auto_decomposition

Applicability

Product	Supported
Atlas 350 Accelerator Card	x
Atlas A3 training product/Atlas A3 inference product	x
Atlas A2 training product/Atlas A2 inference product	x
Atlas 200I/500 A2 inference product	√
Atlas inference product	√
Atlas training product	√

Description

Generates a decomposed model file and its weight file from a given Caffe model file (.prototxt) and its weight file (.caffemodel).

Prototype

auto_decomposition(model_file,weights_file,new_model_file,new_weights_file)

Parameters

Parameter	Input/Output	Description
model_file	Input	Definition file (.prototxt) of the Caffe model. A string.
weights_file	Input	Weight file (.caffemodel) of the trained Caffe model. A string.
new_model_file	Input	Caffe model definition file (.prototxt) after tensor decomposition, for example, xx_tensor_decomposition.prototxt. A string.
new_weights_file	Input	Caffe model weight file (.caffemodel) after tensor decomposition, for example, xx_tensor_decomposition.caffemodel. A string.

Returns

None

Restrictions

Ensure that the input Caffe .prototxt file matches the input .caffemodel file.
Pass the path of the original model and the output path to the tensor decomposition API. This API automatically decomposes the convolutional layers that meet the decomposition conditions. For details about the decomposition conditions, see Restrictions.

Example

from amct_caffe.tensor_decompose import auto_decomposition
auto_decomposition(model_file='ResNet-50-deploy.prototxt',
                  weights_file='ResNet-50-weights.caffemodel',
                  new_model_file='ResNet-50-deploy_tensor_decomposition.prototxt',
                  new_weights_file='ResNet-50-deploy_tensor_decomposition.caffemodel')

Flush files:

Model definition file (.prototxt) after tensor decompression.
Model weight file (.caffemodel) after tensor decompression.

Parent topic: Tensor Decomposition API