qq_37791134的博客

API:https://pytorch.org/docs/master/optim.html#how-to-adjust-learning-rate在该网页中搜索lr_scheduler的话，可以看到：lr_scheduler含15种类，参考这个博文可以可视化看这些API的使用:https://www.jianshu.com/p/a20d5a7ed6f3我这里主要用的是：MultiStepLR然后对比一下：MultiStepLR和StepLR，主要区别就是在于前者可以指定区间...

1.model.train()与model.eval()的用法看别人的面经时，浏览到一题，问的就是这个。自己刚接触pytorch时套用别人的框架，会在训练开始之前写上model.trian()，在测试时写上model.eval()。然后自己写的时候也就保留了这个习惯，没有去想其中原因。在经过一番查阅之后，总结如下：如果模型中有BN层(Batch Normalization）和Dropout，需要在训练时添加model.train()，在测试时添加model.eval()。其中model.train

模型vgg：VGG( (features): Sequential( (0): Conv2d(3, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)) (1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) (2): ReLU(inplace=True) (3): Conv2d(64, 64, k

题外话：这里还是得会用python用的6呀，谁说用深度学习不需要啥高深的python用法，简单就可以，其实还是要对python的各种库非常熟悉的。 # print(arr[:, 0]) # [3 3 5] # print(Counter(arr[:, 0])) # Counter({3: 2, 5: 1}) # print(Counter(arr[:, 0]).most_common(1)) # [(3, 2)] # pri...

前面在（二）中已经得到下面的结果啦：[10, 2000] loss: 0.367[10, 4000] loss: 0.382[10, 6000] loss: 0.417[10, 8000] loss: 0.472[10, 10000] loss: 0.452[10, 12000] loss: 0.488Finished TrainingAccuracy of the network on the 10000 test images: 68 %Accuracy of plane

前面用的网络是pytorch官方给的一个实例网络，本文参照书本换了一个网络，如下：代码如下：class CNNnet(nn.Module): def __init__(self): super(CNNnet,self).__init__() self.conv1 = nn.Conv2d(in_channels=3, out_channels=16, kernel_size=5,stride=1) self.pool1 = nn.MaxPo

不能在GPU上用哈，# net = net.to(device)# 像keras一样显示各层参数import collectionsdef paras_summary(input_size, model): def register_hook(module): def hook(module, input, output): class_name = str(module.__class__).split('.')[-1].split("'"

利用tensorflowX对特征图进行可视化，不同卷积层的特征图的抽取程度不一样的，# 利用tensorboardX可视化特征图import torchvision.utils as vutilsfrom tensorboardX import SummaryWriterwriter = SummaryWriter(log_dir='logs', comment='feature map')for i,data in enumerate(trainloader, 0): # 获取训练数

# -*- coding: utf-8 -*-'''@Time : 2020/8/11 17:14@Author : HHNa@FileName: add.image.py@Software: PyCharm '''# 导入库及下载数据import torchimport torchvisionimport torchvision.transforms as transformsimport torch.utils.data as datatransform = tr.

该错误就是说没有在GPU是计算怎么纠正？首先# 检测是否有用的GPU，如果有使用GPU，否则CPUdevice = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")net = Net() #实例化网络为netnet = net.to(device)然后在for data in testLoader: images, labels = data images, labels = imag..

https://pytorch.org/docs/master/generated/torch.nn.AdaptiveMaxPool2d.html#torch.nn.AdaptiveMaxPool2dtorch.nn.AdaptiveMaxPool2d output_size– the target output size of the image of the form H x W. Can be a tuple (H, W) or a single H for a square ima..

https://pytorch.org/docs/master/generated/torch.nn.ConvTranspose2d.html#torch.nn.ConvTranspose2d in_channels(int) – Number of channels in the input image out_channels(int) – Number of channels produced by the convolution kernel_size(intor...

参考链接，我真的觉得看官网就是最棒的参考！清清楚楚，明明白白https://pytorch.org/docs/master/generated/torch.nn.Conv2d.html#torch.nn.Conv2d# 测试一下输出尺寸# With square kernels and equal stridem = nn.Conv2d(16, 33, 3, stride=2)# non-square kernels and unequal stride and with paddi.

运行Pytorch tutorial代码报错：BrokenPipeError: [Errno 32] Broken pipe源代码地址：Training a classifier (CIFAR10)该问题的产生是由于windows下多线程的问题，和DataLoader类有关，具体细节点这里Fix memory leak when using multiple workers on Windows。解决方案：修改调用torch.utils.data.DataLoader()函数时的...