def __init__(self, in_channels, out_channels, has_bias, has_bn):
super(LastQuantLayer, self).__init__()
self.dense_inner = nn.DenseQuant(in_channels,
out_channels,
has_bias=has_bias,
quant_config=quant_config,
quant_dtype=QuantDtype.INT8)
self.fake_quant_act = nn.FakeQuantWithMinMaxObserver(
min_init=-16,
max_init=16,
ema=True,
quant_dtype=QuantDtype.INT8,
per_channel=False,
symmetric=True,
narrow_range=True,
mode="LEARNED_SCALE")
def __init__(self, block, layer_nums, in_channels, out_channels, strides,
num_classes):
super(ResNet, self).__init__()
if not len(layer_nums) == len(in_channels) == len(out_channels) == 4:
raise ValueError(
"the length of layer_num, in_channels, out_channels list must be 4!"
)
self.conv1 = ConvBNReLU(3, 64, kernel_size=7, stride=2)
self.maxpool = nn.MaxPool2d(kernel_size=3, stride=2, pad_mode="same")
self.layer1 = self._make_layer(block,
layer_nums[0],
in_channel=in_channels[0],
out_channel=out_channels[0],
stride=strides[0])
self.layer2 = self._make_layer(block,
layer_nums[1],
in_channel=in_channels[1],
out_channel=out_channels[1],
stride=strides[1])
self.layer3 = self._make_layer(block,
layer_nums[2],
in_channel=in_channels[2],
out_channel=out_channels[2],
stride=strides[2])
self.layer4 = self._make_layer(block,
layer_nums[3],
in_channel=in_channels[3],
out_channel=out_channels[3],
stride=strides[3])
self.mean = P.ReduceMean(keep_dims=True)
self.flatten = nn.Flatten()
self.end_point = nn.DenseQuant(out_channels[3],
num_classes,
has_bias=True,
quant_config=_quant_config)
self.output_fake = nn.FakeQuantWithMinMaxObserver(ema=True,
ema_decay=_ema_decay)
# init weights
self._initialize_weights()