PyTorch

(未完) import torch import torch.nn as nn import numpy as np def softmax(x): return np.exp(x) / n

收获（源于学长代码讲解） 打断点 之前也知道程序要打断点，就比如刷力扣的题目。不过对于深度学习就没有想起来打断点的习惯。一个特别大的好处是打断点之后，就可以清晰地知道各个变量的数据类型，和具体的情况，

Transforms can be applied to PIL images, tensors, ndarrays, or custom data during creation of the Da

--- highlight: a11y-dark --- ``` import torch import torchvision from torch.utils.data import Datase

--- highlight: a11y-light --- ```python import torch import torch.nn as nn import numpy as np from s

``` # gradient_numpy.py import numpy as np # prepare for data X = np.array([1, 2, 3, 4], dtype=np.fl