代码：
import torch
import torch.nn as nn
import torch.nn.functional as F
class inceptionModuleV1(nn.Module):
def __init__(self,inputs, outputs1 , outputs2 ,outputs3 ,outputs4 ,outputs5 ,outputs6):
super(inceptionModuleV1,self).__init__()
self.blocks1 = nn.Sequential(nn.Conv2d(in_channels= inputs, out_channels= outputs1 ,kernel_size=1 ,stride= 1 ))
self.blocks2 = nn.Sequential(nn.Conv2d(in_channels= inputs ,out_channels= outputs2 ,kernel_size= 1 ,stride = 1 ),
nn.ReLU(inplace= True),
nn.Conv2d(in_channels= outputs2 ,out_channels= outputs3 ,kernel_size=3 ,stride=1,padding= 1 ),
nn.ReLU(inplace=True)
)
self.blocks3 = nn.Sequential(nn.Conv2d(in_channels= inputs ,out_channels= outputs4 ,kernel_size= 1 ,stride = 1),
nn.ReLU(inplace= True),
nn.Conv2d(in_channels= outputs4 ,out_channels= outputs5 ,kernel_size=5 ,stride=1,padding=2 ),
nn.ReLU(inplace=True)
)
self.blocks4= nn.Sequential(nn.MaxPool2d(kernel_size=3 ,stride=1,padding= 1 ),
nn.Conv2d(in_channels= inputs ,out_channels= outputs6 ,kernel_size=1 ,stride=1),
nn.ReLU(inplace=True)
)
def forward(self, x):
x1 = self.blocks1(x)
x2 = self.blocks2(x)
x3 = self.blocks3(x)
x4 = self.blocks4(x)
print(x1.shape,x2.shape,x3.shape,x4.shape)
xM = torch.cat([x1 , x2 , x3 , x4],dim=1)
return xM
class classifier(nn.Module):
def __init__(self,inputs ,outs):
super(classifier,self).__init__()
self.block = nn.AvgPool2d(kernel_size=5 ,stride = 3)
self.conv = nn.Conv2d(in_channels=inputs ,out_channels=128 ,kernel_size=1 ,stride = 1 )
self.fc1 = nn.Linear(128*4*4 ,1024)
self.fc2 = nn.Linear(1024, outs)
def forward(self, x):
x = self.block(x)
x = self.conv(x)
x = F.relu(x)
x = torch.flatten(x ,start_dim=1)
print(x.shape)
x = F.relu(self.fc1(x))
x = F.dropout(x , p = 0.5 )
x = self.fc2(x)
return x
class inceptionMid(nn.Module):
def __init__(self,outputs_Numclass =1000 ,trains = True):
super(inceptionMid,self).__init__()
self.inputsinceptions = nn.Sequential(
nn.Conv2d(
in_channels= 3, out_channels=64 ,kernel_size=7 ,stride = 2 ,padding=3),
nn.ReLU(inplace=True),
nn.MaxPool2d(kernel_size=3 , stride=2 ,padding= 1 ),
nn.Conv2d(in_channels= 64, out_channels=64 ,kernel_size=1 ,stride = 1 ),
nn.ReLU(inplace=True),
nn.Conv2d(in_channels=64, out_channels=192, kernel_size=3, stride=1,padding=1),
nn.ReLU(inplace=True),
nn.MaxPool2d(kernel_size=3 , stride=2 , padding=1)
)
self.traini = trains
self.block1 = inceptionModuleV1(inputs=192, outputs1 = 64 , outputs2 = 96,outputs3 = 128,outputs4 = 16,outputs5 = 32,outputs6 = 32)
self.block2 =inceptionModuleV1(inputs=256, outputs1 = 128 , outputs2 = 128,outputs3 = 192,outputs4 = 32,outputs5 = 96,outputs6 = 64)
self.block3 = inceptionModuleV1(inputs=480, outputs1 = 192 , outputs2 = 96,outputs3 = 208,outputs4 = 16,outputs5 = 48,outputs6 = 64)
self.classif1 = classifier(512,1000)
self.classif2 = classifier(528, 1000)
self.maxpooling1 = nn.MaxPool2d(kernel_size=3, stride=2, padding=1)
self.block4 = inceptionModuleV1(inputs=512, outputs1=160, outputs2=112, outputs3=224, outputs4=24, outputs5=64,
outputs6=64)
self.block5 = inceptionModuleV1(inputs=512, outputs1=128, outputs2=128, outputs3=128*2, outputs4=24, outputs5=64,
outputs6=64)
self.block6 = inceptionModuleV1(inputs=512, outputs1=112, outputs2=144, outputs3=288, outputs4=32, outputs5=32*2,
outputs6=32*2)
self.block7 = inceptionModuleV1(inputs=528, outputs1=256, outputs2=160, outputs3=320, outputs4=32, outputs5=128,
outputs6=128)
self.block8 = inceptionModuleV1(inputs=832, outputs1=256, outputs2=160, outputs3=320, outputs4=32, outputs5=128,
outputs6=128)
self.block9 = inceptionModuleV1(inputs=832, outputs1=384, outputs2=192, outputs3=384, outputs4=48, outputs5=128,
outputs6=128)
self.lastFc = nn.Linear(1024,outputs_Numclass)
def forward(self,x):
x = self.inputsinceptions(x)
x = self.block1(x)
x =self.block2(x)
x = self.maxpooling1(x)
x =self.block3(x)
aur = self.classif1(x)
x = self.block4(x)
x =self.block5(x)
x =self.block6(x)
aur1 = self.classif2(x)
x = self.block7(x)
x = nn.MaxPool2d(kernel_size=3, stride=2, padding=1)(x)
x =self.block8(x)
x =self.block9(x)
x= nn.AvgPool2d(kernel_size=7,stride=1 )(x)
x= nn.Dropout(p= 0.4)(x)
x = torch.flatten(x , start_dim=1)
x = self.lastFc(x)
if self.traini == True:
return aur,aur1,x
else:
return x
if __name__ == '__main__':
model = inceptionMid()
input = torch.randn(1, 3, 224, 224)
aux1, aux2, out = model(input)
print(aux1.shape)
print(aux2.shape)
print(out.shape)
print(out)
结果：
torch.Size([1, 64, 28, 28]) torch.Size([1, 128, 28, 28]) torch.Size([1, 32, 28, 28]) torch.Size([1, 32, 28, 28])
torch.Size([1, 128, 28, 28]) torch.Size([1, 192, 28, 28]) torch.Size([1, 96, 28, 28]) torch.Size([1, 64, 28, 28])
torch.Size([1, 192, 14, 14]) torch.Size([1, 208, 14, 14]) torch.Size([1, 48, 14, 14]) torch.Size([1, 64, 14, 14])
torch.Size([1, 2048])
torch.Size([1, 160, 14, 14]) torch.Size([1, 224, 14, 14]) torch.Size([1, 64, 14, 14]) torch.Size([1, 64, 14, 14])
torch.Size([1, 128, 14, 14]) torch.Size([1, 256, 14, 14]) torch.Size([1, 64, 14, 14]) torch.Size([1, 64, 14, 14])
torch.Size([1, 112, 14, 14]) torch.Size([1, 288, 14, 14]) torch.Size([1, 64, 14, 14]) torch.Size([1, 64, 14, 14])
torch.Size([1, 2048])
torch.Size([1, 256, 14, 14]) torch.Size([1, 320, 14, 14]) torch.Size([1, 128, 14, 14]) torch.Size([1, 128, 14, 14])
torch.Size([1, 256, 7, 7]) torch.Size([1, 320, 7, 7]) torch.Size([1, 128, 7, 7]) torch.Size([1, 128, 7, 7])
torch.Size([1, 384, 7, 7]) torch.Size([1, 384, 7, 7]) torch.Size([1, 128, 7, 7]) torch.Size([1, 128, 7, 7])
torch.Size([1, 1000])
torch.Size([1, 1000])
torch.Size([1, 1000])
tensor([[ 2.4082e-02, -5.8420e-02, -1.6949e-02, 2.4449e-02, -3.1221e-02,
9.9499e-03, -1.0006e-02, -2.0253e-02, -1.8088e-02, -3.7963e-02,
-3.7275e-02, -3.3681e-02, 3.5420e-02, 2.8644e-02, 2.4509e-02,
3.2399e-02, -1.0624e-02, -5.6492e-03, -2.3976e-02, 1.2449e-02,
-3.3331e-03, -3.3036e-02, -2.6920e-03, -4.6949e-02, -5.7959e-03,
-4.0529e-02, 7.5937e-03, 3.9678e-02, -1.2029e-02, 2.1007e-03,
6.8112e-03, -2.6430e-02, 7.0184e-03, -3.8686e-03, 2.1889e-02,
4.3211e-03, -1.2345e-02, -1.8592e-02, -1.5193e-02, 5.0065e-03,
7.8268e-03, -1.8961e-02, 1.8172e-04, -2.0164e-02, 2.8331e-02,
1.0635e-02, 6.0798e-04, -1.7938e-02, 1.5952e-02, -1.0309e-02,
-1.6886e-02, 4.7581e-03, 1.7903e-02, -1.0933e-02, 2.5766e-02,
5.7573e-03, -5.2550e-02, -3.3916e-02, -1.6154e-03, 2.6200e-02,
-1.2264e-02, -2.0989e-02, -1.1595e-02, -3.4834e-02, -4.2556e-03,
3.4324e-02, -5.0266e-03, 4.0287e-02, -2.2876e-02, 1.3520e-02,
-3.0992e-02, -1.7167e-02, -8.6823e-03, -2.8599e-03, -7.4686e-03,
-3.6761e-03, -1.3066e-02, 4.6699e-03, 3.1325e-02, 2.3370e-02,
-4.2785e-02, -2.2272e-03, -2.2222e-02, -1.5821e-02, 1.9573e-02,
-2.2068e-02, -6.5074e-03, -8.1939e-03, -1.2682e-02, 8.4411e-03,
-1.2017e-02, 3.0149e-02, -1.1643e-02, 1.8636e-02, 4.7695e-02,
3.9128e-02, 2.7035e-02, -4.1406e-02, 1.2780e-02, -4.7799e-04,
-6.5371e-03, 4.4641e-04, -9.4797e-03, -2.3611e-02, -3.7593e-02,
-3.1111e-02, 2.1029e-02, 2.0499e-02, 3.8366e-02, 2.3151e-02,
-1.3156e-02, -1.7605e-03, 1.1638e-02, -4.0335e-02, -3.5895e-03,
7.7831e-03, -3.5065e-02, -1.4641e-02, 1.7871e-02, -3.1669e-02,
5.0952e-03, -7.8447e-03, 1.0265e-02, 4.6557e-03, -2.5233e-05,
3.3268e-02, -3.9162e-03, 7.2370e-03, 9.4397e-03, 1.3352e-02,
-1.9267e-03, -1.1092e-02, -3.2677e-02, -5.1903e-02, 1.5745e-03,
2.7011e-02, -1.0265e-02, 1.1868e-02, -3.2690e-03, 4.1474e-03,
1.6440e-02, 4.7689e-03, 3.7905e-02, 2.9972e-02, -4.2048e-02,
-3.6497e-03, 7.8156e-03, -3.9843e-02, 6.7347e-03, -8.7800e-03,
3.0305e-02, 5.0274e-04, 4.7436e-03, -3.5154e-02, 1.9998e-02,
-6.0062e-03, 1.7108e-02, 1.0258e-02, 1.0531e-02, -1.4720e-02,
1.8512e-02, -1.2778e-02, -2.4172e-03, 5.8789e-03, 1.0503e-02,
1.9236e-02, 9.2774e-03, 1.1328e-02, -3.2919e-02, -2.3924e-03,
-8.5038e-03, 3.1433e-03, 1.8359e-02, -1.1192e-02, -3.1357e-02,
2.0731e-02, 1.5249e-03, 2.0635e-03, 3.8798e-02, 3.0231e-03,
-1.8950e-03, 8.7791e-03, -1.2833e-03, 2.0644e-02, -5.6835e-03,
1.1199e-02, -7.8529e-03, 5.6543e-03, -1.1213e-02, -3.1228e-02,
4.5107e-02, -1.6331e-02, 2.5534e-02, -8.7112e-03, 1.5752e-02,
1.0887e-02, 1.2107e-02, -3.8128e-02, -4.4536e-02, -7.7093e-03,
2.3829e-02, -1.5492e-02, -2.1253e-02, 3.7076e-02, 4.6774e-02,
-1.7859e-02, 9.1062e-03, 1.6448e-02, -2.7026e-02, 4.7439e-02,
-3.3028e-02, -1.5446e-02, 7.6355e-04, -2.6603e-02, 2.4906e-02,
-2.0782e-02, -1.5107e-02, -1.9324e-02, -1.2933e-03, 1.5820e-02,
-1.5549e-02, -1.3298e-02, -2.8954e-03, 9.9213e-04, -1.1579e-03,
-1.8003e-02, -1.9594e-02, -1.8866e-02, -1.7737e-02, 2.7968e-02,
-2.3817e-02, -6.5641e-03, 1.7233e-02, 1.1122e-02, 3.5230e-03,
3.2654e-02, 2.7185e-02, 2.4406e-02, 1.4909e-02, -4.1884e-02,
1.9934e-02, -3.5589e-03, 2.1564e-02, 2.7957e-02, 1.8883e-02,
8.8828e-03, 1.5761e-02, 1.2281e-02, -2.5506e-02, -1.2245e-02,
-1.8268e-02, 2.0923e-02, -4.7589e-02, -4.1264e-02, -4.1494e-02,
-3.8508e-02, 2.0369e-02, -1.7287e-02, 1.7207e-02, 8.1962e-03,
-5.4090e-03, -2.0081e-02, -3.0943e-02, 9.3603e-03, 1.2516e-03,
9.1543e-03, -4.1601e-02, 1.6724e-02, 1.9197e-02, 1.5865e-02,
-2.4194e-02, -5.7647e-04, -2.0852e-02, -3.0386e-02, 1.0097e-02,
4.6852e-03, 3.7471e-02, 1.9684e-02, -7.1819e-03, -4.4601e-03,
2.4531e-02, -6.4913e-02, -1.2066e-02, 4.4049e-03, 1.0368e-02,
-3.0523e-02, -2.1976e-02, 1.3212e-02, -4.3818e-04, -1.0769e-02,
6.0931e-03, -1.0415e-02, -3.1166e-03, -2.8357e-02, 2.7949e-02,
8.1592e-03, -3.5564e-02, 2.3272e-02, -3.0470e-03, 2.1489e-02,
-1.2392e-02, 2.9945e-03, -2.5286e-02, 2.7782e-02, 2.3024e-02,
2.4525e-02, 1.5654e-02, 1.5213e-02, -1.2619e-02, -1.6058e-02,
1.4045e-02, -3.0519e-02, 8.2257e-03, -1.4888e-02, -1.2208e-02,
-2.9892e-02, -7.7522e-03, 3.2680e-03, 1.4804e-02, -1.6078e-02,
-2.2648e-02, 1.0364e-02, 2.5817e-02, 2.1088e-02, 3.6622e-02,
2.8599e-02, -2.4781e-02, -2.2218e-02, -9.0425e-03, 1.4814e-02,
-7.5499e-03, 3.1399e-02, -1.5732e-03, 1.6703e-02, -4.1085e-02,
-1.5341e-02, 1.2894e-02, 1.2118e-02, 1.3661e-02, 4.8410e-02,
-2.7868e-02, 1.9562e-02, -4.9497e-03, -1.0930e-02, -1.8428e-02,
1.6511e-02, 2.1969e-02, -1.2764e-02, -4.3133e-02, -5.9951e-03,
-3.0336e-03, 8.5734e-03, -4.9862e-03, 1.1979e-02, -2.4210e-03,
1.6173e-03, -2.7066e-03, -2.7072e-02, -1.5158e-02, 1.5197e-02,
-1.9974e-02, 2.0532e-03, 2.5726e-02, -1.5820e-02, 4.0023e-03,
-8.4175e-04, 3.2635e-02, -2.0302e-02, -1.3907e-02, 3.0658e-02,
3.2124e-02, 7.1663e-03, -7.8291e-03, -3.6056e-02, 2.2520e-02,
6.5366e-03, 4.3307e-03, -9.8971e-03, 1.1610e-02, -1.9075e-02,
4.6235e-03, 2.0744e-02, -2.7762e-02, 1.8821e-02, -1.7459e-03,
-7.6045e-03, -9.3994e-03, 2.3197e-02, 1.7146e-02, 9.3226e-03,
-4.0184e-02, -8.7527e-03, 2.2346e-02, -6.8706e-03, 3.5168e-02,
-2.6045e-02, -1.7890e-02, -3.3777e-02, -3.8513e-02, -2.9700e-02,
1.1137e-02, 3.9390e-03, -4.0386e-02, 2.5594e-03, -2.5501e-02,
-3.2641e-02, 3.7139e-02, -4.5358e-02, 6.5998e-04, 5.0904e-02,
-3.8759e-03, -1.1821e-02, 3.8974e-02, -4.1670e-03, 1.3982e-02,
-2.9981e-02, 1.2272e-02, 4.5514e-02, -1.2491e-02, 1.7688e-02,
2.9717e-02, 3.9861e-02, -1.6042e-02, 1.6126e-02, -1.4905e-02,
-3.0308e-02, -3.0771e-02, 1.6590e-02, 2.8907e-03, 1.7373e-02,
6.5046e-03, 1.4448e-02, 2.9400e-02, -3.7911e-02, -6.5282e-03,
-2.8108e-02, -1.8571e-02, 1.6104e-02, 1.0774e-02, -3.0158e-02,
2.2438e-02, -3.7657e-03, 3.3016e-02, 3.0791e-02, -8.1840e-03,
2.7840e-02, 3.1950e-02, -3.8733e-02, 1.6285e-02, 3.7405e-02,
3.4211e-02, -7.4677e-03, -2.9737e-03, 5.1064e-03, 1.7600e-02,
2.3516e-02, 2.3141e-02, 6.7744e-03, -2.2302e-02, -7.2515e-03,
2.7171e-02, -8.1749e-04, -9.0510e-03, -1.6749e-02, -2.8215e-02,
3.2985e-02, 4.5769e-03, -3.2471e-03, 2.4396e-02, 2.5836e-02,
-2.8360e-02, 5.2908e-03, 4.1436e-02, -2.7840e-02, 2.2693e-02,
1.4262e-02, 7.7141e-03, -2.6910e-03, 1.3212e-02, 5.8913e-03,
1.4510e-02, 3.4849e-02, -7.7778e-03, -1.2928e-02, -4.2563e-02,
-1.1851e-02, -2.4793e-03, -1.6587e-03, -1.1851e-04, 1.8163e-02,
-7.5610e-03, -2.5529e-02, -2.0527e-02, 3.0055e-02, -4.6836e-03,
-4.8483e-03, -1.7288e-02, -1.6413e-02, 3.7989e-04, 1.8169e-02,
2.5964e-02, 2.0588e-02, -1.5517e-02, -2.4541e-02, -1.9917e-03,
-8.5226e-03, -1.0326e-02, 2.6165e-03, 2.6116e-02, -9.8867e-03,
-4.2200e-02, -8.5166e-03, -2.6917e-02, 1.3766e-02, 2.6235e-03,
-1.9894e-03, 6.3684e-03, -1.3894e-02, -9.3206e-03, 1.2365e-02,
-3.1835e-02, -2.7892e-02, -7.7311e-03, -1.1453e-02, 2.1802e-02,
6.3711e-03, -2.9177e-02, -3.8801e-02, 2.8015e-02, 4.0248e-02,
-1.6877e-02, 2.7930e-02, -2.0983e-02, 2.1100e-02, -9.0967e-03,
-2.3590e-02, -2.1185e-02, -5.4761e-04, -2.3765e-02, -5.9543e-03,
1.0147e-02, -4.1056e-02, -2.6868e-02, -3.2207e-02, -1.2326e-02,
-1.1984e-02, -1.1907e-02, -2.0721e-02, -3.1955e-02, -2.0342e-03,
-2.8468e-02, -1.0702e-02, -9.4689e-03, 2.7801e-02, 1.8132e-02,
3.8859e-03, 9.8106e-03, 3.4932e-02, -1.8404e-02, -1.1318e-02,
-2.7875e-02, 2.7088e-02, 1.0845e-02, 2.1144e-02, -2.4799e-02,
-4.2368e-03, -2.1209e-02, 9.8159e-03, -3.0673e-02, -8.6852e-03,
-2.6721e-02, 3.6820e-03, 8.3970e-04, 1.1959e-02, -2.2295e-02,
1.0240e-04, -1.0409e-02, -2.4296e-03, 3.1209e-02, 6.3481e-03,
-2.0317e-02, -1.7248e-02, 1.3495e-02, 2.9935e-02, -1.0744e-02,
1.1985e-02, 3.7952e-03, 6.9590e-03, 4.5128e-02, -3.5511e-02,
8.2757e-03, 1.9478e-02, 1.2501e-02, -1.7899e-02, 2.3701e-02,
1.4647e-02, 2.1040e-02, 2.7586e-02, -2.8423e-02, -7.3693e-03,
4.4682e-02, -2.9940e-02, 3.9848e-02, 6.2207e-03, 2.4694e-02,
2.3230e-02, -3.1932e-03, -9.9655e-03, 1.0513e-02, 2.4302e-02,
-5.4736e-03, 1.4624e-02, -1.7328e-02, 7.9377e-03, 1.6687e-02,
3.6305e-02, -1.7780e-02, 2.3812e-02, -6.3378e-03, -3.5877e-02,
3.3251e-03, -1.7411e-02, 6.8248e-04, -2.5844e-02, -4.5487e-03,
-1.8799e-02, -5.4912e-03, -8.8579e-03, -3.1248e-02, 3.6441e-03,
3.4262e-02, -3.4388e-02, -1.0399e-03, 9.7923e-03, -2.3480e-02,
4.2403e-02, -1.8477e-02, 2.2839e-02, -3.1252e-02, 8.9852e-03,
-2.6305e-02, -1.6721e-02, -1.9544e-02, -2.3927e-02, -1.5472e-02,
1.5737e-02, 1.5991e-02, -2.2882e-02, 1.3390e-02, -2.7449e-02,
-1.7273e-02, -6.9247e-03, -1.6713e-02, 1.4949e-02, 1.5176e-02,
1.9066e-02, -2.1817e-02, -4.0542e-03, -9.2348e-03, 1.4565e-02,
1.2855e-02, 7.3751e-03, -1.4878e-02, 2.5662e-02, -3.0453e-02,
-3.9791e-03, 7.5653e-03, -9.2595e-03, 2.9804e-03, -1.4995e-02,
2.6876e-02, 1.4967e-03, 8.7974e-03, 1.6473e-02, 2.6183e-02,
3.0166e-03, -1.6064e-03, -2.6358e-02, -2.2579e-02, 8.0824e-03,
2.1325e-02, 1.3667e-02, -1.0321e-02, 1.7947e-02, 1.4116e-02,
1.8065e-02, -2.6304e-02, 2.6044e-02, -1.5716e-02, -6.7091e-03,
2.0704e-02, 8.3390e-03, 4.3058e-02, 5.7722e-03, 1.6593e-02,
-1.1851e-02, 2.1086e-02, 1.3658e-02, 3.9412e-02, 1.3713e-02,
3.7449e-02, 8.7189e-03, 1.6376e-02, 1.4716e-02, -9.5834e-03,
-1.8180e-02, 1.7997e-02, 1.2260e-03, 4.9386e-02, -2.9460e-02,
2.1820e-02, -2.3730e-02, 2.8945e-02, 1.6169e-02, -2.7648e-03,
1.6288e-02, -1.7214e-02, 4.2345e-02, -8.3445e-03, 3.5156e-02,
3.9470e-02, 1.0655e-02, 5.4865e-03, -1.7921e-03, 1.7450e-02,
-2.7785e-02, -1.4330e-02, -3.1285e-02, 2.6268e-02, -2.7772e-02,
-2.4397e-02, -2.3181e-02, 2.6009e-02, -1.5131e-02, 2.8456e-02,
-1.6606e-02, 5.1522e-03, 2.7270e-02, -3.8541e-02, -2.1967e-02,
2.9303e-02, 1.4878e-02, 1.5034e-02, -2.1997e-03, -8.3801e-03,
-2.1234e-02, -2.4439e-03, 3.8561e-02, 1.3133e-02, 1.1736e-02,
2.4341e-02, -3.2635e-02, 3.1007e-02, 3.3495e-02, -7.7501e-03,
-7.1284e-03, 1.0267e-02, -2.8204e-02, 1.2758e-02, 5.1837e-03,
-2.1703e-02, 2.4386e-02, 1.1329e-02, -3.2334e-02, 5.4065e-03,
-2.3549e-02, -2.6736e-02, 1.8833e-02, 3.3755e-03, 4.2148e-02,
2.2605e-02, -2.4928e-03, -2.9420e-02, -4.7755e-02, -3.1332e-02,
-2.9352e-02, 1.7874e-02, 2.9931e-02, 2.2340e-02, 1.6732e-02,
5.3264e-03, -3.3778e-02, -1.9956e-02, 4.9978e-03, -5.2016e-03,
-1.6985e-03, 1.0104e-02, -5.1241e-02, -8.4232e-03, -1.3560e-02,
-1.9405e-02, -4.8525e-02, -4.3083e-02, -2.9999e-02, 6.0505e-02,
8.3615e-03, 1.6270e-02, -2.6771e-02, 2.5608e-03, -1.4557e-02,
1.5263e-02, -1.4419e-02, -5.4917e-04, -3.2982e-02, -1.6840e-02,
-1.1572e-02, 1.5651e-02, 1.3509e-02, 7.3461e-03, -2.8315e-02,
2.0102e-02, -4.4708e-02, 1.9183e-02, 7.8982e-05, 1.7029e-04,
-2.0587e-02, 5.9534e-03, 9.7653e-03, -3.6275e-04, 1.7876e-02,
-1.7093e-02, 3.7239e-02, -1.0755e-02, -3.5184e-04, -1.1173e-02,
4.6020e-02, 7.5721e-03, -1.7287e-02, -3.7060e-02, 7.8803e-03,
3.7895e-02, -1.3081e-02, 2.7829e-02, -4.9862e-02, 2.3073e-02,
1.0341e-02, -1.7600e-02, 3.4911e-02, 3.4776e-02, 3.2101e-02,
1.0664e-03, -3.0447e-02, 2.4802e-03, -1.3010e-02, -2.8162e-03,
-1.4231e-02, 2.6457e-02, 1.1781e-02, -3.9268e-04, 8.4195e-04,
1.4934e-02, -2.6485e-02, 1.8859e-02, 1.3969e-02, -2.0413e-03,
2.2274e-02, 5.0586e-03, 8.9480e-03, 2.5266e-02, -4.9164e-03,
-1.7366e-02, 1.4795e-02, 3.0111e-02, -1.0350e-02, 2.2511e-02,
8.0158e-03, -1.5161e-02, -2.7918e-03, -9.1841e-03, 5.6364e-03,
-1.9910e-02, 9.2131e-03, -3.3319e-03, -3.7577e-03, 2.2115e-03,
-6.1464e-02, -1.8719e-02, 5.1098e-02, 2.3493e-02, 7.1576e-03,
-3.8655e-02, -8.6109e-03, 2.6675e-02, 1.7054e-02, 3.6374e-02,
-1.7054e-02, 1.6608e-02, -1.4358e-02, 7.2648e-04, -1.3183e-03,
-1.6677e-02, 1.1783e-03, -9.1894e-04, 1.6137e-03, 7.7208e-03,
1.1354e-02, 4.2095e-03, 2.1681e-02, -4.3674e-04, -1.3702e-02,
-2.4548e-02, 4.2742e-03, -3.2218e-03, 1.7599e-02, -3.5182e-02,
-3.0997e-02, 1.0245e-02, -4.2554e-03, -2.4211e-02, 7.2424e-03,
6.2503e-03, 4.5980e-04, -1.4893e-02, 1.6730e-02, -2.3941e-02,
1.9243e-02, -2.8669e-02, 6.0019e-03, 1.0314e-02, -2.0390e-02,
1.7210e-02, -2.6136e-02, -2.4784e-02, 1.6919e-02, -4.6081e-04,
-3.5554e-02, 2.2545e-02, -6.1369e-03, -1.6966e-02, 7.0338e-03,
2.8618e-02, -4.4364e-02, -1.5062e-02, -3.5687e-02, 1.2352e-03,
-4.0268e-04, -1.2330e-03, 8.1580e-03, -7.7175e-03, 2.1781e-02,
-8.4020e-03, -2.6037e-02, -1.0351e-03, 2.6721e-02, -1.9837e-02,
6.4983e-03, -1.8126e-03, -1.5037e-02, -1.3538e-02, -1.1233e-02,
2.2226e-02, 6.3896e-03, 7.2887e-03, 3.3260e-02, -7.0485e-03,
1.6398e-02, 3.9379e-03, 1.4868e-02, -1.1855e-02, -1.9606e-02,
-1.9075e-02, 1.0427e-02, 2.6717e-02, -1.0573e-02, -7.0019e-03,
1.8224e-02, -2.7268e-03, -3.6840e-02, 5.5233e-03, 6.4812e-03,
1.0367e-03, -1.3217e-02, -2.6354e-03, 1.2521e-02, -1.5189e-02,
2.6507e-02, -3.7729e-02, -1.8050e-02, -3.2470e-02, -3.7903e-02,
2.4325e-02, -5.1281e-03, 1.6413e-02, -3.3414e-02, 1.8317e-02,
1.6425e-02, -3.8652e-02, 1.6691e-02, 1.4437e-02, -1.9905e-02,
3.2776e-02, -2.7744e-02, -1.2522e-02, -5.7744e-03, -2.4705e-02,
-1.2601e-02, -1.4709e-02, -1.5809e-02, -2.5390e-02, 4.0849e-02]],
grad_fn=<AddmmBackward>)
调试了很久 终于弄明白了
644
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
