书籍

​​https://github.com/zergtant/pytorch-handbook​​ Pytorch指南

​​PyTorch 中文手册（pytorch handbook） - Pytorch中文手册​​

​​https://github.com/yunjey/pytorch-tutorial​​ Pytorch指南3

​​Practical Deep Learning for Coders - Practical Deep Learning​​

吴恩达老师的 ​​deeplearning.ai​​ 课程板书

示例

​​GitHub - espectre/Kaggle_Dogs_vs_Cats_PyTorch: kaggle competition: Dogs_vs_Cats_PyTorch Presentation（Getting started with PyTorch）​​

​​9 个超酷的深度学习案例 - 走看看​​

​​GitHub - jcjohnson/neural-style: Torch implementation of neural style algorithm​​

基础

​​Python字符串的处理​​

Tensor


import torch
a = torch.tensor([1, 2, 3, 4])
b = torch.tensor([1, 2, 3, 4])
print(a.equal(b))
 
# 返回结果：True


import torch
c = torch.tensor([1, 2, 2, 3])
d = torch.tensor([2, 2, 3, 3])
print(c.eq(d))
 
# 返回结果：tensor([0, 1, 0, 1], dtype=torch.uint8)

​​torch.max()使用讲解​​

​​Pytorch笔记torch.max() - 知乎​​

函数

​​Pytorch autograd,backward详解 - 知乎​​

​​python魔法方法之__setattr__() - 知乎​​

class Module:
    def __setattr__(self, name: str, value: Union[Tensor, 'Module']) -> None:
        def remove_from(*dicts_or_sets):
            for d in dicts_or_sets:
                if name in d:
                    if isinstance(d, dict):
                        del d[name]
                    else:
                        d.discard(name)

        params = self.__dict__.get('_parameters')
        if isinstance(value, Parameter):
            if params is None:
                raise AttributeError(
                    "cannot assign parameters before Module.__init__() call")
            remove_from(self.__dict__, self._buffers, self._modules, self._non_persistent_buffers_set)
            self.register_parameter(name, value)
        elif params is not None and name in params:
            if value is not None:
                raise TypeError("cannot assign '{}' as parameter '{}' "
                                "(torch.nn.Parameter or None expected)"
                                .format(torch.typename(value), name))
            self.register_parameter(name, value)
        else:
            modules = self.__dict__.get('_modules')
            if isinstance(value, Module):
                if modules is None:
                    raise AttributeError(
                        "cannot assign module before Module.__init__() call")
                remove_from(self.__dict__, self._parameters, self._buffers, self._non_persistent_buffers_set)
                modules[name] = value
            elif modules is not None and name in modules:
                if value is not None:
                    raise TypeError("cannot assign '{}' as child module '{}' "
                                    "(torch.nn.Module or None expected)"
                                    .format(torch.typename(value), name))
                modules[name] = value
            else:
                buffers = self.__dict__.get('_buffers')
                if buffers is not None and name in buffers:
                    if value is not None and not isinstance(value, torch.Tensor):
                        raise TypeError("cannot assign '{}' as buffer '{}' "
                                        "(torch.Tensor or None expected)"
                                        .format(torch.typename(value), name))
                    buffers[name] = value
                else:
                    object.__setattr__(self, name, value)

从上可以看出，Module中对__setattr__函数进行了重载，它会收集里面的Parameter，Module，Tensor，分别放在_parameters，_modules，_buffers中，便于后续操作里面的参数，模块。

卷积


​​PyTorch中的nn.Conv1d与nn.Conv2d​​

​​Pytorch.nn.conv2d 过程验证（单，多通道卷积过程） - 知乎​​

​​PyTorch中torch.linspace的详细用法 - 知乎​​

​​Python __new__()方法详解​​

torchvision


torchvision.transforms.ToTensor ：把一个取值范围是[0,255]的PIL.Image或者shape为(H,W,C)的numpy.ndarray，转换成形状为[C,H,W]，取值范围是[0,1.0]的torch.FloadTensor

torchvision.transforms.ToPILImage：将shape为(C,H,W)的Tensor或shape为(H,W,C)的numpy.ndarray转换成PIL.Image，值不变。

torchvision.transforms.Normalize(mean, std)：用给定的均值和标准差分别对每个通道的数据进行正则化。具体来说，给定均值(M1,…,Mn)，给定标准差(S1,…,Sn)，其中n是通道数（一般是3），对每个通道进行如下操作：
output[channel] = (input[channel] - mean[channel]) / std[channel]

比如：原来的tensor是三个维度的，值在[0,1]之间，经过变换之后就到了[-1，1]
计算如下：(（0,1）-0.5）/0.5=(-1,1)

PIL 库

​​Python 学习笔记之—— PIL 库 - 知乎​​

​​Pillow（PIL）入门教程（非常详细）​​

from PIL import Image

# Numpy 转 Image
img = Image.fromarray(array)


# Image 转 Numpy
array = Image.fromarray(img)

matplotlib.pyplot

​​Matplotlib Pyplot | 菜鸟教程​​

​​matplotlib.pyplot的使用总结大全（入门加进阶） - 知乎​​