def test_add_graph_symbol():
data = mx.sym.Variable('data')
conv = mx.sym.Convolution(data, kernel=(2, 2), num_filter=2)
nodes = _get_nodes_from_symbol(conv)
expected_nodes = [NodeDef(name='data', op='null'),
NodeDef(name='convolution0/convolution0_weight', op='null',
attr={'param': AttrValue(
s='{ kernel : (2, 2) , num_filter : 2 }'.encode(encoding='utf-8'))}),
NodeDef(name='convolution0/convolution0_bias', op='null',
attr={'param': AttrValue(
s='{ kernel : (2, 2) , num_filter : 2 }'.encode(encoding='utf-8'))}),
NodeDef(name='convolution0/convolution0', op='Convolution',
input=['data', 'convolution0/convolution0_weight', 'convolution0/convolution0_bias'],
attr={'param': AttrValue(
s='{ kernel : (2, 2) , num_filter : 2 }'.encode(encoding='utf-8'))})]
# check _get_nodes_from_symbol
for expected_node, node in zip(expected_nodes, nodes):
assert expected_node == node
# check _sym2pb
expected_graph = GraphDef(node=expected_nodes, versions=VersionDef(producer=100))
graph = _net2pb(conv)
assert expected_graph == graph
# check add_graph
with SummaryWriter(logdir=_LOGDIR) as sw:
sw.add_graph(conv)
check_event_file_and_remove_logdir()
def test_add_graph_gluon():
net = nn.HybridSequential()
with net.name_scope():
net.add(nn.Dense(128, activation='relu'))
net.hybridize()
net.initialize()
net.forward(mx.nd.ones(1,))
_, sym = net._cached_graph
nodes = _get_nodes_from_symbol(sym)
expected_nodes = [NodeDef(name='data', op='null'),
NodeDef(name='hybridsequential0_dense0_fwd/hybridsequential0_dense0_weight', op='null',
attr={'param': AttrValue(
s='{ __dtype__ : 0 , __lr_mult__ : 1.0 , __shape__ : '
'(128, 0) , __wd_mult__ : 1.0 }'.encode(encoding='utf-8'))}),
NodeDef(name='hybridsequential0_dense0_fwd/hybridsequential0_dense0_bias', op='null',
attr={'param': AttrValue(
s='{ __dtype__ : 0 , __init__ : zeros , __lr_mult__ : 1.0 , __shape__ : '
'(128,) , __wd_mult__ : 1.0 }'.encode(encoding='utf-8'))}),
NodeDef(name='hybridsequential0_dense0_fwd/hybridsequential0_dense0_fwd', op='FullyConnected',
input=['data', 'hybridsequential0_dense0_fwd/hybridsequential0_dense0_weight',
'hybridsequential0_dense0_fwd/hybridsequential0_dense0_bias'],
attr={'param': AttrValue(
s='{ flatten : True , no_bias : False , '
'num_hidden : 128 }'.encode(encoding='utf-8'))}),
NodeDef(name='hybridsequential0_dense0_relu_fwd/hybridsequential0_dense0_relu_fwd',
op='Activation', input=['hybridsequential0_dense0_fwd/hybridsequential0_dense0_fwd'],
attr={'param': AttrValue(
s='{ act_type : relu }'.encode(encoding='utf-8'))})
]
# check _get_nodes_from_symbol
for expected_node, node in zip(expected_nodes, nodes):
assert expected_node == node
# check _sym2pb
expected_graph = GraphDef(node=expected_nodes, versions=VersionDef(producer=100))
graph = _net2pb(net)
assert expected_graph == graph
# check add_graph
with SummaryWriter(logdir=_LOGDIR) as sw:
sw.add_graph(net)
check_event_file_and_remove_logdir()