gpt4 book ai didi

python - 难以理解修改后的 VGG16 前向方法 (Pytorch) 的行为

转载 作者:行者123 更新时间:2023-12-04 07:32:54 26 4
gpt4 key购买 nike

我在 pytorch 中修改了 VGG16 以在特征提取器中插入 BN 和 dropout 等内容。当我将 forward 方法的定义从以下位置更改时,我偶然发现了一些奇怪的东西:

def forward(self, x):
x = self.model(x)
return x
到:
def forward(self, x):
x = self.model.features(x)
x = self.model.avgpool(x)
x = self.model.classifier(x)
return x
在第二种方法中,我现在收到一个错误,即矩阵的大小不匹配
( mat1 dim 1 must match mat2 dim 0)
以下是我一直在使用的 VGG 编辑版本的完整代码。
class Vgg(nn.Module):

def __init__(self, n_classes, bias= None, dropout = 0.3):

super().__init__()

self.model = models.vgg16()
#self.bn64 = nn.BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
#self.bn128 = nn.BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
#self.bn256 = nn.BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
#self.bn512 = nn.BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)

# change to allow 4 channels input
self.model.features[0] = nn.Conv2d(4, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3))

# remove/edit some of the first layers to make it more similar to Resnet
del self.model.features[2]
del self.model.features[2]
del self.model.features[-1]
self.model.features[2] = nn.MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False)

# add dropout
for m in self.model.modules():
if isinstance(m, nn.Dropout):
m.p = dropout
else:
pass

self.dropout = nn.Dropout(p=dropout)
self.r = nn.ReLU(inplace=True)

modules = nn.Sequential(*[self.model.features[0],
#self.bn64,
self.model.features[1:3],
#self.bn64,
self.model.features[3:5],
#self.dropout,

self.model.features[5],
#self.bn128,
self.model.features[6:8],
#self.bn128,
self.model.features[8:10],
#self.dropout,

self.model.features[10],
#self.bn256,
self.model.features[11:13],
#self.bn256,
self.model.features[13:15],
#self.bn256,
self.model.features[15:17],
#self.dropout,

self.model.features[17],
#self.bn512,
self.model.features[18:20],
#self.bn512,
self.model.features[20:22],
#self.bn512,
self.model.features[22:24],
#self.dropout,

self.model.features[24],
#self.bn512,
self.model.features[25:27],
#self.bn512,
self.model.features[27:29],
#self.bn512,
#self.dropout
nn.MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
])


self.model.features = modules

# change the pooling layer
self.model.avgpool = nn.AdaptiveAvgPool2d(output_size=(1, 1))

# set output to correct num classes
self.model.classifier = nn.Linear(in_features=512, out_features=n_classes, bias=True)

# use predefined bias
if bias is not None:
assert isinstance(bias, torch.Tensor), 'bias must be tensor'
self.model.classifier.bias = nn.Parameter(bias)

def forward(self, x):
x = self.model.features(x)
x = self.model.avgpool(x)
x = self.model.classifier(x)
return x
我知道它看起来非常丑陋和hacky。我试图重新编写它,但重新编写的版本由于某种原因也不起作用,我假设当前的问题也与此有关。我认为输入并没有像我认为的那样通过 forward 方法提供。我的假设是调用 x = self.model(x)不会在我制作的所有已编辑图层中运行“x”,否则我会在上面的两个版本的 forward 方法中得到相同的行为。但我的问题是,当我调用 self.model(x) 时会发生什么?在前进?它是通过 pytorch 的原始 vgg16 运行输入吗?因为当我打印 self.model在我的控制台中,它显示了我对 self.model.features 的架构所做的更改以及 self.model.avgpoolself.model.classifier .
编辑:
以下是错误的整个跟踪。一些额外的信息。 t是我为处理训练步骤而制作的类(class)(因此循环训练和验证模式等)
---------------------------------------------------------------------------
RuntimeError Traceback (most recent call last)
<ipython-input-7-0b3983ae9702> in <module>
77 print_cl_met = True
78 )
---> 79 model = t.run()
80 t.save_to_json()
81 print(np.max(np.array(t.f1_tracker)))

/home/stevea/treesat/TreeSat/TreeSat/trainers/basetrainer.py in run(self)
342 for phase in ['training', 'testing']:
343 self.phase = phase
--> 344 self.model_mode()
345
346 if self.phase == 'testing':

/home/stevea/treesat/TreeSat/TreeSat/trainers/basetrainer.py in model_mode(self)
154 if self.phase == 'training':
155 print('*********TRAINING PHASE*********')
--> 156 self.trainModel()
157 else:
158 print('*********VALIDATION PHASE*********')

/home/stevea/treesat/TreeSat/TreeSat/trainers/basetrainer.py in trainModel(self)
234 # loop through all batches to perform an epoch
235 for loaded in self.loaders[self.phase]:
--> 236 epoch_loss = self.train_step(loaded, epoch_loss)
237
238 mean_loss = np.mean(np.array(epoch_loss))

/home/stevea/treesat/TreeSat/TreeSat/trainers/basetrainer.py in train_step(self, loaded, epoch_loss)
262
263 # process batch through network
--> 264 self.out = self.model(self.img_batch.float())
265
266 # get loss value

/usr/local/lib/python3.6/dist-packages/torch/nn/modules/module.py in _call_impl(self, *input, **kwargs)
887 result = self._slow_forward(*input, **kwargs)
888 else:
--> 889 result = self.forward(*input, **kwargs)
890 for hook in itertools.chain(
891 _global_forward_hooks.values(),

/home/stevea/treesat/TreeSat/TreeSat/models/vgg.py in forward(self, x)
143 x = self.model.features(x)
144 x = self.model.avgpool(x)
--> 145 x = self.model.classifier(x)
146 return x
147

/usr/local/lib/python3.6/dist-packages/torch/nn/modules/module.py in _call_impl(self, *input, **kwargs)
887 result = self._slow_forward(*input, **kwargs)
888 else:
--> 889 result = self.forward(*input, **kwargs)
890 for hook in itertools.chain(
891 _global_forward_hooks.values(),

/usr/local/lib/python3.6/dist-packages/torch/nn/modules/linear.py in forward(self, input)
92
93 def forward(self, input: Tensor) -> Tensor:
---> 94 return F.linear(input, self.weight, self.bias)
95
96 def extra_repr(self) -> str:

/usr/local/lib/python3.6/dist-packages/torch/nn/functional.py in linear(input, weight, bias)
1751 if has_torch_function_variadic(input, weight):
1752 return handle_torch_function(linear, (input, weight), input, weight, bias=bias)
-> 1753 return torch._C._nn.linear(input, weight, bias)
1754
1755

RuntimeError: mat1 dim 1 must match mat2 dim 0
我还尝试在 forward 方法的每一步打印出 x 的形状:
def forward(self, x):
x = self.model.features(x)
print(x.shape)
x = self.model.avgpool(x)
print(x.shape)
x = self.model.classifier(x)
print(x.shape)
return x
它向我展示了形状似乎很好,因为分类器应该采用 512 个特征:
torch.Size([64, 512, 4, 4])
torch.Size([64, 512, 1, 1])

最佳答案

我无法运行您的代码,但我认为问题在于线性层需要 2d 数据输入(因为它实际上是矩阵乘法),而您提供 4d 输入(尺寸为 1 的暗淡 2 和 3)。
请尝试squeeze

def forward(self, x):
x = self.model.features(x)
x = self.model.avgpool(x)
x = torch.squeeze(x)
x = self.model.classifier(x)
return x

压缩部分中看起来不那么 hacky 的代码,请参阅 torch einops .

关于python - 难以理解修改后的 VGG16 前向方法 (Pytorch) 的行为,我们在Stack Overflow上找到一个类似的问题: https://stackoverflow.com/questions/67870887/

26 4 0
Copyright 2021 - 2024 cfsdn All Rights Reserved 蜀ICP备2022000587号
广告合作:1813099741@qq.com 6ren.com