卷积神经网络随记

miodi

于 2024-07-31 19:30:00 发布

阅读量1.2k

点赞数 26

文章标签： cnn 深度学习神经网络

本文链接：https://blog.csdn.net/miodi/article/details/140814447

版权

1.问题描述：一般而言，几个小滤波器卷积层的组合比一个大滤波器卷积层要好，比如层层堆叠了3个3x3的卷积层，中间含有非线性激活层，在这种排列下面，第一个卷积层中每个神经元对输入数据的感受野是3x3，第二层卷积层对第一层卷积层的感受野也是3x3，这样对于输入数据的感受野就是5x5，同样，第三层卷积层上对第二层卷积层的感受野是3x3，这样第三层卷积层对于第一层输入数据的感受野就是7x7。

问题：如何理解“对于输入数据的感受野就是5x5”“第三层卷积层对于第一层输入数据的感受野就是7x7”用计算公式说明

解答：

在卷积神经网络中，感受野是指卷积神经网络每一层输出的特征图（feature map）上的像素点在输入图像上映射的区域大小。换句话说，它定义了输出特征图中的一个元素与输入图像之间的空间范围。

对于堆叠的卷积层，每一层的感受野可以通过以下公式计算：

$RF_{n+1} = RF_{n}+\left ( K_{n}-1 \right )$

其中， $RF_{n}$ 是第 n 层的感受野大小， $K_{n}$ 是第 n 层的核（滤波器）大小。

以3x3的卷积层为例，如果连续堆叠3个这样的卷积层，我们可以这样计算每层的感受野：

1. 第一层卷积层使用3x3的滤波器，所以其感受野为 $RF_{1} = 3 * 3 = 3$ 。
2. 第二层卷积层对第一层的输出应用另一个3x3的滤波器，因此感受野增加到 $RF_{2} = RF_{1}+(K_{1}-1) = 3 + (3-1) = 5$ 。
3. 第三层卷积层再次应用一个3x3的滤波器，感受野进一步增加到 $RF_{3}=RF_{2}+(k_{2}-1) = 3 + (5-1) =7$ 。

因此，第三层卷积层对于第一层输入数据的感受野是7x7，这意味着第三层的每一个元素都对应于原始输入图像中的一个7x7的区域。

这种层层堆叠小滤波器的策略可以增加网络的深度和非线性，同时保持参数数量相对较少，这通常可以提高网络的性能，因为它能够捕捉到更复杂的模式和特征。

2.问题描述：Traceback (most recent call last): File "/home/xc/PycharmProjects/Liuyafei/cnn/net.py", line 57, in <module> conv_model.add_module(layer[0], layer[1]) File "/home/xc/anaconda3/envs/share_env/lib/python3.8/site-packages/torch/nn/modules/module.py", line 390, in add_module raise KeyError("module name can't contain \".\", got: {}".format(name)) KeyError: 'module name can\'t contain ".", got: layer1.conv1'

解决方法：这个错误是因为在添加模块时，模块名称不能包含"."。你可以尝试将层的名称中的"."替换为其他字符，例如下划线"_"。

conv_model = SimpleCNN()
for layer in model.named_modules():
    if isinstance(layer[1], nn.Conv2d):
        # 将层名称中的"."替换为"_"
        new_name = layer[0].replace('.', '_')
        conv_model.add_module(new_name, layer[1])

3.问题：怎么提取卷积模块，使用print打印出来

解决方法：要提取卷积模块并使用print打印出来，可以遍历模型的层，检查每一层是否为卷积层（nn.Conv2d），如果是，则打印出该层的详细信息。

import torch
import torch.nn as nn

# 假设你已经定义了一个名为model的卷积神经网络模型
model = ...

# 遍历模型的所有层
for name, module in model.named_modules():
    # 检查当前层是否为卷积层
    if isinstance(module, nn.Conv2d):
        # 打印卷积层的名称和参数
        print(f"Layer: {name}")
        print(f"Kernel size: {module.kernel_size}")
        print(f"Stride: {module.stride}")
        print(f"Padding: {module.padding}")
        print(f"Dilation: {module.dilation}")
        print(f"Groups: {module.groups}")
        print("-----")

这段代码将遍历模型的所有层，如果发现某一层是卷积层（nn.Conv2d），就会打印出该层的名称、核大小、步长、填充、扩张和分组等信息。

4.问题：/home/xc/PycharmProjects/Liuyafei/cnn/net.py:68: UserWarning: nn.init.normal is now deprecated in favor of nn.init.normal_. init.normal(m.weight.data) /home/xc/PycharmProjects/Liuyafei/cnn/net.py:69: UserWarning: nn.init.xavier_normal is now deprecated in favor of nn.init.xavier_normal_. init.xavier_normal(m.weight.data) /home/xc/PycharmProjects/Liuyafei/cnn/net.py:70: UserWarning: nn.init.kaiming_normal is now deprecated in favor of nn.init.kaiming_normal_. init.kaiming_normal(m.weight.data)

解决方法：要修改这些警告，需要将nn.init.normal、nn.init.xavier_normal和nn.init.kaiming_normal替换为它们的下划线版本，即nn.init.normal_、nn.init.xavier_normal_和nn.init.kaiming_normal_。

5.问题：Traceback (most recent call last): File "/home/xc/PycharmProjects/Liuyafei/MNIST/cnn.py", line 68, in <module> model = CNN(28 * 28, 300, 100, 10) TypeError: __init__() takes 1 positional argument but 5 were given

解决方法：CNN类定义中并没有接受这些参数的构造函数。实际上，已经在__init__方法中初始化了网络的各个层，因此不需要在创建CNN对象时传递任何参数。

model = CNN()

6.问题：

Traceback (most recent call last): File "/home/xc/PycharmProjects/Liuyafei/MNIST/cnn.py", line 85, in <module> outputs = model(inputs.view(-1, 28 * 28)) File "/home/xc/anaconda3/envs/share_env/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1130, in _call_impl return forward_call(*input, **kwargs) File "/home/xc/PycharmProjects/Liuyafei/MNIST/cnn.py", line 42, in forward x = self.layer1(x) File "/home/xc/anaconda3/envs/share_env/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1130, in _call_impl return forward_call(*input, **kwargs) File "/home/xc/anaconda3/envs/share_env/lib/python3.8/site-packages/torch/nn/modules/container.py", line 139, in forward input = module(input) File "/home/xc/anaconda3/envs/share_env/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1130, in _call_impl return forward_call(*input, **kwargs) File "/home/xc/anaconda3/envs/share_env/lib/python3.8/site-packages/torch/nn/modules/conv.py", line 457, in forward return self._conv_forward(input, self.weight, self.bias) File "/home/xc/anaconda3/envs/share_env/lib/python3.8/site-packages/torch/nn/modules/conv.py", line 453, in _conv_forward return F.conv2d(input, weight, bias, self.stride, RuntimeError: Expected 3D (unbatched) or 4D (batched) input to conv2d, but got input of size: [64, 784]

解决方法：

这个错误是因为卷积层期望输入的数据维度为3D（未批量化）或4D（批量化），但实际输入的数据维度为[64, 784]。要解决这个问题，需要将输入数据调整为正确的维度。可以通过在输入数据上添加一个额外的维度来实现这一点。

首先，确保输入数据的维度为[batch_size, channels, height, width]。在这个例子中，channels应该为1，因为MNIST数据集是灰度图像。
修改代码，将输入数据调整为正确的维度。

# 假设 inputs 是一个形状为 [batch_size, 784] 的张量
inputs = inputs.view(-1, 1, 28, 28)  # 将输入数据调整为 [batch_size, 1, 28, 28]
outputs = model(inputs)

7.问题：

Traceback (most recent call last): File "/home/xc/PycharmProjects/Liuyafei/MNIST/cnn.py", line 118, in <module> out = model(img) File "/home/xc/anaconda3/envs/share_env/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1130, in _call_impl return forward_call(*input, **kwargs) File "/home/xc/PycharmProjects/Liuyafei/MNIST/cnn.py", line 42, in forward x = self.layer1(x) File "/home/xc/anaconda3/envs/share_env/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1130, in _call_impl return forward_call(*input, **kwargs) File "/home/xc/anaconda3/envs/share_env/lib/python3.8/site-packages/torch/nn/modules/container.py", line 139, in forward input = module(input) File "/home/xc/anaconda3/envs/share_env/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1130, in _call_impl return forward_call(*input, **kwargs) File "/home/xc/anaconda3/envs/share_env/lib/python3.8/site-packages/torch/nn/modules/conv.py", line 457, in forward return self._conv_forward(input, self.weight, self.bias) File "/home/xc/anaconda3/envs/share_env/lib/python3.8/site-packages/torch/nn/modules/conv.py", line 453, in _conv_forward return F.conv2d(input, weight, bias, self.stride, RuntimeError: Expected 3D (unbatched) or 4D (batched) input to conv2d, but got input of size: [64, 784]

解决方法：

在测试阶段，需要将输入数据调整为适合卷积层的形状。具体来说，卷积层期望的输入形状是 (batch_size, channels, height, width)，而当前的输入数据形状是 (batch_size, 784)。

要解决这个问题，需要在测试阶段的输入数据上添加一个通道维度。可以使用 unsqueeze() 函数来实现这一点。

# eval
eval_loss = 0
eval_acc = 0
for data in test_loader:
    img, label = data
    img = img.view(img.size(0), 1, 28, 28)  # 添加通道维度
    if torch.cuda.is_available():
        img = Variable(img).cuda()
        label = Variable(label).cuda()
    else:
        img = Variable(img)
        label = Variable(label)
    out = model(img)
    loss = criterion(out, label)
    eval_loss += loss.item() * label.size(0)
    _, pred = torch.max(out, 1)
    num_correct = (pred == label).sum()
    eval_acc += num_correct.item()

print('Test Loss: {:.6f}, ACC: {:.6f}'.format(eval_loss / (len(test_dataset)), eval_acc / (len(test_dataset))))

8.问题：

或者这个：

Traceback (most recent call last):
  File "/home/xc/anaconda3/envs/share_env/lib/python3.8/urllib/request.py", line 1354, in do_open
    h.request(req.get_method(), req.selector, req.data, headers,
  File "/home/xc/anaconda3/envs/share_env/lib/python3.8/http/client.py", line 1256, in request
    self._send_request(method, url, body, headers, encode_chunked)
  File "/home/xc/anaconda3/envs/share_env/lib/python3.8/http/client.py", line 1302, in _send_request
    self.endheaders(body, encode_chunked=encode_chunked)
  File "/home/xc/anaconda3/envs/share_env/lib/python3.8/http/client.py", line 1251, in endheaders
    self._send_output(message_body, encode_chunked=encode_chunked)
  File "/home/xc/anaconda3/envs/share_env/lib/python3.8/http/client.py", line 1011, in _send_output
    self.send(msg)
  File "/home/xc/anaconda3/envs/share_env/lib/python3.8/http/client.py", line 951, in send
    self.connect()
  File "/home/xc/anaconda3/envs/share_env/lib/python3.8/http/client.py", line 1425, in connect
    self.sock = self._context.wrap_socket(self.sock,
  File "/home/xc/anaconda3/envs/share_env/lib/python3.8/ssl.py", line 500, in wrap_socket
    return self.sslsocket_class._create(
  File "/home/xc/anaconda3/envs/share_env/lib/python3.8/ssl.py", line 1040, in _create
    self.do_handshake()
  File "/home/xc/anaconda3/envs/share_env/lib/python3.8/ssl.py", line 1309, in do_handshake
    self._sslobj.do_handshake()
ssl.SSLCertVerificati torchvision.datasets.CIFAR10(root='./data', train=True, download=True, transform=train_transform)
  File "/home/xc/anaconda3/envs/share_env/lib/python3.8/site-packages/torchvision/datasets/cifar.py", line 65, in __init__
    self.download()
  File "/home/xc/anaconda3/envs/share_env/lib/python3.8/site-packages/torchvision/datasets/cifar.py", line 141, in download
    download_and_extract_archive(self.url, self.root, filename=self.filename, md5=self.tgz_md5)
  File "/home/xc/anaconda3/envs/share_env/lib/python3.8/site-packages/torchvision/datasets/utils.py", line 446, in download_and_extract_archive
    download_url(url, download_root, filename, md5)
  File "/home/xc/anaconda3/envs/share_env/lib/python3.8/site-packages/torchvision/datasets/utils.py", line 146, in download_url
    url = _get_redirect_url(url, max_hops=max_redirect_hops)
  File "/home/xc/anaconda3/envs/share_env/lib/python3.8/site-packages/torchvision/datasets/utils.py", line 94, in _get_redirect_url
    with urllib.request.urlopen(urllib.request.Request(url, headers=headers)) as response:
  File "/home/xc/anaconda3/envs/share_env/lib/python3.8/urllib/request.py", line 222, in urlopen
    return opener.open(url, data, timeout)
  File "/home/xc/anaconda3/envs/share_env/lib/python3.8/urllib/request.py", line 525, in open
    response = self._open(req, data)
  File "/home/xc/anaconda3/envs/share_env/lib/python3.8/urllib/request.py", line 542, in _open
    result = self._call_chain(self.handle_open, protocol, protocol +
  File "/home/xc/anaconda3/envs/share_env/lib/python3.8/urllib/request.py", line 502, in _call_chain
    result = func(*args)
  File "/home/xc/anaconda3/envs/share_env/lib/python3.8/urllib/request.py", line 1397, in https_open
    return self.do_open(http.client.HTTPSConnection, req,
  File "/home/xc/anaconda3/envs/share_env/lib/python3.8/urllib/request.py", line 1357, in do_open
    raise URLError(err)
urllib.error.URLError: <urlopen error [SSL: CERTIFICATE_VERIFY_FAILED] certificate verify failed: unable to get local issuer certificate (_ssl.c:1131)>

进程已结束，退出代码 1

解决办法：

这个错误是由于SSL证书验证失败导致的。可以尝试在代码中禁用SSL证书验证，但请注意这样做可能会导致安全风险。如果仍然想要尝试禁用SSL证书验证，可以在创建urllib.request.urlopen对象之前添加以下代码：

import ssl
ssl._create_default_https_context = ssl._create_unverified_context

将这段代码添加到你的脚本的开头，然后再次运行脚本。

9.问题：

Downloading http://192.168.0.2:80/ac_portal/proxy.html?template=disclaimer&tabs=pwd&vlanid=0&_ID_=0&switch_url=&url=https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz&controller_type=&mac=04-7c-16-5e-85-7c to ./data/cifar-10-python.tar.gz 100%|█████████████████████████████████| 2250/2250 [00:00<00:00, 10922666.67it/s] Traceback (most recent call last): File "/home/xc/PycharmProjects/Liuyafei/cifar/resnet.py", line 25, in <module> trainset = torchvision.datasets.CIFAR10(root='./data', train=True, download=True, transform=train_transform) File "/home/xc/anaconda3/envs/share_env/lib/python3.8/site-packages/torchvision/datasets/cifar.py", line 65, in __init__ self.download() File "/home/xc/anaconda3/envs/share_env/lib/python3.8/site-packages/torchvision/datasets/cifar.py", line 141, in download download_and_extract_archive(self.url, self.root, filename=self.filename, md5=self.tgz_md5) File "/home/xc/anaconda3/envs/share_env/lib/python3.8/site-packages/torchvision/datasets/utils.py", line 446, in download_and_extract_archive download_url(url, download_root, filename, md5) File "/home/xc/anaconda3/envs/share_env/lib/python3.8/site-packages/torchvision/datasets/utils.py", line 167, in download_url raise RuntimeError("File not found or corrupted.") RuntimeError: File not found or corrupted.

这个错误表明在下载过程中出现了问题，导致文件未找到或损坏。可以尝试重新运行代码以重新下载文件。如果问题仍然存在，可能是网络连接问题或者文件链接失效。

10.问题：

Traceback (most recent call last): File "/home/xc/PycharmProjects/Liuyafei/cifar/resnet.py", line 25, in <module> trainset = torchvision.datasets.CIFAR10(root='./data', train=True, download=True, transform=train_transform) File "/home/xc/anaconda3/envs/share_env/lib/python3.8/site-packages/torchvision/datasets/cifar.py", line 65, in __init__ self.download() File "/home/xc/anaconda3/envs/share_env/lib/python3.8/site-packages/torchvision/datasets/cifar.py", line 141, in download download_and_extract_archive(self.url, self.root, filename=self.filename, md5=self.tgz_md5) File "/home/xc/anaconda3/envs/share_env/lib/python3.8/site-packages/torchvision/datasets/utils.py", line 446, in download_and_extract_archive download_url(url, download_root, filename, md5) File "/home/xc/anaconda3/envs/share_env/lib/python3.8/site-packages/torchvision/datasets/utils.py", line 146, in download_url url = _get_redirect_url(url, max_hops=max_redirect_hops) File "/home/xc/anaconda3/envs/share_env/lib/python3.8/site-packages/torchvision/datasets/utils.py", line 94, in _get_redirect_url with urllib.request.urlopen(urllib.request.Request(url, headers=headers)) as response: File "/home/xc/anaconda3/envs/share_env/lib/python3.8/urllib/request.py", line 222, in urlopen return opener.open(url, data, timeout) File "/home/xc/anaconda3/envs/share_env/lib/python3.8/urllib/request.py", line 525, in open response = self._open(req, data) File "/home/xc/anaconda3/envs/share_env/lib/python3.8/urllib/request.py", line 542, in _open result = self._call_chain(self.handle_open, protocol, protocol + File "/home/xc/anaconda3/envs/share_env/lib/python3.8/urllib/request.py", line 502, in _call_chain result = func(*args) File "/home/xc/anaconda3/envs/share_env/lib/python3.8/urllib/request.py", line 1397, in https_open return self.do_open(http.client.HTTPSConnection, req, File "/home/xc/anaconda3/envs/share_env/lib/python3.8/urllib/request.py", line 1358, in do_open r = h.getresponse() File "/home/xc/anaconda3/envs/share_env/lib/python3.8/http/client.py", line 1348, in getresponse response.begin() File "/home/xc/anaconda3/envs/share_env/lib/python3.8/http/client.py", line 316, in begin version, status, reason = self._read_status() File "/home/xc/anaconda3/envs/share_env/lib/python3.8/http/client.py", line 285, in _read_status raise RemoteDisconnected("Remote end closed connection without" http.client.RemoteDisconnected: Remote end closed connection without response

解决方法：

这个错误是由于在下载CIFAR-10数据集时，远程服务器关闭了连接而没有响应。这可能是由于网络问题或服务器问题导致的。你可以尝试以下方法解决这个问题：

检查你的网络连接是否正常。
确保你的防火墙或代理设置不会阻止下载。
稍后再尝试下载，可能是服务器暂时不可用。
如果问题仍然存在，可以考虑手动下载CIFAR-10数据集，然后将其解压缩到指定的目录（在这个例子中是./data）。

11.问题：

Traceback (most recent call last): File "/home/xc/anaconda3/envs/share_env/lib/python3.8/urllib/request.py", line 1354, in do_open h.request(req.get_method(), req.selector, req.data, headers, File "/home/xc/anaconda3/envs/share_env/lib/python3.8/http/client.py", line 1256, in request self._send_request(method, url, body, headers, encode_chunked) File "/home/xc/anaconda3/envs/share_env/lib/python3.8/http/client.py", line 1302, in _send_request self.endheaders(body, encode_chunked=encode_chunked) File "/home/xc/anaconda3/envs/share_env/lib/python3.8/http/client.py", line 1251, in endheaders self._send_output(message_body, encode_chunked=encode_chunked) File "/home/xc/anaconda3/envs/share_env/lib/python3.8/http/client.py", line 1011, in _send_output self.send(msg) File "/home/xc/anaconda3/envs/share_env/lib/python3.8/http/client.py", line 951, in send self.connect() File "/home/xc/anaconda3/envs/share_env/lib/python3.8/http/client.py", line 1425, in connect self.sock = self._context.wrap_socket(self.sock, File "/home/xc/anaconda3/envs/share_env/lib/python3.8/ssl.py", line 500, in wrap_socket return self.sslsocket_class._create( File "/home/xc/anaconda3/envs/share_env/lib/python3.8/ssl.py", line 1040, in _create self.do_handshake() File "/home/xc/anaconda3/envs/share_env/lib/python3.8/ssl.py", line 1309, in do_handshake self._sslobj.do_handshake() Connecti torchvision.datasets.CIFAR10(root='./data', train=True, download=True, transform=train_transform) File "/home/xc/anaconda3/envs/share_env/lib/python3.8/site-packages/torchvision/datasets/cifar.py", line 65, in __init__ self.download() File "/home/xc/anaconda3/envs/share_env/lib/python3.8/site-packages/torchvision/datasets/cifar.py", line 141, in download download_and_extract_archive(self.url, self.root, filename=self.filename, md5=self.tgz_md5) File "/home/xc/anaconda3/envs/share_env/lib/python3.8/site-packages/torchvision/datasets/utils.py", line 446, in download_and_extract_archive download_url(url, download_root, filename, md5) File "/home/xc/anaconda3/envs/share_env/lib/python3.8/site-packages/torchvision/datasets/utils.py", line 146, in download_url url = _get_redirect_url(url, max_hops=max_redirect_hops) File "/home/xc/anaconda3/envs/share_env/lib/python3.8/site-packages/torchvision/datasets/utils.py", line 94, in _get_redirect_url with urllib.request.urlopen(urllib.request.Request(url, headers=headers)) as response: File "/home/xc/anaconda3/envs/share_env/lib/python3.8/urllib/request.py", line 222, in urlopen return opener.open(url, data, timeout) File "/home/xc/anaconda3/envs/share_env/lib/python3.8/urllib/request.py", line 525, in open response = self._open(req, data) File "/home/xc/anaconda3/envs/share_env/lib/python3.8/urllib/request.py", line 542, in _open result = self._call_chain(self.handle_open, protocol, protocol + File "/home/xc/anaconda3/envs/share_env/lib/python3.8/urllib/request.py", line 502, in _call_chain result = func(*args) File "/home/xc/anaconda3/envs/share_env/lib/python3.8/urllib/request.py", line 1397, in https_open return self.do_open(http.client.HTTPSConnection, req, File "/home/xc/anaconda3/envs/share_env/lib/python3.8/urllib/request.py", line 1357, in do_open raise URLError(err) urllib.error.URLError: <urlopen error [Errno 104] Connection reset by peer>

解决方法：

这个错误是由于在尝试下载CIFAR-10数据集时，远程服务器关闭了连接。这可能是由于网络问题或服务器问题导致的。可以尝试以下方法解决这个问题：

检查你的网络连接是否正常。
确保防火墙或代理设置不会阻止下载。
稍后再尝试下载，可能是服务器暂时不可用。
如果问题仍然存在，可以考虑手动下载CIFAR-10数据集，然后将其解压缩到指定的目录（在这个例子中是./data）。

12.问题：

通过这个网址http://192.168.0.2:80/ac_portal/proxy.html?template=disclaimer&tabs=pwd&vlanid=0&_ID_=0&switch_url=&url=https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz&controller_type=&mac=04-7c-16-5e-85-7c下载了名为“cifar-10-python.tar.gz”这个的文件，将下载的文件“cifar-10-python.tar.gz”放到当前运行文件data下面后，运行当前代码，显示以下错误：

Downloading http://192.168.0.2:80/ac_portal/proxy.html?template=disclaimer&tabs=pwd&vlanid=0&_ID_=0&switch_url=&url=https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz&controller_type=&mac=04-7c-16-5e-85-7c to ./data/cifar-10-python.tar.gz 100%|██████████████████████████████████| 2250/2250 [00:00<00:00, 9779465.28it/s] Traceback (most recent call last): File "/home/xc/PycharmProjects/Liuyafei/cifar/resnet.py", line 25, in <module> trainset = torchvision.datasets.CIFAR10(root='./data', train=True, download=True, transform=train_transform) File "/home/xc/anaconda3/envs/share_env/lib/python3.8/site-packages/torchvision/datasets/cifar.py", line 65, in __init__ self.download() File "/home/xc/anaconda3/envs/share_env/lib/python3.8/site-packages/torchvision/datasets/cifar.py", line 141, in download download_and_extract_archive(self.url, self.root, filename=self.filename, md5=self.tgz_md5) File "/home/xc/anaconda3/envs/share_env/lib/python3.8/site-packages/torchvision/datasets/utils.py", line 446, in download_and_extract_archive download_url(url, download_root, filename, md5) File "/home/xc/anaconda3/envs/share_env/lib/python3.8/site-packages/torchvision/datasets/utils.py", line 167, in download_url raise RuntimeError("File not found or corrupted.") RuntimeError: File not found or corrupted.

解决方法：

从错误信息来看，下载的文件可能已经损坏或者文件路径不正确。请尝试以下步骤：

1. 确保文件 "cifar-10-python.tar.gz" 已经被正确下载到 "./data" 目录下。
2. 如果文件不存在，请重新运行代码以下载文件。
3. 如果文件已存在但仍然出现错误，请检查文件是否完整且未损坏。可以尝试重新下载文件。
4. 确保在创建 `CIFAR10` 数据集时使用正确的文件路径。在这个例子中，文件应该位于 "./data/cifar-10-python.tar.gz"。

如果问题仍然存在，可以尝试手动下载文件并将其移动到正确的位置。以下是手动下载和解压文件的步骤：

1. 访问 https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz 并下载文件。
2. 将下载的文件移动到 "./data" 目录下。
3. 确保文件名正确（应为 "cifar-10-python.tar.gz"）。
4. 解压文件：在终端中运行 `tar -xzvf ./data/cifar-10-python.tar.gz -C ./data`。
5. 重新运行代码。

13.问题：

Traceback (most recent call last): File "/home/xc/PycharmProjects/Liuyafei/cifar/resnet.py", line 67, in <module> outputs = net(inputs) File "/home/xc/anaconda3/envs/share_env/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1130, in _call_impl return forward_call(*input, **kwargs) File "/home/xc/PycharmProjects/Liuyafei/cifar/resnet.py", line 46, in forward x = self.pool(F.relu(self.conv1(x))) AttributeError: 'TypeVar' object has no attribute 'relu'

解决方法：

这个错误是因为在代码中使用了`F.relu`，但是没有正确导入`torch.nn.functional`模块。要解决这个问题，需要在代码开头添加以下导入语句：

import torch.nn.functional as F

这样，`F.relu`就可以正常工作了。

miodi

关注

26
点赞
踩
22

收藏

觉得还不错? 一键收藏
0
评论
卷积神经网络随记

一般而言，几个小滤波器卷积层的组合比一个大滤波器卷积层要好，比如层层堆叠了3个3x3的卷积层，中间含有非线性激活层，在这种排列下面，第一个卷积层中每个神经元对输入数据的感受野是3x3，第二层卷积层对第一层卷积层的感受野也是3x3，这样对于输入数据的感受野就是5x5，同样，第三层卷积层上对第二层卷积层的感受野是3x3，这样第三层卷积层对于第一层输入数据的感受野就是7x7。要提取卷积模块并使用print打印出来，可以遍历模型的层，检查每一层是否为卷积层（nn.Conv2d），如果是，则打印出该层的详细信息。
复制链接

扫一扫