MCPcopy Create free account
hub / github.com/HanHan55/Graph2plan / build_cnn

Function build_cnn

PostProcess/g2p/layers.py:149–233  ·  view source on GitHub ↗

Build a CNN from an architecture string, which is a list of layer specification strings. The overall architecture can be given as a list or as a comma-separated string. All convolutions *except for the first* are preceeded by normalization and nonlinearity. All other layer

(arch, normalization='batch', activation='leakyrelu', padding='same',
              pooling='max', init='default')

Source from the content-addressed store, hash-verified

source not stored for this graph (policy: none)

Callers 1

__init__Method · 0.70

Calls 7

get_normalization_2dFunction · 0.70
get_activationFunction · 0.70
_get_paddingFunction · 0.70
_init_convFunction · 0.70
ResidualBlockClass · 0.70
InterpolateClass · 0.70
FlattenClass · 0.70

Tested by

no test coverage detected