You can not select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
206 lines
4.5 KiB
206 lines
4.5 KiB
{
|
|
"cells": [
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"# 3.7 softmax回归的简洁实现"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 1,
|
|
"metadata": {},
|
|
"outputs": [
|
|
{
|
|
"name": "stdout",
|
|
"output_type": "stream",
|
|
"text": [
|
|
"0.4.1\n"
|
|
]
|
|
}
|
|
],
|
|
"source": [
|
|
"import torch\n",
|
|
"from torch import nn\n",
|
|
"from torch.nn import init\n",
|
|
"import numpy as np\n",
|
|
"import sys\n",
|
|
"sys.path.append(\"..\") \n",
|
|
"import d2lzh_pytorch as d2l\n",
|
|
"\n",
|
|
"print(torch.__version__)"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"## 3.7.1 获取和读取数据"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 2,
|
|
"metadata": {
|
|
"collapsed": true
|
|
},
|
|
"outputs": [],
|
|
"source": [
|
|
"batch_size = 256\n",
|
|
"train_iter, test_iter = d2l.load_data_fashion_mnist(batch_size)"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"## 3.7.2 定义和初始化模型"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 3,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"num_inputs = 784\n",
|
|
"num_outputs = 10\n",
|
|
"\n",
|
|
"# class LinearNet(nn.Module):\n",
|
|
"# def __init__(self, num_inputs, num_outputs):\n",
|
|
"# super(LinearNet, self).__init__()\n",
|
|
"# self.linear = nn.Linear(num_inputs, num_outputs)\n",
|
|
"# def forward(self, x): # x shape: (batch, 1, 28, 28)\n",
|
|
"# y = self.linear(x.view(x.shape[0], -1))\n",
|
|
"# return y\n",
|
|
" \n",
|
|
"# net = LinearNet(num_inputs, num_outputs)\n",
|
|
"\n",
|
|
"class FlattenLayer(nn.Module):\n",
|
|
" def __init__(self):\n",
|
|
" super(FlattenLayer, self).__init__()\n",
|
|
" def forward(self, x): # x shape: (batch, *, *, ...)\n",
|
|
" return x.view(x.shape[0], -1)\n",
|
|
"\n",
|
|
"from collections import OrderedDict\n",
|
|
"net = nn.Sequential(\n",
|
|
" # FlattenLayer(),\n",
|
|
" # nn.Linear(num_inputs, num_outputs)\n",
|
|
" OrderedDict([\n",
|
|
" ('flatten', FlattenLayer()),\n",
|
|
" ('linear', nn.Linear(num_inputs, num_outputs))])\n",
|
|
" )"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 4,
|
|
"metadata": {},
|
|
"outputs": [
|
|
{
|
|
"data": {
|
|
"text/plain": [
|
|
"Parameter containing:\n",
|
|
"tensor([0., 0., 0., 0., 0., 0., 0., 0., 0., 0.], requires_grad=True)"
|
|
]
|
|
},
|
|
"execution_count": 4,
|
|
"metadata": {},
|
|
"output_type": "execute_result"
|
|
}
|
|
],
|
|
"source": [
|
|
"init.normal_(net.linear.weight, mean=0, std=0.01)\n",
|
|
"init.constant_(net.linear.bias, val=0) "
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"## 3.7.3 softmax和交叉熵损失函数"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 5,
|
|
"metadata": {
|
|
"collapsed": true
|
|
},
|
|
"outputs": [],
|
|
"source": [
|
|
"loss = nn.CrossEntropyLoss()"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"## 3.7.4 定义优化算法"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 6,
|
|
"metadata": {
|
|
"collapsed": true
|
|
},
|
|
"outputs": [],
|
|
"source": [
|
|
"optimizer = torch.optim.SGD(net.parameters(), lr=0.1)"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"## 3.7.5 训练模型"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 7,
|
|
"metadata": {},
|
|
"outputs": [
|
|
{
|
|
"name": "stdout",
|
|
"output_type": "stream",
|
|
"text": [
|
|
"epoch 1, loss 0.0031, train acc 0.748, test acc 0.785\n",
|
|
"epoch 2, loss 0.0022, train acc 0.813, test acc 0.802\n",
|
|
"epoch 3, loss 0.0021, train acc 0.824, test acc 0.808\n",
|
|
"epoch 4, loss 0.0020, train acc 0.833, test acc 0.824\n",
|
|
"epoch 5, loss 0.0019, train acc 0.837, test acc 0.806\n"
|
|
]
|
|
}
|
|
],
|
|
"source": [
|
|
"num_epochs = 5\n",
|
|
"d2l.train_ch3(net, train_iter, test_iter, loss, num_epochs, batch_size, None, None, optimizer)"
|
|
]
|
|
}
|
|
],
|
|
"metadata": {
|
|
"kernelspec": {
|
|
"display_name": "Python [default]",
|
|
"language": "python",
|
|
"name": "python3"
|
|
},
|
|
"language_info": {
|
|
"codemirror_mode": {
|
|
"name": "ipython",
|
|
"version": 3
|
|
},
|
|
"file_extension": ".py",
|
|
"mimetype": "text/x-python",
|
|
"name": "python",
|
|
"nbconvert_exporter": "python",
|
|
"pygments_lexer": "ipython3",
|
|
"version": "3.6.3"
|
|
}
|
|
},
|
|
"nbformat": 4,
|
|
"nbformat_minor": 2
|
|
}
|