pytorch_逻辑回归实现图像分类.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "https://mp.weixin.qq.com/s/nyCJ9TtpctLS3PW7t6Cmgg"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-07-12T12:31:03.267519Z",
     "start_time": "2019-07-12T12:31:03.264069Z"
    }
   },
   "outputs": [],
   "source": [
    "import torch\n",
    "import torchvision\n",
    "from torchvision.datasets import MNIST"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-07-12T12:31:14.280899Z",
     "start_time": "2019-07-12T12:31:14.275206Z"
    }
   },
   "outputs": [],
   "source": [
    "print(torch.__version__)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-07-12T12:34:00.113771Z",
     "start_time": "2019-07-12T12:33:12.290016Z"
    }
   },
   "outputs": [],
   "source": [
    "dataset = MNIST(root='../dataset/',download=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_dataset = MNIST(root='../dataset/',train=False)\n",
    "len(test_dataset)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "dataset[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "image,label = dataset[0]\n",
    "plt.imshow(image)\n",
    "print('Label:',label)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Pytorch无法直接处理图像，需要将图像转换成tensor\n",
    "import torchvision.transforms as transforms"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "dataset = MNIST(root='../dataset/',train=True,transform=transforms.ToTensor())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "img_tensor,label = dataset[0]\n",
    "print(img_tensor.shape,label)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(img_tensor[:,10:15,10:15])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "plt.imshow(img_tensor[0],cmap='gray')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 训练和验证数据集\n",
    "import numpy as np\n",
    "def split_indices(n,val_pct):\n",
    "    # 定义验证集的大小\n",
    "    n_val = int(val_pct*n)\n",
    "    # 创建随机0到n-1的数列\n",
    "    idxs = np.random.permutation(n)\n",
    "    # 挑选第一个n_val为验证集的开始\n",
    "    return idxs[n_val:],idxs[:n_val]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_indices, val_indices = split_indices(len(dataset), val_pct=0.2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "48000 12000\n",
      "Sample val indices:  [18440 11347 30943 56540 34263 30533 42999  3012 45960 43659  5591 54299\n",
      " 13217 47201 17430 18758 50043 37672 37188 41691]\n"
     ]
    }
   ],
   "source": [
    "print(len(train_indices), len(val_indices))\n",
    "print('Sample val indices: ', val_indices[:20])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "from torch.utils.data.sampler import SubsetRandomSampler\n",
    "from torch.utils.data.dataloader import DataLoader"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "batch_size=100\n",
    "\n",
    "# Training sampler and data loader\n",
    "train_sampler = SubsetRandomSampler(train_indices)\n",
    "train_loader = DataLoader(dataset, \n",
    "                          batch_size, \n",
    "                          sampler=train_sampler)\n",
    "\n",
    "# Validation sampler and data loader\n",
    "val_sampler = SubsetRandomSampler(val_indices)\n",
    "val_loader = DataLoader(dataset,\n",
    "                        batch_size, \n",
    "                        sampler=val_sampler)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 模型\n",
    "import torch.nn as nn"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "input_size = 28*28\n",
    "num_classes = 10\n",
    "# 逻辑回归模型\n",
    "model = nn.Linear(input_size,num_classes)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "torch.Size([10, 784])\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "Parameter containing:\n",
       "tensor([[ 0.0259,  0.0195, -0.0271,  ...,  0.0318, -0.0343, -0.0080],\n",
       "        [-0.0091,  0.0149,  0.0291,  ..., -0.0253,  0.0189, -0.0234],\n",
       "        [-0.0185, -0.0136, -0.0034,  ...,  0.0228, -0.0335, -0.0210],\n",
       "        ...,\n",
       "        [ 0.0348,  0.0333, -0.0329,  ..., -0.0343, -0.0147,  0.0090],\n",
       "        [ 0.0244, -0.0094, -0.0194,  ...,  0.0144, -0.0071,  0.0046],\n",
       "        [-0.0004, -0.0088,  0.0307,  ..., -0.0025,  0.0252, -0.0095]],\n",
       "       requires_grad=True)"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(model.weight.shape)\n",
    "model.weight"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "torch.Size([10])\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "Parameter containing:\n",
       "tensor([-0.0106, -0.0085, -0.0010,  0.0169,  0.0084,  0.0105, -0.0253, -0.0155,\n",
       "        -0.0034, -0.0231], requires_grad=True)"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(model.bias.shape)\n",
    "model.bias"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "# reshape\n",
    "class MnistModel(nn.Module):\n",
    "    def __init__(self):\n",
    "        super().__init__()\n",
    "        self.linear = nn.Linear(input_size,num_classes)\n",
    "    def forward(self,xb):\n",
    "        xb = xb.reshape(-1,784)\n",
    "        out = self.linear(xb)\n",
    "        return out\n",
    "model = MnistModel()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "torch.Size([10, 784]) torch.Size([10])\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[Parameter containing:\n",
       " tensor([[-0.0295, -0.0256, -0.0008,  ..., -0.0099,  0.0066, -0.0283],\n",
       "         [ 0.0247, -0.0237,  0.0022,  ..., -0.0187,  0.0031, -0.0152],\n",
       "         [-0.0068, -0.0222,  0.0136,  ...,  0.0223,  0.0204,  0.0235],\n",
       "         ...,\n",
       "         [-0.0268,  0.0287, -0.0179,  ...,  0.0133, -0.0013, -0.0319],\n",
       "         [ 0.0247,  0.0338,  0.0251,  ...,  0.0099,  0.0102,  0.0308],\n",
       "         [ 0.0128,  0.0104, -0.0193,  ..., -0.0063,  0.0169, -0.0017]],\n",
       "        requires_grad=True), Parameter containing:\n",
       " tensor([ 0.0187, -0.0189,  0.0176, -0.0108,  0.0257,  0.0147, -0.0167,  0.0319,\n",
       "         -0.0262, -0.0069], requires_grad=True)]"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(model.linear.weight.shape, model.linear.bias.shape)\n",
    "list(model.parameters()) # 返回包含权重和偏差的列表"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "outputs.shape :  torch.Size([100, 10])\n",
      "Sample outputs :\n",
      " tensor([[ 0.1846, -0.0555,  0.1163,  0.0377, -0.2811,  0.0452,  0.0603,  0.0935,\n",
      "          0.1397, -0.0840],\n",
      "        [-0.1666,  0.0475,  0.4868, -0.2980, -0.0412,  0.0668, -0.0506, -0.1591,\n",
      "         -0.0080,  0.3366]])\n"
     ]
    }
   ],
   "source": [
    "for images, labels in train_loader:\n",
    "    outputs = model(images)\n",
    "    break\n",
    "\n",
    "print('outputs.shape : ', outputs.shape)\n",
    "print('Sample outputs :\\n', outputs[:2].data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "outputs.shape :  torch.Size([100, 10])\n",
      "Sample outputs :\n",
      " tensor([[-0.0859, -0.0300,  0.1691, -0.1258, -0.2122,  0.1627, -0.2197, -0.1946,\n",
      "          0.1369,  0.2111],\n",
      "        [ 0.1207,  0.2300,  0.4857, -0.0645,  0.2487, -0.0302,  0.1075,  0.1674,\n",
      "          0.0404,  0.1104]])\n"
     ]
    }
   ],
   "source": [
    "for images, labels in train_loader:\n",
    "    outputs = model(images)\n",
    "    break\n",
    "\n",
    "print('outputs.shape : ', outputs.shape)\n",
    "print('Sample outputs :\\n', outputs[:2].data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch.nn.functional as F"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Sample probabilities:\n",
      " tensor([[0.0923, 0.0976, 0.1190, 0.0886, 0.0813, 0.1183, 0.0807, 0.0828, 0.1153,\n",
      "         0.1242],\n",
      "        [0.0968, 0.1080, 0.1394, 0.0804, 0.1100, 0.0832, 0.0955, 0.1014, 0.0893,\n",
      "         0.0958]])\n",
      "Sum:  1.0\n"
     ]
    }
   ],
   "source": [
    "# Apply softmax for each output row\n",
    "probs = F.softmax(outputs, dim=1)\n",
    "\n",
    "# Look at sample probabilities\n",
    "print(\"Sample probabilities:\\n\", probs[:2].data)\n",
    "\n",
    "# Add up the probabilities of an output row\n",
    "print(\"Sum: \", torch.sum(probs[0]).item())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor([9, 2, 0, 9, 8, 2, 2, 2, 2, 5, 2, 5, 2, 3, 2, 9, 0, 2, 1, 5, 2, 2, 9, 3,\n",
      "        0, 2, 2, 9, 0, 2, 2, 5, 2, 2, 2, 9, 3, 9, 2, 9, 9, 2, 7, 8, 2, 2, 2, 9,\n",
      "        2, 2, 6, 2, 9, 2, 2, 0, 1, 2, 2, 1, 1, 2, 0, 2, 2, 8, 6, 2, 2, 2, 5, 9,\n",
      "        9, 8, 1, 2, 8, 3, 2, 0, 2, 2, 2, 3, 9, 2, 7, 2, 6, 2, 2, 2, 8, 9, 2, 9,\n",
      "        2, 6, 9, 9])\n"
     ]
    }
   ],
   "source": [
    "max_probs,preds = torch.max(probs,dim=1)\n",
    "print(preds)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([9, 2, 2, 7, 9, 4, 6, 6, 9, 4, 5, 4, 4, 8, 2, 1, 1, 1, 9, 7, 1, 1, 8, 5,\n",
       "        5, 2, 8, 4, 3, 0, 1, 8, 1, 5, 2, 4, 9, 7, 3, 6, 1, 6, 5, 7, 6, 8, 1, 1,\n",
       "        2, 0, 0, 8, 1, 0, 1, 2, 8, 2, 4, 5, 5, 1, 3, 1, 7, 8, 9, 6, 1, 2, 7, 4,\n",
       "        5, 7, 3, 2, 5, 4, 1, 3, 4, 6, 3, 4, 8, 5, 0, 8, 7, 1, 9, 6, 4, 1, 4, 4,\n",
       "        4, 0, 1, 8])"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 评估度量和损失函数\n",
    "def accuracy(l1,l2):\n",
    "    return torch.sum(l1 == l2).item() / len(l1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.1"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "accuracy(preds,labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "loss_fn = F.cross_entropy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor(2.3327, grad_fn=<NllLossBackward>)\n"
     ]
    }
   ],
   "source": [
    "loss = loss_fn(outputs,labels)\n",
    "print(loss)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 优化\n",
    "learning_rate = 0.001\n",
    "optimizer = torch.optim.SGD(model.parameters(),lr=learning_rate)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 训练模型\n",
    "def loss_batch(model,loss_func,xb,yb,opt=None,metric=None):\n",
    "    # 计算损失\n",
    "    preds = model(xb)\n",
    "    loss = loss_func(preds,yb)\n",
    "    if opt is not None:\n",
    "        # 计算梯度\n",
    "        loss.backward()\n",
    "        # 更新参数\n",
    "        opt.step()\n",
    "        # 重置梯度\n",
    "        opt.zero_grad()\n",
    "    metric_result = None\n",
    "    if metric is not None:\n",
    "        # 计算矩阵\n",
    "        metric_result = metric(preds,yb)\n",
    "    return loss.item(),len(xb),metric_result"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [],
   "source": [
    "#  计算验证集的总体损失\n",
    "def evaluate(model, loss_fn, valid_dl, metric=None):\n",
    "    with torch.no_grad():\n",
    "        # Pass each batch through the model\n",
    "        results = [loss_batch(model, loss_fn, xb, yb, metric=metric)\n",
    "                   for xb,yb in valid_dl]\n",
    "        # Separate losses, counts and metrics\n",
    "        losses, nums, metrics = zip(*results)\n",
    "        # Total size of the dataset\n",
    "        total = np.sum(nums)\n",
    "        # Avg. loss across batches \n",
    "        avg_loss = np.sum(np.multiply(losses, nums)) / total\n",
    "        avg_metric = None\n",
    "        if metric is not None:\n",
    "            # Avg. of metric across batches\n",
    "            avg_metric = np.sum(np.multiply(metrics, nums)) / total\n",
    "    return avg_loss, total, avg_metric"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [],
   "source": [
    "def accuracy(outputs, labels):\n",
    "    _, preds = torch.max(outputs, dim=1)\n",
    "    return torch.sum(preds == labels).item() / len(preds)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loss: 2.3186, Accuracy: 0.1120\n"
     ]
    }
   ],
   "source": [
    "val_loss, total, val_acc = evaluate(model, loss_fn, val_loader, metric=accuracy)\n",
    "print('Loss: {:.4f}, Accuracy: {:.4f}'.format(val_loss, val_acc))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [],
   "source": [
    "def fit(epochs, model, loss_fn, opt, train_dl, valid_dl, metric=None):\n",
    "    for epoch in range(epochs):\n",
    "        # Training\n",
    "        for xb,yb in train_dl:\n",
    "            loss,_,_ = loss_batch(model, loss_fn, xb, yb, opt)\n",
    "\n",
    "        # Evaluation\n",
    "        result = evaluate(model, loss_fn, valid_dl, metric)\n",
    "        val_loss, total, val_metric = result\n",
    "        \n",
    "        # Print progress\n",
    "        if metric is None:\n",
    "            print('Epoch [{}/{}], Loss: {:.4f}'\n",
    "                  .format(epoch+1, epochs, val_loss))\n",
    "        else:\n",
    "            print('Epoch [{}/{}], Loss: {:.4f}, {}: {:.4f}'\n",
    "                  .format(epoch+1, epochs, val_loss, metric.__name__, val_metric))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Redifine model and optimizer\n",
    "model = MnistModel()\n",
    "optimizer = torch.optim.SGD(model.parameters(), lr=learning_rate)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [1/5], Loss: 1.8831, accuracy: 0.6896\n",
      "Epoch [2/5], Loss: 1.5811, accuracy: 0.7637\n",
      "Epoch [3/5], Loss: 1.3696, accuracy: 0.7887\n",
      "Epoch [4/5], Loss: 1.2181, accuracy: 0.8004\n",
      "Epoch [5/5], Loss: 1.1062, accuracy: 0.8117\n"
     ]
    }
   ],
   "source": [
    "fit(5, model, F.cross_entropy, optimizer, train_loader, val_loader, accuracy)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [1/5], Loss: 1.0209, accuracy: 0.8203\n",
      "Epoch [2/5], Loss: 0.9536, accuracy: 0.8255\n",
      "Epoch [3/5], Loss: 0.8994, accuracy: 0.8294\n",
      "Epoch [4/5], Loss: 0.8548, accuracy: 0.8328\n",
      "Epoch [5/5], Loss: 0.8175, accuracy: 0.8352\n"
     ]
    }
   ],
   "source": [
    "fit(5, model, F.cross_entropy, optimizer, train_loader, val_loader, accuracy)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [1/5], Loss: 0.7856, accuracy: 0.8377\n",
      "Epoch [2/5], Loss: 0.7582, accuracy: 0.8401\n",
      "Epoch [3/5], Loss: 0.7343, accuracy: 0.8423\n",
      "Epoch [4/5], Loss: 0.7131, accuracy: 0.8450\n",
      "Epoch [5/5], Loss: 0.6944, accuracy: 0.8475\n"
     ]
    }
   ],
   "source": [
    "fit(5, model, F.cross_entropy, optimizer, train_loader, val_loader, accuracy)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [1/5], Loss: 0.6777, accuracy: 0.8488\n",
      "Epoch [2/5], Loss: 0.6626, accuracy: 0.8507\n",
      "Epoch [3/5], Loss: 0.6489, accuracy: 0.8531\n",
      "Epoch [4/5], Loss: 0.6364, accuracy: 0.8541\n",
      "Epoch [5/5], Loss: 0.6250, accuracy: 0.8552\n"
     ]
    }
   ],
   "source": [
    "fit(5, model, F.cross_entropy, optimizer, train_loader, val_loader, accuracy)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Replace these values with your results\n",
    "accuracies = [0.1076, 0.6486, 0.7362, 0.7737, 0.7925, 0.8069, \n",
    "              0.8165, 0.8227, 0.8269, 0.8325, 0.8367,\n",
    "              0.8399, 0.8438, 0.8463, 0.8482, 0.8512,\n",
    "              0.8529, 0.8552, 0.8563, 0.8587, 0.8606]\n",
    "plt.plot(accuracies, '-x')\n",
    "plt.xlabel('epoch')\n",
    "plt.ylabel('accuracy')\n",
    "plt.title('Accuracy vs. No. of epochs');"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Define test dataset\n",
    "test_dataset = MNIST(root='../dataset/', \n",
    "                     train=False,\n",
    "                     transform=transforms.ToTensor())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape: torch.Size([1, 28, 28])\n",
      "Label: 7\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "img, label = test_dataset[0]\n",
    "plt.imshow(img[0], cmap='gray')\n",
    "print('Shape:', img.shape)\n",
    "print('Label:', label)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [],
   "source": [
    "def predict_image(img, model):\n",
    "    xb = img.unsqueeze(0)\n",
    "    yb = model(xb)\n",
    "    _, preds  = torch.max(yb, dim=1)\n",
    "    return preds[0].item()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Label: 7 , Predicted: 7\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "img, label = test_dataset[0]\n",
    "plt.imshow(img[0], cmap='gray')\n",
    "print('Label:', label, ', Predicted:', predict_image(img, model))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Label: 0 , Predicted: 0\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "img, label = test_dataset[10]\n",
    "plt.imshow(img[0], cmap='gray')\n",
    "print('Label:', label, ', Predicted:', predict_image(img, model))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Label: 9 , Predicted: 4\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "img, label = test_dataset[193]\n",
    "plt.imshow(img[0], cmap='gray')\n",
    "print('Label:', label, ', Predicted:', predict_image(img, model))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Label: 3 , Predicted: 5\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "img, label = test_dataset[1889]\n",
    "plt.imshow(img[0], cmap='gray')\n",
    "print('Label:', label, ', Predicted:', predict_image(img, model))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loss: 0.5947, Accuracy: 0.8682\n"
     ]
    }
   ],
   "source": [
    "test_loader = DataLoader(test_dataset, batch_size=200)\n",
    "\n",
    "test_loss, total, test_acc = evaluate(model, loss_fn, test_loader, metric=accuracy)\n",
    "print('Loss: {:.4f}, Accuracy: {:.4f}'.format(test_loss, test_acc))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 保存加载模型\n",
    "torch.save(model.state_dict(),'../dataset/mnist_logistic.pth')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "OrderedDict([('linear.weight',\n",
       "              tensor([[ 2.8604e-02,  3.0332e-02,  3.2241e-02,  ...,  2.7478e-02,\n",
       "                        4.4436e-03, -1.1815e-02],\n",
       "                      [ 2.1928e-02,  2.1416e-02,  3.0131e-03,  ..., -2.9282e-02,\n",
       "                        1.5006e-02,  2.1047e-02],\n",
       "                      [-2.8020e-02,  8.7929e-03,  3.2058e-02,  ...,  4.5490e-05,\n",
       "                       -4.3436e-03,  2.9610e-02],\n",
       "                      ...,\n",
       "                      [ 1.3719e-02, -2.9503e-02,  3.6410e-03,  ...,  1.0788e-02,\n",
       "                       -5.8315e-03,  2.0756e-02],\n",
       "                      [-7.5518e-03,  2.4706e-02,  8.5890e-03,  ..., -1.3711e-02,\n",
       "                        4.5113e-03,  2.5724e-03],\n",
       "                      [ 7.0339e-04, -2.4474e-02, -1.7951e-02,  ..., -2.4362e-02,\n",
       "                        3.4991e-02, -2.3155e-02]])),\n",
       "             ('linear.bias',\n",
       "              tensor([-0.0268,  0.0751, -0.0590, -0.0202,  0.0337,  0.0573, -0.0072,  0.0452,\n",
       "                      -0.1083, -0.0294]))])"
      ]
     },
     "execution_count": 65,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.state_dict()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "OrderedDict([('linear.weight',\n",
       "              tensor([[ 2.8604e-02,  3.0332e-02,  3.2241e-02,  ...,  2.7478e-02,\n",
       "                        4.4436e-03, -1.1815e-02],\n",
       "                      [ 2.1928e-02,  2.1416e-02,  3.0131e-03,  ..., -2.9282e-02,\n",
       "                        1.5006e-02,  2.1047e-02],\n",
       "                      [-2.8020e-02,  8.7929e-03,  3.2058e-02,  ...,  4.5490e-05,\n",
       "                       -4.3436e-03,  2.9610e-02],\n",
       "                      ...,\n",
       "                      [ 1.3719e-02, -2.9503e-02,  3.6410e-03,  ...,  1.0788e-02,\n",
       "                       -5.8315e-03,  2.0756e-02],\n",
       "                      [-7.5518e-03,  2.4706e-02,  8.5890e-03,  ..., -1.3711e-02,\n",
       "                        4.5113e-03,  2.5724e-03],\n",
       "                      [ 7.0339e-04, -2.4474e-02, -1.7951e-02,  ..., -2.4362e-02,\n",
       "                        3.4991e-02, -2.3155e-02]])),\n",
       "             ('linear.bias',\n",
       "              tensor([-0.0268,  0.0751, -0.0590, -0.0202,  0.0337,  0.0573, -0.0072,  0.0452,\n",
       "                      -0.1083, -0.0294]))])"
      ]
     },
     "execution_count": 66,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model2  = MnistModel()\n",
    "model2.load_state_dict(torch.load('../dataset/mnist_logistic.pth'))\n",
    "model2.state_dict()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loss: 0.5947, Accuracy: 0.8682\n"
     ]
    }
   ],
   "source": [
    "test_loss, total, test_acc = evaluate(model, loss_fn, test_loader, metric=accuracy)\n",
    "print('Loss: {:.4f}, Accuracy: {:.4f}'.format(test_loss, test_acc))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "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": "3.6.5"
  },
  "toc": {
   "base_numbering": 1,
   "nav_menu": {},
   "number_sections": true,
   "sideBar": true,
   "skip_h1_title": false,
   "title_cell": "Table of Contents",
   "title_sidebar": "Contents",
   "toc_cell": false,
   "toc_position": {},
   "toc_section_display": true,
   "toc_window_display": false
  },
  "varInspector": {
   "cols": {
    "lenName": 16,
    "lenType": 16,
    "lenVar": 40
   },
   "kernels_config": {
    "python": {
     "delete_cmd_postfix": "",
     "delete_cmd_prefix": "del ",
     "library": "var_list.py",
     "varRefreshCmd": "print(var_dic_list())"
    },
    "r": {
     "delete_cmd_postfix": ") ",
     "delete_cmd_prefix": "rm(",
     "library": "var_list.r",
     "varRefreshCmd": "cat(var_dic_list()) "
    }
   },
   "types_to_exclude": [
    "module",
    "function",
    "builtin_function_or_method",
    "instance",
    "_Feature"
   ],
   "window_display": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}