我正在尝试尝试为什么我们有消失和爆炸梯度，以及为什么Resnet对于避免上述两个问题如此有帮助。因此，我决定训练一个具有多层的普通卷积网络，只是为了知道为什么当我训练多层（例如 20 层）时模型LOSS会增加。但我在某个时候收到此错误，我可以找出可能是什么问题，但我知道它来自我的模型架构。
images.shape: torch.Size([128, 3, 32, 32])
---------------------------------------------------------------------------
RuntimeError Traceback (most recent call last)
<ipython-input-80-0ad7109b33c1> in <module>
1 for images, labels in train_dl:
2 print('images.shape:', images.shape)
----> 3 out = model(images)
4 print('out.shape:', out.shape)
5 print('out[0]:', out[0])
/opt/conda/lib/python3.7/site-packages/torch/nn/modules/module.py in _call_impl(self, *input, **kwargs)
725 result = self._slow_forward(*input, **kwargs)
726 else:
--> 727 result = self.forward(*input, **kwargs)
728 for hook in itertools.chain(
729 _global_forward_hooks.values(),
<ipython-input-78-81b21c16ed79> in forward(self, xb)
31
32 def forward(self, xb):
---> 33 return self.network(xb)
/opt/conda/lib/python3.7/site-packages/torch/nn/modules/module.py in _call_impl(self, *input, **kwargs)
725 result = self._slow_forward(*input, **kwargs)
726 else:
--> 727 result = self.forward(*input, **kwargs)
728 for hook in itertools.chain(
729 _global_forward_hooks.values(),
/opt/conda/lib/python3.7/site-packages/torch/nn/modules/container.py in forward(self, input)
115 def forward(self, input):
116 for module in self:
--> 117 input = module(input)
118 return input
119
/opt/conda/lib/python3.7/site-packages/torch/nn/modules/module.py in _call_impl(self, *input, **kwargs)
725 result = self._slow_forward(*input, **kwargs)
726 else:
--> 727 result = self.forward(*input, **kwargs)
728 for hook in itertools.chain(
729 _global_forward_hooks.values(),
/opt/conda/lib/python3.7/site-packages/torch/nn/modules/conv.py in forward(self, input)
421
422 def forward(self, input: Tensor) -> Tensor:
--> 423 return self._conv_forward(input, self.weight)
424
425 class Conv3d(_ConvNd):
/opt/conda/lib/python3.7/site-packages/torch/nn/modules/conv.py in _conv_forward(self, input, weight)
418 _pair(0), self.dilation, self.groups)
419 return F.conv2d(input, weight, self.bias, self.stride,
--> 420 self.padding, self.dilation, self.groups)
421
422 def forward(self, input: Tensor) -> Tensor:
RuntimeError: Given groups=1, weight of size [64, 32, 3, 3], expected input[128, 64, 32, 32] to have 32 channels, but got 64 channels instead
我的模型架构是
class Cifar10CnnModel(ImageClassificationBase):
def __init__(self):
super().__init__()
self.network = nn.Sequential(
nn.Conv2d(3, 32, kernel_size=3, padding=1),
nn.Conv2d(32, 32, kernel_size=3, padding=1),
nn.ReLU(),
nn.Conv2d(32, 64, kernel_size=3, stride=1, padding=1),
nn.Conv2d(32, 64, kernel_size=3, stride=1, padding=1),
nn.ReLU(),
nn.MaxPool2d(2, 2), # output: 64 x 16 x 16
nn.Conv2d(64, 128, kernel_size=3, stride=1, padding=1),
nn.ReLU(),
nn.Conv2d(128, 128, kernel_size=3, stride=1, padding=1),
nn.ReLU(),
nn.MaxPool2d(2, 2), # output: 128 x 8 x 8
nn.Conv2d(128, 256, kernel_size=3, stride=1, padding=1),
nn.ReLU(),
nn.Conv2d(256, 256, kernel_size=3, stride=1, padding=1),
nn.ReLU(),
nn.MaxPool2d(2, 2), # output: 256 x 4 x 4
nn.Flatten(),
nn.Linear(256*4*4, 1024),
nn.ReLU(),
nn.Linear(1024, 512),
nn.ReLU(),
nn.Linear(512, 10))
def forward(self, xb):
return self.network(xb)
for images, labels in train_dl:
print('images.shape:', images.shape)
out = model(images)
print('out.shape:', out.shape)
print('out[0]:', out[0])
break
我可以通过模型看到，看起来您在序列中的第四个转换块上犯了一个拼写错误。 你有
nn.Conv2d(32, 64, kernel_size=3, stride=1, padding=1),
nn.Conv2d(32, 64, kernel_size=3, stride=1, padding=1),
但是，您已经将图像转换为 64 个通道，然后将其作为具有 32 个通道的图像传递到下一个转换块，这会导致上面的错误。
将此问题修复为：
self.network = nn.Sequential(
nn.Conv2d(3, 32, kernel_size=3, padding=1),
nn.Conv2d(32, 32, kernel_size=3, padding=1),
nn.ReLU(),
nn.Conv2d(32, 64, kernel_size=3, stride=1, padding=1),
# Change this from 32 to now 64 like I did here.
nn.Conv2d(64, 64, kernel_size=3, stride=1, padding=1),
nn.ReLU(),
nn.MaxPool2d(2, 2), # output: 64 x 16 x 16
沙克·耆那教
请告诉我我必须更改什么，我不是编码人员，所以请使其简单，以便我能够理解
!!! Exception during processing!!! Given groups=1, weight of size [320, 5, 3, 3], expected input[16, 4, 144, 256] to have 5 channels, but got 4 channels instead
回溯（最近一次调用最后一次）： 文件“E:\ComfyUI_windows_portable\ComfyUI xecution.py”，第 151 行，在 recursive_execute 中 输出数据，输出UI = get_output_data(obj, input_data_all) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 文件“E:\ComfyUI_windows_portable\ComfyUI xecution.py”，第 81 行，在 get_output_data 中 return_values = map_node_over_list(obj, input_data_all, obj.FUNCTION, allowed_interrupt=True) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ ^^^^^^^^^^^^^^^^^^^^^^^^^^^ 文件“E:\ComfyUI_windows_portable\ComfyUI xecution.py”，第 74 行，map_node_over_list 结果.append(getattr(obj, func)(**slice_dict(input_data_all, i))) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ ^ 文件“E:\ComfyUI_windows_portable\ComfyUI odes.py”，第 1405 行，样本中 返回common_ksampler（模型，noise_seed，步骤，cfg，sampler_name，调度程序，正值，负值，latent_image，降噪=降噪，disable_noise=disable_noise，start_step=start_at_step，last_step=end_at_step，force_full_denoise=force_full_denoise） ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 文件“E:\ComfyUI_windows_portable\ComfyUI odes.py”，第 1341 行，在 common_ksampler 中 样本= comfy.sample.sample（模型，噪声，步骤，cfg，sampler_name，调度程序，正数，负数，潜在图像， ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ ^^^^ 文件“E:\ComfyUI_windows_portable\ComfyU