onnx.hub#

ModelInfo#

class onnx.hub.ModelInfo(raw_model_info: dict[str, Any])[source]#

A class to represent a model’s property and metadata in the ONNX Hub. It extracts model name, path, sha, tags, etc. from the passed in raw_model_info dict.

model#: The name of the model.

model_path#: The path to the model, relative to the model zoo (https://github.com/onnx/models/) repo root.

metadata#: Additional metadata of the model, such as the size of the model, IO ports, etc.

model_sha#: The SHA256 digest of the model file.

tags#: A set of tags associated with the model.

opset#: The opset version of the model.

download_model_with_test_data#

onnx.hub.download_model_with_test_data(model: str, repo: str = 'onnx/models:main', opset: int | None = None, force_reload: bool = False, silent: bool = False) → str | None[source]#

Downloads a model along with test data by name from the onnx model hub and returns the directory to which the files have been extracted.

Parameters:

model – The name of the onnx model in the manifest. This field is case-sensitive
repo – The location of the model repo in format “user/repo[:branch]”. If no branch is found will default to “main”
opset – The opset of the model to download. The default of None automatically chooses the largest opset
force_reload – Whether to force the model to re-download even if its already found in the cache
silent – Whether to suppress the warning message if the repo is not trusted.

Returns:

str or None

get_model_info#

onnx.hub.get_model_info(model: str, repo: str = 'onnx/models:main', opset: int | None = None) → ModelInfo[source]#

Gets the model info matching the given name and opset.

Parameters:

model – The name of the onnx model in the manifest. This field is case-sensitive
repo – The location of the model repo in format “user/repo[:branch]”. If no branch is found will default to “main”
opset – The opset of the model to get. The default of None will return the model with largest opset.

Returns:

ModelInfo.

list_models#

onnx.hub.list_models(repo: str = 'onnx/models:main', model: str | None = None, tags: list[str] | None = None) → list[ModelInfo][source]#

Gets the list of model info consistent with a given name and tags

Parameters:

repo – The location of the model repo in format “user/repo[:branch]”. If no branch is found will default to “main”
model – The name of the model to search for. If None, will return all models with matching tags.
tags – A list of tags to filter models by. If None, will return all models with matching name.

Returns:

``ModelInfo``s.

load#

onnx.hub.load(model: str, repo: str = 'onnx/models:main', opset: int | None = None, force_reload: bool = False, silent: bool = False) → ModelProto | None[source]#

Downloads a model by name from the onnx model hub.

Parameters:

model – The name of the onnx model in the manifest. This field is case-sensitive
repo – The location of the model repo in format “user/repo[:branch]”. If no branch is found will default to “main”
opset – The opset of the model to download. The default of None automatically chooses the largest opset
force_reload – Whether to force the model to re-download even if its already found in the cache
silent – Whether to suppress the warning message if the repo is not trusted.

Returns:

ModelProto or None

load_composite_model#

onnx.hub.load_composite_model(network_model: str, preprocessing_model: str, network_repo: str = 'onnx/models:main', preprocessing_repo: str = 'onnx/models:main', opset: int | None = None, force_reload: bool = False, silent: bool = False) → ModelProto | None[source]#

Builds a composite model including data preprocessing by downloading a network and a preprocessing model and combine it into a single model

Parameters:

network_model – The name of the onnx model in the manifest.
preprocessing_model – The name of the preprocessing model.
network_repo – The location of the model repo in format “user/repo[:branch]”. If no branch is found will default to “main”
preprocessing_repo – The location of the proprocessing model repo in format “user/repo[:branch]”. If no branch is found will default to “main”
opset – The opset of the model to download. The default of None automatically chooses the largest opset
force_reload – Whether to force the model to re-download even if its already found in the cache
silent – Whether to suppress the warning message if the repo is not trusted.

Returns:

ModelProto or None