mindspore.hub

mindspore.hub.load_weights(network, network_name=None, force_reload=True, **kwargs)

Load a model from mindspore, with pretrained weights.

Parameters

• network (Cell) – Cell network.

• network_name (string, optional) – Cell network name get from network. Default: None.

• force_reload (bool, optional) – Whether to force a fresh download unconditionally. Default: False.

• kwargs (dict, optional) –

  The corresponding kwargs for download for model.

  • device_target (str, optional): Runtime device target. Default: ‘ascend’.

  • dataset (str, optional): Dataset to train the network. Default: ‘cifar10’.

  • version (str, optional): MindSpore version to save the checkpoint. Default: Latest version.

Example

>>> hub.load(network, network_name='lenet',
             **{'device_target': 'ascend', 'dataset':'mnist', 'version': '0.5.0'})