickle')[source]

Save an statsforecast model to a path on the local file system. Produces an MLflow Model containing the following flavors:

Parameters

  • statsforecast_model – Fitted statsforecast model object.

  • path – Local path where the model is to be saved.

  • conda_env –

    Either a dictionary representation of a Conda environment or the path to a conda environment yaml file. If provided, this describes the environment this model should be run in. At minimum, it should specify the dependencies contained in get_default_conda_env(). If None, a conda environment with pip requirements inferred by mlflow.models.infer_pip_requirements() is added to the model. If the requirement inference fails, it falls back to using get_default_pip_requirements(). pip requirements from conda_env are written to a pip requirements.txt file and the full conda environment is written to conda.yaml. The following is an example dictionary representation of a conda environment:

    {
    "name": "mlflow-env",
    "channels": ["conda-forge"],
    "dependencies": [
        "python=3.8.15",
        {
            "pip": [
                "statsforecast==x.y.z"
            ],
        },
    ],
}

  • code_paths – A list of local filesystem paths to Python file dependencies (or directories containing file dependencies). These files are prepended to the system path when the model is loaded.

  • mlflow_model – mlflow.models.Model configuration to which to add the python_function flavor.

  • signature –

    Model Signature mlflow.models.ModelSignature describes model input and output Schema. The model signature can be inferred from datasets with valid model input (e.g. the training dataset with target column omitted) and valid model output (e.g. model predictions generated on the training dataset), for example:

    from mlflow.models.signature import infer_signature

train = df.drop_column("target_label")
predictions = ...  # compute model predictions
signature = infer_signature(train, predictions)

  • input_example – Input example provides one or several instances of valid model input.The example can be used as a hint of what data to feed the model. The given example will be converted to a Pandas DataFrame and then serialized to json using the Pandas split-oriented format. Bytes are base64-encoded.

  • pip_requirements – Either an iterable of pip requirement strings (e.g. ["statsforecast", "-r requirements.txt", "-c constraints.txt"]) or the string path to a pip requirements file on the local filesystem (e.g. "requirements.txt"). If provided, this describes the environment this model should be run in. If None, a default list of requirements is inferred by mlflow.models.infer_pip_requirements() from the current software environment. If the requirement inference fails, it falls back to using get_default_pip_requirements(). Both requirements and constraints are automatically parsed and written to requirements.txt and constraints.txt files, respectively, and stored as part of the model. Requirements are also written to the pip section of the model’s conda environment (conda.yaml) file.

  • extra_pip_requirements –

    Either an iterable of pip requirement strings (e.g. ["pandas", "-r requirements.txt", "-c constraints.txt"]) or the string path to a pip requirements file on the local filesystem (e.g. "requirements.txt"). If provided, this describes additional pip requirements that are appended to a default set of pip requirements generated automatically based on the user’s current software environment. Both requirements and constraints are automatically parsed and written to requirements.txt and constraints.txt files, respectively, and stored as part of the model. Requirements are also written to the pip section of the model’s conda environment (conda.yaml) file.

    Warning

    The following arguments can’t be specified at the same time:

    • conda_env

    • pip_requirements

    • extra_pip_requirements

    This example demonstrates how to specify pip requirements using pip_requirements and extra_pip_requirements.

  • serialization_format – The format in which to serialize the model. This should be one of the formats “pickle” or “cloudpickle”

PyOD

The mlflavors.pyod module provides an API for logging and loading pyod models. This module exports pyod models with the following flavors:

pyod (native) format

This is the main flavor that can be loaded back into pyod, which relies on pickle internally to serialize a model.

Note that pickle serialization requires using the same python environment (version) in whatever environment you’re going to use this model for inference to ensure that the model will load with appropriate version of pickle.

mlflow.pyfunc format

Produced for use by generic pyfunc-based deployment tools and batch inference.

The interface for utilizing an pyod model loaded as a pyfunc type for generating predictions uses a single-row Pandas DataFrame configuration argument. The following columns in this configuration Pandas DataFrame are supported:

Column

Type

Description

predict_method

str (required)
Specifies the pyod predict method. The supported predict methods are
predict, predict_proba, predict_confidence, and
decision_function.

X

numpy ndarray or list (required)
The input samples.
For more information, read the underlying library explanation:
https://pyod.readthedocs.io/en/latest/index.html.

return_confidence

bool (optional)
If True, returns prediction confidence as well.
Can only be provided in combination with predict method predict and
predict_proba.
(Default: False)

method

str (optional)
The probability conversion method.
Can only be provided in combination with predict method predict_proba.
(Default: linear)
mlflavors.pyod.get_default_conda_env(include_cloudpickle=False)[source]
Returns

The default Conda environment for MLflow Models produced by calls to save_model() and log_model().

mlflavors.pyod.get_default_pip_requirements(include_cloudpickle=False)[source]
Returns

A list of default pip requirements for MLflow Models produced by this flavor. Calls to save_model() and log_model() produce a pip environment that, at minimum, contains these requirements.

mlflavors.pyod.load_model(model_uri, dst_path=None)[source]

Load an pyod model from a local file or a run.

Parameters

  • model_uri –

    The location, in URI format, of the MLflow model, for example:

    • /Users/me/path/to/local/model

    • relative/path/to/local/model

    • s3://my_bucket/path/to/model

    • runs:/<mlflow_run_id>/run-relative/path/to/model

    • models:/<model_name>/<model_version>

    • models:/<model_name>/<stage>

    For more information about supported URI schemes, see Referencing Artifacts.

  • dst_path – The local filesystem path to which to download the model artifact. This directory must already exist. If unspecified, a local output path will be created.

Returns

An pyod model.

mlflavors.pyod.log_model(pyod_model, artifact_path, conda_env=None, code_paths=None, registered_model_name=None, signature=None, input_example=None, await_registration_for=300, pip_requirements=None, extra_pip_requirements=None, serialization_format='pickle', **kwargs)[source]

Log an pyod model as an MLflow artifact for the current run. Produces an MLflow Model containing the following flavors:

Parameters

  • pyod_model – Fitted pyod model object.

  • artifact_path – Run-relative artifact path to save the model instance to.

  • conda_env –

    Either a dictionary representation of a Conda environment or the path to a conda environment yaml file. If provided, this describes the environment this model should be run in. At minimum, it should specify the dependencies contained in get_default_conda_env(). If None, a conda environment with pip requirements inferred by mlflow.models.infer_pip_requirements() is added to the model. If the requirement inference fails, it falls back to using get_default_pip_requirements(). pip requirements from conda_env are written to a pip requirements.txt file and the full conda environment is written to conda.yaml. The following is an example dictionary representation of a conda environment:

    {
    "name": "mlflow-env",
    "channels": ["conda-forge"],
    "dependencies": [
        "python=3.8.15",
        {
            "pip": [
                "pyod==x.y.z"
            ],
        },
    ],
}

  • code_paths – A list of local filesystem paths to Python file dependencies (or directories containing file dependencies). These files are prepended to the system path when the model is loaded.

  • registered_model_name – This argument may change or be removed in a future release without warning. If given, create a model version under registered_model_name, also creating a registered model if one with the given name does not exist.

  • signature –

    Model Signature mlflow.models.ModelSignature describes model input and output Schema. The model signature can be inferred from datasets with valid model input (e.g. the training dataset with target column omitted) and valid model output (e.g. model predictions generated on the training dataset), for example:

    from mlflow.models.signature import infer_signature

train = df.drop_column("target_label")
predictions = ...  # compute model predictions
signature = infer_signature(train, predictions)

  • input_example – Input example provides one or several instances of valid model input. The example can be used as a hint of what data to feed the model. The given example will be converted to a Pandas DataFrame and then serialized to json using the Pandas split-oriented format. Bytes are base64-encoded.

  • await_registration_for – Number of seconds to wait for the model version to finish being created and is in READY status. By default, the function waits for five minutes. Specify 0 or None to skip waiting.

  • pip_requirements – Either an iterable of pip requirement strings (e.g. ["pyod", "-r requirements.txt", "-c constraints.txt"]) or the string path to a pip requirements file on the local filesystem (e.g. "requirements.txt"). If provided, this describes the environment this model should be run in. If None, a default list of requirements is inferred by mlflow.models.infer_pip_requirements() from the current software environment. If the requirement inference fails, it falls back to using get_default_pip_requirements(). Both requirements and constraints are automatically parsed and written to requirements.txt and constraints.txt files, respectively, and stored as part of the model. Requirements are also written to the pip section of the model’s conda environment (conda.yaml) file.

  • extra_pip_requirements –

    Either an iterable of pip requirement strings (e.g. ["pandas", "-r requirements.txt", "-c constraints.txt"]) or the string path to a pip requirements file on the local filesystem (e.g. "requirements.txt"). If provided, this describes additional pip requirements that are appended to a default set of pip requirements generated automatically based on the user’s current software environment. Both requirements and constraints are automatically parsed and written to requirements.txt and constraints.txt files, respectively, and stored as part of the model. Requirements are also written to the pip section of the model’s conda environment (conda.yaml) file.

    Warning

    The following arguments can’t be specified at the same time:

    • conda_env

    • pip_requirements

    • extra_pip_requirements

    This example demonstrates how to specify pip requirements using pip_requirements and extra_pip_requirements.

  • serialization_format – The format in which to serialize the model. This should be one of the formats “pickle” or “cloudpickle”

Returns

A ModelInfo instance that contains the metadata of the logged model.

mlflavors.pyod.save_model(pyod_model, path, conda_env=None, code_paths=None, mlflow_model=None, signature=None, input_example=None, pip_requirements=None, extra_pip_requirements=None, serialization_format='pickle')[source]

Save an pyod model to a path on the local file system. Produces an MLflow Model containing the following flavors:

Parameters

  • pyod_model – Fitted pyod model object.

  • path – Local path where the model is to be saved.

  • conda_env –

    Either a dictionary representation of a Conda environment or the path to a conda environment yaml file. If provided, this describes the environment this model should be run in. At minimum, it should specify the dependencies contained in get_default_conda_env(). If None, a conda environment with pip requirements inferred by mlflow.models.infer_pip_requirements() is added to the model. If the requirement inference fails, it falls back to using get_default_pip_requirements(). pip requirements from conda_env are written to a pip requirements.txt file and the full conda environment is written to conda.yaml. The following is an example dictionary representation of a conda environment:

    {
    "name": "mlflow-env",
    "channels": ["conda-forge"],
    "dependencies": [
        "python=3.8.15",
        {
            "pip": [
                "pyod==x.y.z"
            ],
        },
    ],
}

  • code_paths – A list of local filesystem paths to Python file dependencies (or directories containing file dependencies). These files are prepended to the system path when the model is loaded.

  • mlflow_model – mlflow.models.Model configuration to which to add the python_function flavor.

  • signature –

    Model Signature mlflow.models.ModelSignature describes model input and output Schema. The model signature can be inferred from datasets with valid model input (e.g. the training dataset with target column omitted) and valid model output (e.g. model predictions generated on the training dataset), for example:

    from mlflow.models.signature import infer_signature

train = df.drop_column("target_label")
predictions = ...  # compute model predictions
signature = infer_signature(train, predictions)

  • input_example – Input example provides one or several instances of valid model input.The example can be used as a hint of what data to feed the model. The given example will be converted to a Pandas DataFrame and then serialized to json using the Pandas split-oriented format. Bytes are base64-encoded.

  • pip_requirements – Either an iterable of pip requirement strings (e.g. ["pyod", "-r requirements.txt", "-c constraints.txt"]) or the string path to a pip requirements file on the local filesystem (e.g. "requirements.txt"). If provided, this describes the environment this model should be run in. If None, a default list of requirements is inferred by mlflow.models.infer_pip_requirements() from the current software environment. If the requirement inference fails, it falls back to using get_default_pip_requirements(). Both requirements and constraints are automatically parsed and written to requirements.txt and constraints.txt files, respectively, and stored as part of the model. Requirements are also written to the pip section of the model’s conda environment (conda.yaml) file.

  • extra_pip_requirements –

    Either an iterable of pip requirement strings (e.g. ["pandas", "-r requirements.txt", "-c constraints.txt"]) or the string path to a pip requirements file on the local filesystem (e.g. "requirements.txt"). If provided, this describes additional pip requirements that are appended to a default set of pip requirements generated automatically based on the user’s current software environment. Both requirements and constraints are automatically parsed and written to requirements.txt and constraints.txt files, respectively, and stored as part of the model. Requirements are also written to the pip section of the model’s conda environment (conda.yaml) file.

    Warning

    The following arguments can’t be specified at the same time:

    • conda_env

    • pip_requirements

    • extra_pip_requirements

    This example demonstrates how to specify pip requirements using pip_requirements and extra_pip_requirements.

  • serialization_format – The format in which to serialize the model. This should be one of the formats “pickle” or “cloudpickle”

SDV

The mlflavors.sdv module provides an API for logging and loading sdv models. This module exports sdv models with the following flavors:

sdv (native) format

This is the main flavor that can be loaded back into sdv, which relies on pickle internally to serialize a model.

Note that pickle serialization requires using the same python environment (version) in whatever environment you’re going to use this model for inference to ensure that the model will load with appropriate version of pickle.

mlflow.pyfunc format

Produced for use by generic pyfunc-based deployment tools and batch inference.

Currently only the sample method is supported in the pyfunc flavor. Additional methods (e.g. sample_remaining_columns) could be added in a similar fashion.

The interface for utilizing an sdv model loaded as a pyfunc type for generating predictions uses a single-row Pandas DataFrame configuration argument. The following columns in this configuration Pandas DataFrame are supported:

Column

Type

Description

modality

str (required)
Specifies the sdv table modalities. The supported modalities are
single_table, multi_table, and sequential.
For more information, read the underlying library explanation:
https://docs.sdv.dev/sdv/.

num_rows

int (required)
An integer >0, describing the number of rows to sample.
Can only be provided in combination with modality single_table.

num_sequences

int (required)
An integer >0, describing the number of sequences to sample.
Can only be provided in combination with modality sequential.

batch_size

int (optional)
An integer >0, describing the number of rows to sample at a time.
Can only be provided in combination with modality single_table.
(Default: num_rows)

max_tries_per_batch

int (optional)
An integer >0, describing the number of sampling attempts to make per batch.
Can only be provided in combination with modality single_table.
(Default: 100)

output_file_path

str (optional)
A string describing a CSV filepath for writing the synthetic data.
Can only be provided in combination with modality single_table.
(Default: None)

scale

float (optional)
A float >0.0 that describes how much to scale the data by.
Can only be provided in combination with modality multi_table.
(Default: 1.0)

sequence_length

int (optional)
An integer >0 describing the length of each sequence.
Can only be provided in combination with modality sequential.
(Default: None)
mlflavors.sdv.get_default_conda_env(include_cloudpickle=False)[source]
Returns

The default Conda environment for MLflow Models produced by calls to save_model() and log_model().

mlflavors.sdv.get_default_pip_requirements(include_cloudpickle=False)[source]
Returns

A list of default pip requirements for MLflow Models produced by this flavor. Calls to save_model() and log_model() produce a pip environment that, at minimum, contains these requirements.

mlflavors.sdv.load_model(model_uri, dst_path=None)[source]

Load an sdv model from a local file or a run.

Parameters

  • model_uri –

    The location, in URI format, of the MLflow model, for example:

    • /Users/me/path/to/local/model

    • relative/path/to/local/model

    • s3://my_bucket/path/to/model

    • runs:/<mlflow_run_id>/run-relative/path/to/model

    • models:/<model_name>/<model_version>

    • models:/<model_name>/<stage>

    For more information about supported URI schemes, see Referencing Artifacts.

  • dst_path – The local filesystem path to which to download the model artifact. This directory must already exist. If unspecified, a local output path will be created.

Returns

An sdv model.

mlflavors.sdv.log_model(sdv_model, artifact_path, conda_env=None, code_paths=None, registered_model_name=None, signature=None, input_example=None, await_registration_for=300, pip_requirements=None, extra_pip_requirements=None, serialization_format='pickle', **kwargs)[source]

Log an sdv model as an MLflow artifact for the current run. Produces an MLflow Model containing the following flavors:

Parameters

  • sdv_model – Fitted sdv model object.

  • artifact_path – Run-relative artifact path to save the model instance to.

  • conda_env –

    Either a dictionary representation of a Conda environment or the path to a conda environment yaml file. If provided, this describes the environment this model should be run in. At minimum, it should specify the dependencies contained in get_default_conda_env(). If None, a conda environment with pip requirements inferred by mlflow.models.infer_pip_requirements() is added to the model. If the requirement inference fails, it falls back to using get_default_pip_requirements(). pip requirements from conda_env are written to a pip requirements.txt file and the full conda environment is written to conda.yaml. The following is an example dictionary representation of a conda environment:

    {
    "name": "mlflow-env",
    "channels": ["conda-forge"],
    "dependencies": [
        "python=3.8.15",
        {
            "pip": [
                "sdv==x.y.z"
            ],
        },
    ],
}

  • code_paths – A list of local filesystem paths to Python file dependencies (or directories containing file dependencies). These files are prepended to the system path when the model is loaded.

  • registered_model_name – This argument may change or be removed in a future release without warning. If given, create a model version under registered_model_name, also creating a registered model if one with the given name does not exist.

  • signature –

    Model Signature mlflow.models.ModelSignature describes model input and output Schema. The model signature can be inferred from datasets with valid model input (e.g. the training dataset with target column omitted) and valid model output (e.g. model predictions generated on the training dataset), for example:

    from mlflow.models.signature import infer_signature

train = df.drop_column("target_label")
predictions = ...  # compute model predictions
signature = infer_signature(train, predictions)

  • input_example – Input example provides one or several instances of valid model input. The example can be used as a hint of what data to feed the model. The given example will be converted to a Pandas DataFrame and then serialized to json using the Pandas split-oriented format. Bytes are base64-encoded.

  • await_registration_for – Number of seconds to wait for the model version to finish being created and is in READY status. By default, the function waits for five minutes. Specify 0 or None to skip waiting.

  • pip_requirements – Either an iterable of pip requirement strings (e.g. ["sdv", "-r requirements.txt", "-c constraints.txt"]) or the string path to a pip requirements file on the local filesystem (e.g. "requirements.txt"). If provided, this describes the environment this model should be run in. If None, a default list of requirements is inferred by mlflow.models.infer_pip_requirements() from the current software environment. If the requirement inference fails, it falls back to using get_default_pip_requirements(). Both requirements and constraints are automatically parsed and written to requirements.txt and constraints.txt files, respectively, and stored as part of the model. Requirements are also written to the pip section of the model’s conda environment (conda.yaml) file.

  • extra_pip_requirements –

    Either an iterable of pip requirement strings (e.g. ["pandas", "-r requirements.txt", "-c constraints.txt"]) or the string path to a pip requirements file on the local filesystem (e.g. "requirements.txt"). If provided, this describes additional pip requirements that are appended to a default set of pip requirements generated automatically based on the user’s current software environment. Both requirements and constraints are automatically parsed and written to requirements.txt and constraints.txt files, respectively, and stored as part of the model. Requirements are also written to the pip section of the model’s conda environment (conda.yaml) file.

    Warning

    The following arguments can’t be specified at the same time:

    • conda_env

    • pip_requirements

    • extra_pip_requirements

    This example demonstrates how to specify pip requirements using pip_requirements and extra_pip_requirements.

  • serialization_format – The format in which to serialize the model. This should be one of the formats “pickle” or “cloudpickle”

Returns

A ModelInfo instance that contains the metadata of the logged model.

mlflavors.sdv.save_model(sdv_model, path, conda_env=None, code_paths=None, mlflow_model=None, signature=None, input_example=None, pip_requirements=None, extra_pip_requirements=None, serialization_format='pickle')[source]

Save an sdv model to a path on the local file system. Produces an MLflow Model containing the following flavors:

Parameters

  • sdv_model – Fitted sdv model object.

  • path – Local path where the model is to be saved.

  • conda_env –

    Either a dictionary representation of a Conda environment or the path to a conda environment yaml file. If provided, this describes the environment this model should be run in. At minimum, it should specify the dependencies contained in get_default_conda_env(). If None, a conda environment with pip requirements inferred by mlflow.models.infer_pip_requirements() is added to the model. If the requirement inference fails, it falls back to using get_default_pip_requirements(). pip requirements from conda_env are written to a pip requirements.txt file and the full conda environment is written to conda.yaml. The following is an example dictionary representation of a conda environment:

    {
    "name": "mlflow-env",
    "channels": ["conda-forge"],
    "dependencies": [
        "python=3.8.15",
        {
            "pip": [
                "sdv==x.y.z"
            ],
        },
    ],
}

  • code_paths – A list of local filesystem paths to Python file dependencies (or directories containing file dependencies). These files are prepended to the system path when the model is loaded.

  • mlflow_model – mlflow.models.Model configuration to which to add the python_function flavor.

  • signature –

    Model Signature mlflow.models.ModelSignature describes model input and output Schema. The model signature can be inferred from datasets with valid model input (e.g. the training dataset with target column omitted) and valid model output (e.g. model predictions generated on the training dataset), for example:

    from mlflow.models.signature import infer_signature

train = df.drop_column("target_label")
predictions = ...  # compute model predictions
signature = infer_signature(train, predictions)

  • input_example – Input example provides one or several instances of valid model input.The example can be used as a hint of what data to feed the model. The given example will be converted to a Pandas DataFrame and then serialized to json using the Pandas split-oriented format. Bytes are base64-encoded.

  • pip_requirements – Either an iterable of pip requirement strings (e.g. ["sdv", "-r requirements.txt", "-c constraints.txt"]) or the string path to a pip requirements file on the local filesystem (e.g. "requirements.txt"). If provided, this describes the environment this model should be run in. If None, a default list of requirements is inferred by mlflow.models.infer_pip_requirements() from the current software environment. If the requirement inference fails, it falls back to using get_default_pip_requirements(). Both requirements and constraints are automatically parsed and written to requirements.txt and constraints.txt files, respectively, and stored as part of the model. Requirements are also written to the pip section of the model’s conda environment (conda.yaml) file.

  • extra_pip_requirements –

    Either an iterable of pip requirement strings (e.g. ["pandas", "-r requirements.txt", "-c constraints.txt"]) or the string path to a pip requirements file on the local filesystem (e.g. "requirements.txt"). If provided, this describes additional pip requirements that are appended to a default set of pip requirements generated automatically based on the user’s current software environment. Both requirements and constraints are automatically parsed and written to requirements.txt and constraints.txt files, respectively, and stored as part of the model. Requirements are also written to the pip section of the model’s conda environment (conda.yaml) file.

    Warning

    The following arguments can’t be specified at the same time:

    • conda_env

    • pip_requirements

    • extra_pip_requirements

    This example demonstrates how to specify pip requirements using pip_requirements and extra_pip_requirements.

  • serialization_format – The format in which to serialize the model. This should be one of the formats “pickle” or “cloudpickle”