jiamenghu的博客

其实文章目录【深度学习基础】梯度下降算法1、批量梯度下降（`BGD`）2、随机梯度下降（`SGD`）2、`Online GD`3、Adam 优化【深度学习基础】梯度下降算法while !(Stop condition) weights_grad = evaluate_gradients(loss_fun,data,weights) weights -= step_size * weights_grad流程1、确定初始状态（初始位置）weights,也就是初始化2、计算当前状态的梯度weig

来自于 https://tangshusen.me/Dive-into-DL-PyTorch/#/官方文档 https://pytorch.org/docs/stable/tensors.html模型构建Module类是nn模块里提供的一个模型构造类，是所有神经网络模块的基类，我们可以继承它来定义我们想要的模型。下面继承Module类构造本节开头提到的多层感知机。这里定义的MLP类重载了Module类的__init__函数和forward函数。它们分别用于创建模型参数和定义前向计算。前向计算也即正向传

来自于 https://tangshusen.me/Dive-into-DL-PyTorch/#/官方文档 https://pytorch.org/docs/stable/tensors.html多层感知机import torchimport numpy as npimport torchvisionimport torchvision.transforms as transformsimport sysmnist_train = torchvision.datasets.FashionM

来自于 https://tangshusen.me/Dive-into-DL-PyTorch/#/官方文档 https://pytorch.org/docs/stable/tensors.htmlsoftmax回归Fashion-MNIST[2]获取数据集import torchimport torchvisionimport torchvision.transforms as transformsimport matplotlib.pyplot as pltimport timeimp

来自于 https://tangshusen.me/Dive-into-DL-### PyTorch/#/官方文档 https://pytorch.org/docs/stable/tensors.html一、线性回归%matplotlib inlineimport torchfrom IPython import displayfrom matplotlib import pyplot as plt import numpy as npimport random生成数据集num_i

来自于 https://tangshusen.me/Dive-into-DL-PyTorch/#/官方文档 https://pytorch.org/docs/stable/tensors.html一、创建tensorimport torchx=torch.empty(5,3)print(x)tensor([[0., 0., 0.], [0., 0., 0.], [0., 0., 0.], [0., 0., 0.], [0., 0