def __init__(self, args, classes, layers, nodes, channels, keep_prob, drop_path_keep_prob, use_aux_head, steps, arch):
super(NASNetworkImageNet, self).__init__()
self.args = args
self.search_space = args.search_space
self.classes = classes
self.layers = layers
self.nodes = nodes
self.channels = channels
self.keep_prob = keep_prob
self.drop_path_keep_prob = drop_path_keep_prob
self.use_aux_head = use_aux_head
self.steps = steps
arch = list(map(int, arch.strip().split()))
self.conv_arch = arch[:4 * self.nodes]
self.reduc_arch = arch[4 * self.nodes:]
self.pool_layers = [self.layers, 2 * self.layers + 1]
self.layers = self.layers * 3
if self.use_aux_head:
self.aux_head_index = self.pool_layers[-1]
channels = self.channels
self.stem0 = nn.Sequential(
nn.Conv2d(3, channels // 2, 3, stride=2, padding=1, bias=False),
nn.BatchNorm2d(channels // 2),
nn.ReLU(inplace=True),
nn.Conv2d(channels // 2, channels, 3, stride=2, padding=1, bias=False),
nn.BatchNorm2d(channels),
)
self.stem1 = nn.Sequential(
nn.ReLU(inplace=True),
nn.Conv2d(channels, channels, 3, stride=2, padding=1, bias=False),
nn.BatchNorm2d(channels),
)
outs = [[56, 56, channels], [28, 28, channels]]
channels = self.channels
self.cells = nn.ModuleList()
for i in range(self.layers + 2):
if i not in self.pool_layers:
cell = Cell(self.search_space, self.conv_arch, outs, channels, False, i, self.layers + 2, self.steps,
self.drop_path_keep_prob)
else:
channels *= 2
cell = Cell(self.search_space, self.reduc_arch, outs, channels, True, i, self.layers + 2, self.steps,
self.drop_path_keep_prob)
self.cells.append(cell)
outs = [outs[-1], cell.out_shape]
if self.use_aux_head and i == self.aux_head_index:
self.auxiliary_head = AuxHeadImageNet(outs[-1][-1], classes)
self.global_pooling = nn.AdaptiveAvgPool2d(1)
self.dropout = nn.Dropout(1 - self.keep_prob)
self.classifier = nn.Linear(outs[-1][-1], classes)
self.init_parameters()
def __init__(self, classes, layers, nodes, channels, keep_prob,
drop_path_keep_prob, use_aux_head, steps):
super(NASWSNetworkImageNet, self).__init__()
self.classes = classes
self.layers = layers
self.nodes = nodes
self.channels = channels
self.keep_prob = keep_prob
self.drop_path_keep_prob = drop_path_keep_prob
self.use_aux_head = use_aux_head
self.steps = steps
self.pool_layers = [self.layers, 2 * self.layers + 1]
self.total_layers = self.layers * 3 + 2
if self.use_aux_head:
self.aux_head_index = self.pool_layers[-1]
channels = self.channels
self.stem0 = nn.Sequential(
nn.Conv2d(3, channels // 2, 3, stride=2, padding=1, bias=False),
nn.BatchNorm2d(channels // 2),
nn.ReLU(inplace=False),
nn.Conv2d(channels // 2,
channels,
3,
stride=2,
padding=1,
bias=False),
nn.BatchNorm2d(channels // 2),
)
self.stem1 = nn.Sequential(
nn.ReLU(inplace=False),
nn.Conv2d(channels, channels, 3, stride=2, padding=1, bias=False),
nn.BatchNorm2d(channels),
)
outs = [[112, 112, channels // 2], [56, 56, channels]]
self.cells = nn.ModuleList()
for i in range(self.total_layers):
if i not in self.pool_layers:
cell = Cell(outs, self.nodes, channels, False, i,
self.total_layers, self.steps,
self.drop_path_keep_prob)
outs = [outs[-1], cell.out_shape]
else:
channels *= 2
cell = Cell(outs, self.nodes, channels, True, i,
self.total_layers, self.steps,
self.drop_path_keep_prob)
outs = [outs[-1], cell.out_shape]
self.cells.append(cell)
if self.use_aux_head and i == self.aux_head_index:
self.auxiliary_head = AuxHeadImageNet(outs[-1][-1], classes)
self.global_pooling = nn.AdaptiveAvgPool2d(1)
self.dropout = nn.Dropout(1 - self.keep_prob)
self.classifier = nn.Linear(outs[-1][-1], classes)
self.init_parameters()
def __init__(self, args, classes, layers, nodes, channels, keep_prob,
drop_path_keep_prob, use_aux_head, steps, arch):
super(NASNetworkImageNet, self).__init__()
self.args = args
self.classes = classes
self.layers = layers
self.nodes = nodes
self.channels = channels
self.keep_prob = keep_prob
self.drop_path_keep_prob = drop_path_keep_prob
self.use_aux_head = use_aux_head
self.steps = steps
try:
arch = eval(arch)
except:
pass
arch = arch[0] + arch[1]
self.conv_arch = arch[:4 * self.nodes]
self.reduc_arch = arch[4 * self.nodes:]
self.pool_layers = [self.layers, 2 * self.layers + 1]
self.layers = self.layers * 3
self.multi_adds = 0
if self.use_aux_head:
self.aux_head_index = self.pool_layers[-1]
channels = self.channels
self.stem0 = nn.Sequential(
nn.Conv2d(3, channels // 2, 3, stride=2, padding=1, bias=False),
nn.BatchNorm2d(channels // 2),
nn.ReLU(inplace=True),
nn.Conv2d(channels // 2,
channels,
3,
stride=2,
padding=1,
bias=False),
nn.BatchNorm2d(channels),
)
self.multi_adds += 3 * 3 * 3 * channels // 2 * 112 * 112 + 3 * 3 * channels // 2 * channels * 56 * 56
self.stem1 = nn.Sequential(
nn.ReLU(inplace=True),
nn.Conv2d(channels, channels, 3, stride=2, padding=1, bias=False),
nn.BatchNorm2d(channels),
)
self.multi_adds += 3 * 3 * channels * channels * 56 * 56
outs = [[224, 224, channels], [112, 112, channels]]
print(outs)
channels = self.channels
self.cells = nn.ModuleList()
for i in range(self.layers + 2):
if i not in self.pool_layers:
cell = Cell(self.conv_arch, outs, channels, False, i,
self.layers + 2, self.steps,
self.drop_path_keep_prob)
else:
channels *= 2
cell = Cell(self.reduc_arch, outs, channels, True, i,
self.layers + 2, self.steps,
self.drop_path_keep_prob)
self.multi_adds += cell.multi_adds
self.cells.append(cell)
outs = [outs[-1], cell.out_shape]
print(outs)
if self.use_aux_head and i == self.aux_head_index:
self.auxiliary_head = AuxHeadImageNet(outs[-1][-1], classes)
self.global_pooling = nn.AdaptiveAvgPool2d(1)
self.dropout = nn.Dropout(1 - self.keep_prob)
self.classifier = nn.Linear(outs[-1][-1], classes)
self.init_parameters()