def load_recipe(
domain: str,
sub_domain: str,
architecture: str,
sub_architecture: Union[str, None],
framework: str,
repo: str,
dataset: str,
training_scheme: Union[str, None],
sparse_name: str,
sparse_category: str,
sparse_target: Union[str, None],
recipe_type: Union[str, None] = None,
release_version: Union[str, None] = None,
override_folder_name: Union[str, None] = None,
override_parent_path: Union[str, None] = None,
force_token_refresh: bool = False,
) -> "Recipe":
"""
Obtains a Recipe from the model repo
:param domain: The domain of the model the object belongs to;
e.g. cv, nlp
:param sub_domain: The sub domain of the model the object belongs to;
e.g. classification, segmentation
:param architecture: The architecture of the model the object belongs to;
e.g. resnet_v1, mobilenet_v1
:param sub_architecture: The sub architecture (scaling factor) of the model
the object belongs to; e.g. 50, 101, 152
:param framework: The framework the model the object belongs to was trained on;
e.g. pytorch, tensorflow
:param repo: The source repo for the model the object belongs to;
e.g. sparseml, torchvision
:param dataset: The dataset the model the object belongs to was trained on;
e.g. imagenet, cifar10
:param training_scheme: The training scheme used on the model the object
belongs to if any; e.g. augmented
:param sparse_name: The name describing the sparsification of the model
the object belongs to, e.g. base, pruned, pruned_quant
:param sparse_category: The degree of sparsification of the model the object
belongs to; e.g. none, conservative (~100% baseline metric),
moderate (>=99% baseline metric), aggressive (<99% baseline metric)
:param sparse_target: The deployment target of sparsification of the model
the object belongs to; e.g. edge, deepsparse, deepsparse_throughput, gpu
:param recipe_type: The recipe type; e.g. original, transfer_learn
:param release_version: The sparsezoo release version for the model
:param override_folder_name: Override for the name of the folder to save
this file under
:param override_parent_path: Path to override the default save path
for where to save the parent folder for this file under
:param force_token_refresh: True to refresh the auth token, False otherwise
:return: The requested Recipe instance
"""
args = ModelArgs(
domain=domain,
sub_domain=sub_domain,
architecture=architecture,
sub_architecture=sub_architecture,
framework=framework,
repo=repo,
dataset=dataset,
training_scheme=training_scheme,
sparse_name=sparse_name,
sparse_category=sparse_category,
sparse_target=sparse_target,
release_version=release_version,
)
_LOGGER.debug("load_recipe: loading recipe")
return Recipe.load_recipe_from_stub(
args,
recipe_type=recipe_type,
override_folder_name=override_folder_name,
override_parent_path=override_parent_path,
force_token_refresh=force_token_refresh,
)
def search_models(
domain: str,
sub_domain: str,
architecture: Union[str, None] = None,
sub_architecture: Union[str, None] = None,
framework: Union[str, None] = None,
repo: Union[str, None] = None,
dataset: Union[str, None] = None,
training_scheme: Union[str, None] = None,
sparse_name: Union[str, None] = None,
sparse_category: Union[str, None] = None,
sparse_target: Union[str, None] = None,
release_version: Union[str, None] = None,
page: int = 1,
page_length: int = 20,
override_folder_name: Union[str, None] = None,
override_parent_path: Union[str, None] = None,
force_token_refresh: bool = False,
) -> List["Model"]:
"""
Obtains a list of Models matching the search parameters
:param domain: The domain of the model the object belongs to;
e.g. cv, nlp
:param sub_domain: The sub domain of the model the object belongs to;
e.g. classification, segmentation
:param architecture: The architecture of the model the object belongs to;
e.g. resnet_v1, mobilenet_v1
:param sub_architecture: The sub architecture (scaling factor) of the model
the object belongs to; e.g. 50, 101, 152
:param framework: The framework the model the object belongs to was trained on;
e.g. pytorch, tensorflow
:param repo: The source repo for the model the object belongs to;
e.g. sparseml, torchvision
:param dataset: The dataset the model the object belongs to was trained on;
e.g. imagenet, cifar10
:param training_scheme: The training scheme used on the model the object
belongs to if any; e.g. augmented
:param sparse_name: The name describing the sparsification of the model
the object belongs to, e.g. base, pruned, pruned_quant
:param sparse_category: The degree of sparsification of the model the object
belongs to; e.g. none, conservative (~100% baseline metric),
moderate (>=99% baseline metric), aggressive (<99% baseline metric)
:param sparse_target: The deployment target of sparsification of the model
the object belongs to; e.g. edge, deepsparse, deepsparse_throughput, gpu
:param release_version: The sparsezoo release version for the model
:param page: the page of values to get
:param page_length: the page length of values to get
:param override_folder_name: Override for the name of the folder to save
this file under
:param override_parent_path: Path to override the default save path
for where to save the parent folder for this file under
:param force_token_refresh: True to refresh the auth token, False otherwise
:return: The requested Model instance
"""
args = ModelArgs(
domain=domain,
sub_domain=sub_domain,
architecture=architecture,
sub_architecture=sub_architecture,
framework=framework,
repo=repo,
dataset=dataset,
training_scheme=training_scheme,
sparse_name=sparse_name,
sparse_category=sparse_category,
sparse_target=sparse_target,
release_version=release_version,
)
_LOGGER.debug(
f"search_models: searching models with args {args.model_url_args}")
response_json = search_model_get_request(
args=args,
page=page,
page_length=page_length,
force_token_refresh=force_token_refresh,
)
return [
Model(
**model,
override_folder_name=override_folder_name,
override_parent_path=override_parent_path,
) for model in response_json["models"]
]