module `tools.zoo`#

Short summary#

module mlprodict.tools.zoo

Tools to test models from the ONNX Zoo.

Functions#

function	truncated documentation
`_download_url`
`download_model_data`	Downloads a model and returns a link to the local ONNX file and data which can be used as inputs.
`load_data`	Restores protobuf data stored in a folder.
`short_list_zoo_models`	Returns a short list from ONNX Zoo.
`verify_model`	Verifies a model.

Documentation#

Tools to test models from the ONNX Zoo.

New in version 0.6.

source on GitHub

mlprodict.tools.zoo._download_url(url, output_path, name, verbose=False)#

mlprodict.tools.zoo.download_model_data(name, model=None, cache=None, verbose=False)#

Downloads a model and returns a link to the local ONNX file and data which can be used as inputs.

Parameters:

name – model name (see short_list_zoo_models)
model – url or empty to get the default value returned by short_list_zoo_models)
cache – folder to cache the downloaded data
verbose – display a progress bar

Returns:

local onnx file, input data

source on GitHub

mlprodict.tools.zoo.load_data(folder)#

Restores protobuf data stored in a folder.

Parameters:: folder – folder
Returns:: dictionary

source on GitHub

mlprodict.tools.zoo.short_list_zoo_models()#

Returns a short list from ONNX Zoo.

Returns:: list of dictionaries.

<<<

import pprint
from mlprodict.tools.zoo import short_list_zoo_models
pprint.pprint(short_list_zoo_models())

>>>

    [{'model': 'https://github.com/onnx/models/raw/main/vision/classification/mobilenet/model/mobilenetv2-7.tar.gz',
      'name': 'mobilenet'},
     {'model': 'https://github.com/onnx/models/raw/main/vision/classification/resnet/model/resnet18-v1-7.tar.gz',
      'name': 'resnet18'},
     {'folder': 'squeezenet',
      'model': 'https://github.com/onnx/models/raw/main/vision/classification/squeezenet/model/squeezenet1.0-9.tar.gz',
      'name': 'squeezenet'},
     {'folder': 'densenet121',
      'model': 'https://github.com/onnx/models/raw/main/vision/classification/densenet-121/model/densenet-9.tar.gz',
      'name': 'densenet121'},
     {'model': 'https://github.com/onnx/models/raw/main/vision/classification/inception_and_googlenet/inception_v2/model/inception-v2-9.tar.gz',
      'name': 'inception2'},
     {'model': 'https://github.com/onnx/models/raw/main/vision/classification/shufflenet/model/shufflenet-9.tar.gz',
      'name': 'shufflenet'},
     {'model': 'https://github.com/onnx/models/raw/main/vision/classification/efficientnet-lite4/model/efficientnet-lite4-11.tar.gz',
      'name': 'efficientnet-lite4'}]

source on GitHub

mlprodict.tools.zoo.verify_model(onnx_file, examples, runtime=None, abs_tol=0.0005, verbose=0, fLOG=None)#

Verifies a model.

Parameters:

onnx_file – ONNX file
examples – list of examples to verify
runtime – a runtime to use
abs_tol – error tolerance when checking the output
verbose – verbosity level for for runtime other than ‘onnxruntime’
fLOG – logging function when verbose > 0

Returns:

errors for every sample

source on GitHub

module tools.zoo#

Short summary#

Functions#

Documentation#

module `tools.zoo`#