dnnsvg

NNSVG inspired network structure drawer for deep learning

How to use

Sample code

dnnsvg has intuitive interface which enables building svgs for common neural network architecture

svg = SVGBuilder(height=height, width=width) \
    .add_layer(Convolution2D(in_channels=None, out_channels=96, ksize=11, stride=4)) \
    .add_layer(MaxPooling(ksize=3, stride=2)) \
    .add_layer(Convolution2D(in_channels=96, out_channels=256, ksize=5, stride=1, pad=2)) \
    .add_layer(MaxPooling(ksize=3, stride=2)) \
    .add_layer(Convolution2D(in_channels=256, out_channels=384, ksize=3, stride=1, pad=1)) \
    .add_layer(Convolution2D(in_channels=384, out_channels=384, ksize=3, stride=1, pad=1)) \
    .add_layer(Convolution2D(in_channels=384, out_channels=256, ksize=3, stride=1, pad=1)) \
    .add_layer(MaxPooling(ksize=3, stride=2)) \
    .add_layer(Reshape(output_shape=(1, 9216))) \
    .add_layer(FullyConnected(output_shape=(1, 4096))) \
    .add_layer(FullyConnected(output_shape=(1, 4096))) \
    .add_layer(FullyConnected(output_shape=(1, 1000))) \
    .build(input_tensor)

Sample svg

Alexnet svg sample

To install the package

python setup.py install

or if you prefer using pip

pip install .

When developing the package

It is recommended to use develop instead of install option to reflect changes in the directory

python setup.py develop

or if you prefer using pip

pip install -e .