from __future__ import annotations
import logging
import typing as t
from types import ModuleType
from typing import TYPE_CHECKING
import numpy as np
import bentoml
from bentoml import Tag
from bentoml.exceptions import InvalidArgument
from bentoml.exceptions import MissingDependencyException
from bentoml.exceptions import NotFound
from ..models.model import ModelContext
from ..utils.pkg import get_pkg_version
if TYPE_CHECKING:
from bentoml.types import ModelSignature
from bentoml.types import ModelSignatureDict
try:
import lightgbm as lgb
except ImportError: # pragma: no cover
raise MissingDependencyException(
"lightgbm is required in order to use module 'bentoml.lightgbm', install lightgbm with 'pip install lightgbm'. For more information, refer to https://github.com/microsoft/LightGBM/tree/master/python-package"
)
MODULE_NAME = "bentoml.lightgbm"
MODEL_FILENAME = "saved_model.ubj"
API_VERSION = "v1"
logger = logging.getLogger(__name__)
[docs]def get(tag_like: str | Tag) -> bentoml.Model:
"""
Get the BentoML model with the given tag.
Args:
tag_like (``str`` ``|`` :obj:`~bentoml.Tag`):
The tag of the model to retrieve from the model store.
Returns:
:obj:`~bentoml.Model`: A BentoML :obj:`~bentoml.Model` with the matching tag.
Example:
.. code-block:: python
import bentoml
# target model must be from the BentoML model store
model = bentoml.lightgbm.get("my_lightgbm_model:latest")
"""
model = bentoml.models.get(tag_like)
if model.info.module not in (MODULE_NAME, __name__):
raise NotFound(
f"Model {model.tag} was saved with module {model.info.module}, not loading with {MODULE_NAME}."
)
return model
[docs]def load_model(bento_model: str | Tag | bentoml.Model) -> lgb.basic.Booster: # type: ignore (incomplete ligthgbm type stubs)
"""
Load the LightGBM model with the given tag from the local BentoML model store.
Args:
bento_model (``str`` ``|`` :obj:`~bentoml.Tag` ``|`` :obj:`~bentoml.Model`):
Either the tag of the model to get from the store, or a BentoML `~bentoml.Model`
instance to load the model from.
Returns:
:obj:`~lightgbm.basic.Booster`: The LightGBM model loaded from the model store or BentoML :obj:`~bentoml.Model`.
Example:
.. code-block:: python
import bentoml
gbm = bentoml.lightgbm.load("my_lightgbm_model:latest")
""" # noqa
if not isinstance(bento_model, bentoml.Model):
bento_model = get(bento_model)
assert isinstance(bento_model, bentoml.Model)
if bento_model.info.module not in (MODULE_NAME, __name__):
raise NotFound(
f"Model {bento_model.tag} was saved with module {bento_model.info.module}, not loading with {MODULE_NAME}."
)
model_file = bento_model.path_of(MODEL_FILENAME)
booster = lgb.basic.Booster(model_file=model_file) # type: ignore (incomplete ligthgbm type stubs)
return booster # type: ignore
[docs]def save_model(
name: Tag | str,
model: lgb.basic.Booster, # type: ignore (incomplete ligthgbm type stubs)
*,
signatures: dict[str, ModelSignatureDict] | None = None,
labels: dict[str, str] | None = None,
custom_objects: dict[str, t.Any] | None = None,
external_modules: t.List[ModuleType] | None = None,
metadata: dict[str, t.Any] | None = None,
) -> bentoml.Model:
"""
Save a LightGBM model instance to the BentoML model store.
Args:
name (``str``):
The name to give to the model in the BentoML store. This must be a valid
:obj:`~bentoml.Tag` name.
model (:obj:`~lgb.basic.Booster`):
The LightGBM model (booster) to be saved.
signatures (``dict[str, ModelSignatureDict]``, optional):
Signatures of predict methods to be used. If not provided, the signatures default to
``{"predict": {"batchable": False}}``. See :obj:`~bentoml.types.ModelSignature` for more
details.
labels (``dict[str, str]``, optional):
A default set of management labels to be associated with the model. An example is
``{"training-set": "data-1"}``.
custom_objects (``dict[str, Any]``, optional):
Custom objects to be saved with the model. An example is
``{"my-normalizer": normalizer}``.
Custom objects are currently serialized with cloudpickle, but this implementation is
subject to change.
external_modules (:code:`List[ModuleType]`, `optional`, default to :code:`None`):
user-defined additional python modules to be saved alongside the model or custom objects,
e.g. a tokenizer module, preprocessor module, model configuration module
metadata (``dict[str, Any]``, optional):
Metadata to be associated with the model. An example is ``{"max_depth": 2}``.
Metadata is intended for display in model management UI and therefore must be a default
Python type, such as ``str`` or ``int``.
Returns:
:obj:`~bentoml.Tag`: A :obj:`tag` with a format `name:version` where `name` is the
user-defined model's name, and a generated `version` by BentoML.
Example:
.. code-block:: python
import bentoml
import lightgbm as lgb
import pandas as pd
# load a dataset
df_train = pd.read_csv("regression.train", header=None, sep="\t")
df_test = pd.read_csv("regression.test", header=None, sep="\t")
y_train = df_train[0]
y_test = df_test[0]
X_train = df_train.drop(0, axis=1)
X_test = df_test.drop(0, axis=1)
# create dataset for lightgbm
lgb_train = lgb.Dataset(X_train, y_train)
lgb_eval = lgb.Dataset(X_test, y_test, reference=lgb_train)
# specify your configurations as a dict
params = {
"boosting_type": "gbdt",
"objective": "regression",
"metric": {"l2", "l1"},
"num_leaves": 31,
"learning_rate": 0.05,
}
# train
gbm = lgb.train(
params, lgb_train, num_boost_round=20, valid_sets=lgb_eval
)
# save the booster to BentoML modelstore:
bento_model = bentoml.lightgbm.save_model("my_lightgbm_model", gbm, booster_params=params)
"""
# Ensure that `model` is actually the Booster object, and not for example one of the scikit-learn wrapper objects.
if not isinstance(model, lgb.basic.Booster): # type: ignore (incomplete ligthgbm type stubs)
try:
# Work around a LightGBM issue (https://github.com/microsoft/LightGBM/issues/3014)
# 'model.booster_' chjecks that the model has been fitted and will error otherwise.
if not hasattr(model, "fitted_"): # type: ignore (incomplete ligthgbm type stubs)
model.fitted_ = True
model = model.booster_ # type: ignore (incomplete ligthgbm type stubs)
except AttributeError as e:
logger.error('Unable to obtain a "lightgbm.basic.Booster" from %s.', model)
raise e
if not isinstance(model, lgb.basic.Booster): # type: ignore (incomplete ligthgbm type stubs)
raise TypeError(f"Given model ({model}) is not a lightgbm.basic.Booster.")
context: ModelContext = ModelContext(
framework_name="lightgbm",
framework_versions={"lightgbm": get_pkg_version("lightgbm")},
)
if signatures is None:
signatures = {
"predict": {"batchable": False},
}
logger.info(
'Using the default model signature for LightGBM (%s) for model "%s".',
signatures,
name,
)
with bentoml.models._create( # type: ignore
name,
module=MODULE_NAME,
api_version=API_VERSION,
signatures=signatures,
labels=labels,
custom_objects=custom_objects,
external_modules=external_modules,
metadata=metadata,
context=context,
) as bento_model:
model.save_model(bento_model.path_of(MODEL_FILENAME))
return bento_model
def get_runnable(bento_model: bentoml.Model) -> t.Type[bentoml.Runnable]:
"""
Private API: use :obj:`~bentoml.Model.to_runnable` instead.
"""
class LightGBMRunnable(bentoml.Runnable):
# LightGBM only supports GPU during training, not for inference.
SUPPORTED_RESOURCES = ("cpu",)
SUPPORTS_CPU_MULTI_THREADING = True
def __init__(self):
super().__init__()
self.model = load_model(bento_model)
self.predict_fns: dict[str, t.Callable[..., t.Any]] = {}
for method_name in bento_model.info.signatures:
try:
self.predict_fns[method_name] = getattr(self.model, method_name) # type: ignore (incomplete ligthgbm type stubs)
except AttributeError:
raise InvalidArgument(
f"No method with name {method_name} found for LightGBM model of type {self.model.__class__}"
)
def add_runnable_method(method_name: str, options: ModelSignature):
def _run(self: LightGBMRunnable, input_data: t.Any) -> t.Any:
res = self.predict_fns[method_name](input_data)
return np.asarray(res) # type: ignore (unknown ndarray types)
LightGBMRunnable.add_method(
_run,
name=method_name,
batchable=options.batchable,
batch_dim=options.batch_dim,
input_spec=options.input_spec,
output_spec=options.output_spec,
)
for method_name, options in bento_model.info.signatures.items():
add_runnable_method(method_name, options)
return LightGBMRunnable