查看“Pytorch”的源代码

*英文文档：[https://pytorch.org/docs/stable/index.html]
*中文文档  [https://pytorch.apachecn.org/][https://pytorch-cn.readthedocs.io/zh/latest/package_references/functional/#convolution]

==网络初始化==
*Xavier and Kaiming initialization [https://www.jianshu.com/p/f2d800388d1c]

==函数==
*torch.clamp(input, min=None, max=None, *, out=None) → Tensor
:Clamps all elements in input into the range [ min, max ]. Letting min_value and max_value be min and max, respectively
*torch.eye(n, m=None, out=None)
:返回一个2维张量，对角线位置全1，其它位置全0
*torch.view()
:相当于numpy的reshape，某个维度上等于-1，就是让计算机自己算一下这一维度上应该有多少
*torch.cat & torch.stack
:orch.stack()沿着一个新维度对输入张量序列进行连接。 序列中所有的张量都应该为相同形状。
:torch.cat()是为了把函数torch.stack()得到tensor进行拼接而存在的 (不增加新的维度)
*gather
:In PyTorch you can perform the same operation using the gather() method. If s is a PyTorch Tensor of shape (N, C) and y is a PyTorch Tensor of shape (N,) containing longs in the range 0 <= y[i] < C, then s.gather(1, y.view(-1, 1)).squeeze() will be a PyTorch Tensor of shape (N,) containing one entry from each row of s, selected according to the indices in y.
*squeeze() 
:删除一个张量中所有维数为1的维度

==Tensor==
* cpu() numpy() detach() item() [https://blog.csdn.net/ODIMAYA/article/details/102892799]
:注意cuda上面的变量类型只能是tensor，不能是其他

==torchvision==
*PyTorch框架中有一个非常重要且好用的包：torchvision，该包主要由3个子包组成，分别是：torchvision.datasets、torchvision.models、torchvision.transforms [https://www.jianshu.com/p/1ae863c1e66d]
*__all__ = ["Compose", "ToTensor", "ToPILImage", "Normalize", "Resize",
"Scale", "CenterCrop", "Pad", "Lambda", "RandomCrop", 
"RandomHorizontalFlip", "RandomVerticalFlip", "RandomResizedCrop", 
"RandomSizedCrop", "FiveCrop", "TenCrop","LinearTransformation", 
"ColorJitter", "RandomRotation", "Grayscale", "RandomGrayscale"]

==第三方库==
*thop
:THOP 是 PyTorch 非常实用的一个第三方库，可以统计模型的 FLOPs 和参数量。
 from thop import clever_format
 from thop import profile
 class YourModule(nn.Module):
    # your definition
 input = torch.randn(10, 128, 128)
 flops, params = profile(model, inputs=(input, ))
 flops, params = clever_format([flops, params], "%.3f")