Lec5/Lab7_CIFAR-10_with_CNN.ipynb

{"nbformat":4,"nbformat_minor":0,"metadata":{"colab":{"name":"Lab7_CIFAR-10_with_CNN.ipynb","version":"0.3.2","provenance":[],"collapsed_sections":[]},"kernelspec":{"name":"python3","display_name":"Python 3"},"accelerator":"GPU"},"cells":[{"metadata":{"id":"LoWtdW8jFNgj","colab_type":"code","colab":{}},"cell_type":"code","source":["!mkdir results"],"execution_count":0,"outputs":[]},{"metadata":{"id":"cPWEXDj1GZwi","colab_type":"code","colab":{}},"cell_type":"code","source":["import torch\n","import torchvision\n","import torchvision.transforms as transforms\n","import torch.nn as nn\n","import torch.nn.functional as F\n","import torch.optim as optim\n","import argparse\n","import numpy as np\n","import time\n","from copy import deepcopy # Add Deepcopy for args\n","import seaborn as sns \n","import matplotlib.pyplot as plt"],"execution_count":0,"outputs":[]},{"metadata":{"id":"r9osn1HVGgxU","colab_type":"code","colab":{"base_uri":"https://localhost:8080/","height":73},"outputId":"23c51f9b-748f-443e-8eac-f6d76e1b9eef","executionInfo":{"status":"ok","timestamp":1548907905528,"user_tz":-540,"elapsed":34462,"user":{"displayName":"김승수","photoUrl":"","userId":"00478032631106904091"}}},"cell_type":"code","source":["transform = transforms.Compose(\n","    [transforms.ToTensor(),\n","     transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))])\n","\n","trainset = torchvision.datasets.CIFAR10(root='./data', train=True,\n","                                        download=True, transform=transform)\n","trainset, valset = torch.utils.data.random_split(trainset, [40000, 10000])\n","testset = torchvision.datasets.CIFAR10(root='./data', train=False,\n","                                       download=True, transform=transform)\n","partition = {'train': trainset, 'val':valset, 'test':testset}"],"execution_count":3,"outputs":[{"output_type":"stream","text":["Downloading https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz to ./data/cifar-10-python.tar.gz\n","Files already downloaded and verified\n"],"name":"stdout"}]},{"metadata":{"id":"5A2DCvUxGnsb","colab_type":"code","colab":{}},"cell_type":"code","source":["class MLP(nn.Module):\n","    def __init__(self, in_dim, out_dim, hid_dim, n_layer, act, dropout, use_bn, use_xavier):\n","        super(MLP, self).__init__()\n","        self.in_dim = in_dim\n","        self.out_dim = out_dim\n","        self.hid_dim = hid_dim\n","        self.n_layer = n_layer\n","        self.act = act\n","        self.dropout = dropout\n","        self.use_bn = use_bn\n","        self.use_xavier = use_xavier\n","        \n","        # ====== Create Linear Layers ====== #\n","        self.fc1 = nn.Linear(self.in_dim, self.hid_dim)\n","        \n","        self.linears = nn.ModuleList()\n","        self.bns = nn.ModuleList()\n","        for i in range(self.n_layer-1):\n","            self.linears.append(nn.Linear(self.hid_dim, self.hid_dim))\n","            if self.use_bn:\n","                self.bns.append(nn.BatchNorm1d(self.hid_dim))\n","                \n","        self.fc2 = nn.Linear(self.hid_dim, self.out_dim)\n","        \n","        # ====== Create Activation Function ====== #\n","        if self.act == 'relu':\n","            self.act = nn.ReLU()\n","        elif self.act == 'tanh':\n","            self.act == nn.Tanh()\n","        elif self.act == 'sigmoid':\n","            self.act = nn.Sigmoid()\n","        else:\n","            raise ValueError('no valid activation function selected!')\n","        \n","        # ====== Create Regularization Layer ======= #\n","        self.dropout = nn.Dropout(self.dropout)\n","        if self.use_xavier:\n","            self.xavier_init()\n","          \n","    def forward(self, x):\n","        x = self.act(self.fc1(x))\n","        for i in range(len(self.linears)):\n","            x = self.act(self.linears[i](x))\n","            if self.use_bn:\n","                x = self.bns[i](x)\n","            x = self.dropout(x)\n","        x = self.fc2(x)\n","        return x\n","    \n","    def xavier_init(self):\n","        for linear in self.linears:\n","            nn.init.xavier_normal_(linear.weight)\n","            linear.bias.data.fill_(0.01)"],"execution_count":0,"outputs":[]},{"metadata":{"id":"Xiyev8moQIEl","colab_type":"code","colab":{}},"cell_type":"code","source":["cfg = {\n","    'VGG11': [64, 'M', 128, 'M', 256, 256, 'M', 512, 512, 'M', 512, 512, 'M'],\n","    'VGG13': [64, 64, 'M', 128, 128, 'M', 256, 256, 'M', 512, 512, 'M', 512, 512, 'M'],\n","    'VGG16': [64, 64, 'M', 128, 128, 'M', 256, 256, 256, 'M', 512, 512, 512, 'M', 512, 512, 512, 'M'],\n","    'VGG19': [64, 64, 'M', 128, 128, 'M', 256, 256, 256, 256, 'M', 512, 512, 512, 512, 'M', 512, 512, 512, 512, 'M'],\n","}"],"execution_count":0,"outputs":[]},{"metadata":{"id":"US83BQgZQT5r","colab_type":"code","colab":{}},"cell_type":"code","source":["class CNN(nn.Module):\n","    \n","    def __init__(self, model_code, in_channels, out_dim, act, use_bn):\n","        super(CNN, self).__init__()\n","        \n","        if act == 'relu':\n","            self.act = nn.ReLU()\n","        elif act == 'sigmoid':\n","            self.act = nn.Sigmoid()\n","        elif act == 'tanh':\n","            self.act = nn.TanH()\n","        else:\n","            raise ValueError(\"Not a valid activation function code\")\n","        \n","        self.layers = self._make_layers(model_code, in_channels, use_bn)\n","        self.classifer = nn.Sequential(nn.Linear(512, 256),\n","                                       self.act,\n","                                       nn.Linear(256, out_dim))\n","        \n","    def forward(self, x):\n","        x = self.layers(x)\n","        x = x.view(x.size(0), -1)\n","        x = self.classifer(x)\n","        return x\n","        \n","    def _make_layers(self, model_code, in_channels, use_bn):\n","        layers = []\n","        for x in cfg[model_code]:\n","            if x == 'M':\n","                layers += [nn.MaxPool2d(kernel_size=2, stride=2)]\n","            else:\n","                layers += [nn.Conv2d(in_channels=in_channels,\n","                                     out_channels=x,\n","                                     kernel_size=3,\n","                                     stride=1,\n","                                     padding=1)]\n","                if use_bn:\n","                    layers += [nn.BatchNorm2d(x)]\n","                layers += [self.act]\n","                in_channels = x\n","        return nn.Sequential(*layers)"],"execution_count":0,"outputs":[]},{"metadata":{"id":"ryIjab0NI_JN","colab_type":"code","colab":{}},"cell_type":"code","source":["class CNN1(nn.Module):\n","    \n","    def __init__(self):\n","        super(CNN, self).__init__()\n","        \n","        self.conv1 = nn.Conv2d(in_channels = 3,\n","                               out_channels = 64,\n","                               kernel_size = 3,\n","                               stride = 1,\n","                               padding = 1)\n","        self.conv2 = nn.Conv2d(in_channels = 64,\n","                               out_channels = 256,\n","                               kernel_size = 5,\n","                               stride = 1,\n","                               padding = 2)\n","        self.act = nn.ReLU()\n","        self.maxpool1 = nn.MaxPool2d(kernel_size = 2,\n","                                     stride = 2)\n","        self.fc = nn.Linear(65536, 10)\n","\n","    def forward(self, x):\n","        x = self.conv1(x)\n","        x = self.act(x)\n","        x = self.conv2(x)\n","        x = self.act(x)\n","        x = self.maxpool1(x)\n","        x = x.view(x.size(0), -1)\n","        x = self.fc(x)\n","        return x"],"execution_count":0,"outputs":[]},{"metadata":{"id":"7bI6A9QxLjtJ","colab_type":"code","colab":{}},"cell_type":"code","source":["def dimension_check():\n","    net = CNN('VGG11', 3)\n","    x = torch.randn(2, 3, 32, 32)\n","    y = net(x)\n","    print(y.size())"],"execution_count":0,"outputs":[]},{"metadata":{"id":"u4Dbi5X-MDGJ","colab_type":"code","colab":{"base_uri":"https://localhost:8080/","height":35},"outputId":"2db0150b-a672-41ca-f87f-9e76b503969f","executionInfo":{"status":"ok","timestamp":1548911309395,"user_tz":-540,"elapsed":1109,"user":{"displayName":"김승수","photoUrl":"","userId":"00478032631106904091"}}},"cell_type":"code","source":["dimension_check()"],"execution_count":42,"outputs":[{"output_type":"stream","text":["torch.Size([2, 10])\n"],"name":"stdout"}]},{"metadata":{"id":"sz9_iCJWGzGH","colab_type":"code","colab":{}},"cell_type":"code","source":["def train(net, partition, optimizer, criterion, args):\n","    trainloader = torch.utils.data.DataLoader(partition['train'], \n","                                              batch_size=args.train_batch_size, \n","                                              shuffle=True, num_workers=2)\n","    net.train()\n","\n","    correct = 0\n","    total = 0\n","    train_loss = 0.0\n","    for i, data in enumerate(trainloader, 0):\n","        optimizer.zero_grad() # [21.01.05 오류 수정] 매 Epoch 마다 .zero_grad()가 실행되는 것을 매 iteration 마다 실행되도록 수정했습니다. \n","\n","        # get the inputs\n","        inputs, labels = data\n","        inputs = inputs.cuda()\n","        labels = labels.cuda()\n","        outputs = net(inputs)\n","\n","        loss = criterion(outputs, labels)\n","        loss.backward()\n","        optimizer.step()\n","\n","        train_loss += loss.item()\n","        _, predicted = torch.max(outputs.data, 1)\n","        total += labels.size(0)\n","        correct += (predicted == labels).sum().item()\n","\n","    train_loss = train_loss / len(trainloader)\n","    train_acc = 100 * correct / total\n","    return net, train_loss, train_acc"],"execution_count":0,"outputs":[]},{"metadata":{"id":"DsS6PQvKG17w","colab_type":"code","colab":{}},"cell_type":"code","source":["def validate(net, partition, criterion, args):\n","    valloader = torch.utils.data.DataLoader(partition['val'], \n","                                            batch_size=args.test_batch_size, \n","                                            shuffle=False, num_workers=2)\n","    net.eval()\n","\n","    correct = 0\n","    total = 0\n","    val_loss = 0 \n","    with torch.no_grad():\n","        for data in valloader:\n","            images, labels = data\n","            images = images.cuda()\n","            labels = labels.cuda()\n","            outputs = net(images)\n","\n","            loss = criterion(outputs, labels)\n","            \n","            val_loss += loss.item()\n","            _, predicted = torch.max(outputs.data, 1)\n","            total += labels.size(0)\n","            correct += (predicted == labels).sum().item()\n","\n","        val_loss = val_loss / len(valloader)\n","        val_acc = 100 * correct / total\n","    return val_loss, val_acc"],"execution_count":0,"outputs":[]},{"metadata":{"id":"Imj5-XqbG4tI","colab_type":"code","colab":{}},"cell_type":"code","source":["def test(net, partition, args):\n","    testloader = torch.utils.data.DataLoader(partition['test'], \n","                                             batch_size=args.test_batch_size, \n","                                             shuffle=False, num_workers=2)\n","    net.eval()\n","    \n","    correct = 0\n","    total = 0\n","    with torch.no_grad():\n","        for data in testloader:\n","            images, labels = data\n","            images = images.cuda()\n","            labels = labels.cuda()\n","\n","            outputs = net(images)\n","            _, predicted = torch.max(outputs.data, 1)\n","            total += labels.size(0)\n","            correct += (predicted == labels).sum().item()\n","\n","        test_acc = 100 * correct / total\n","    return test_acc"],"execution_count":0,"outputs":[]},{"metadata":{"id":"vu_vScpPG61S","colab_type":"code","colab":{}},"cell_type":"code","source":["def experiment(partition, args):\n","  \n","    net = CNN(model_code = args.model_code,\n","              in_channels = args.in_channels,\n","              out_dim = args.out_dim,\n","              act = args.act,\n","              use_bn = args.use_bn)\n","    net.cuda()\n","\n","    criterion = nn.CrossEntropyLoss()\n","    if args.optim == 'SGD':\n","        optimizer = optim.SGD(net.parameters(), lr=args.lr, weight_decay=args.l2)\n","    elif args.optim == 'RMSprop':\n","        optimizer = optim.RMSprop(net.parameters(), lr=args.lr, weight_decay=args.l2)\n","    elif args.optim == 'Adam':\n","        optimizer = optim.Adam(net.parameters(), lr=args.lr, weight_decay=args.l2)\n","    else:\n","        raise ValueError('In-valid optimizer choice')\n","    \n","    train_losses = []\n","    val_losses = []\n","    train_accs = []\n","    val_accs = []\n","        \n","    for epoch in range(args.epoch):  # loop over the dataset multiple times\n","        ts = time.time()\n","        net, train_loss, train_acc = train(net, partition, optimizer, criterion, args)\n","        val_loss, val_acc = validate(net, partition, criterion, args)\n","        te = time.time()\n","        \n","        train_losses.append(train_loss)\n","        val_losses.append(val_loss)\n","        train_accs.append(train_acc)\n","        val_accs.append(val_acc)\n","        \n","        print('Epoch {}, Acc(train/val): {:2.2f}/{:2.2f}, Loss(train/val) {:2.2f}/{:2.2f}. Took {:2.2f} sec'.format(epoch, train_acc, val_acc, train_loss, val_loss, te-ts))\n","        \n","    test_acc = test(net, partition, args)    \n","    \n","    result = {}\n","    result['train_losses'] = train_losses\n","    result['val_losses'] = val_losses\n","    result['train_accs'] = train_accs\n","    result['val_accs'] = val_accs\n","    result['train_acc'] = train_acc\n","    result['val_acc'] = val_acc\n","    result['test_acc'] = test_acc\n","    return vars(args), result"],"execution_count":0,"outputs":[]},{"metadata":{"id":"u6S8kDvhG9zk","colab_type":"code","colab":{}},"cell_type":"code","source":["import hashlib\n","import json\n","from os import listdir\n","from os.path import isfile, join\n","import pandas as pd\n","\n","def save_exp_result(setting, result):\n","    exp_name = setting['exp_name']\n","    del setting['epoch']\n","    del setting['test_batch_size']\n","\n","    hash_key = hashlib.sha1(str(setting).encode()).hexdigest()[:6]\n","    filename = './results/{}-{}.json'.format(exp_name, hash_key)\n","    result.update(setting)\n","    with open(filename, 'w') as f:\n","        json.dump(result, f)\n","\n","    \n","def load_exp_result(exp_name):\n","    dir_path = './results'\n","    filenames = [f for f in listdir(dir_path) if isfile(join(dir_path, f)) if '.json' in f]\n","    list_result = []\n","    for filename in filenames:\n","        if exp_name in filename:\n","            with open(join(dir_path, filename), 'r') as infile:\n","                results = json.load(infile)\n","                list_result.append(results)\n","    df = pd.DataFrame(list_result) # .drop(columns=[])\n","    return df"],"execution_count":0,"outputs":[]},{"metadata":{"id":"50Fc45ykHAnp","colab_type":"code","colab":{}},"cell_type":"code","source":["\n","def plot_acc(var1, var2, df):\n","\n","    fig, ax = plt.subplots(1, 3)\n","    fig.set_size_inches(15, 6)\n","    sns.set_style(\"darkgrid\", {\"axes.facecolor\": \".9\"})\n","\n","    sns.barplot(x=var1, y='train_acc', hue=var2, data=df, ax=ax[0])\n","    sns.barplot(x=var1, y='val_acc', hue=var2, data=df, ax=ax[1])\n","    sns.barplot(x=var1, y='test_acc', hue=var2, data=df, ax=ax[2])\n","    \n","    ax[0].set_title('Train Accuracy')\n","    ax[1].set_title('Validation Accuracy')\n","    ax[2].set_title('Test Accuracy')\n","\n","    \n","def plot_loss_variation(var1, var2, df, **kwargs):\n","\n","    list_v1 = df[var1].unique()\n","    list_v2 = df[var2].unique()\n","    list_data = []\n","\n","    for value1 in list_v1:\n","        for value2 in list_v2:\n","            row = df.loc[df[var1]==value1]\n","            row = row.loc[df[var2]==value2]\n","\n","            train_losses = list(row.train_losses)[0]\n","            val_losses = list(row.val_losses)[0]\n","\n","            for epoch, train_loss in enumerate(train_losses):\n","                list_data.append({'type':'train', 'loss':train_loss, 'epoch':epoch, var1:value1, var2:value2})\n","            for epoch, val_loss in enumerate(val_losses):\n","                list_data.append({'type':'val', 'loss':val_loss, 'epoch':epoch, var1:value1, var2:value2})\n","\n","    df = pd.DataFrame(list_data)\n","    g = sns.FacetGrid(df, row=var2, col=var1, hue='type', **kwargs)\n","    g = g.map(plt.plot, 'epoch', 'loss', marker='.')\n","    g.add_legend()\n","    g.fig.suptitle('Train loss vs Val loss')\n","    plt.subplots_adjust(top=0.89) # 만약 Title이 그래프랑 겹친다면 top 값을 조정해주면 됩니다! 함수 인자로 받으면 그래프마다 조절할 수 있겠죠?\n","\n","\n","def plot_acc_variation(var1, var2, df, **kwargs):\n","    list_v1 = df[var1].unique()\n","    list_v2 = df[var2].unique()\n","    list_data = []\n","\n","    for value1 in list_v1:\n","        for value2 in list_v2:\n","            row = df.loc[df[var1]==value1]\n","            row = row.loc[df[var2]==value2]\n","\n","            train_accs = list(row.train_accs)[0]\n","            val_accs = list(row.val_accs)[0]\n","            test_acc = list(row.test_acc)[0]\n","\n","            for epoch, train_acc in enumerate(train_accs):\n","                list_data.append({'type':'train', 'Acc':train_acc, 'test_acc':test_acc, 'epoch':epoch, var1:value1, var2:value2})\n","            for epoch, val_acc in enumerate(val_accs):\n","                list_data.append({'type':'val', 'Acc':val_acc, 'test_acc':test_acc, 'epoch':epoch, var1:value1, var2:value2})\n","\n","    df = pd.DataFrame(list_data)\n","    g = sns.FacetGrid(df, row=var2, col=var1, hue='type', **kwargs)\n","    g = g.map(plt.plot, 'epoch', 'Acc', marker='.')\n","\n","    def show_acc(x, y, metric, **kwargs):\n","        plt.scatter(x, y, alpha=0.3, s=1)\n","        metric = \"Test Acc: {:1.3f}\".format(list(metric.values)[0])\n","        plt.text(0.05, 0.95, metric,  horizontalalignment='left', verticalalignment='center', transform=plt.gca().transAxes, bbox=dict(facecolor='yellow', alpha=0.5, boxstyle=\"round,pad=0.1\"))\n","    g = g.map(show_acc, 'epoch', 'Acc', 'test_acc')\n","\n","    g.add_legend()\n","    g.fig.suptitle('Train Accuracy vs Val Accuracy')\n","    plt.subplots_adjust(top=0.89)"],"execution_count":0,"outputs":[]},{"metadata":{"id":"kmL0mGJpHEKh","colab_type":"code","colab":{"base_uri":"https://localhost:8080/","height":1073},"outputId":"65073ae2-632a-4a9a-dbe0-c13b35de5990","executionInfo":{"status":"error","timestamp":1548912831887,"user_tz":-540,"elapsed":993063,"user":{"displayName":"김승수","photoUrl":"","userId":"00478032631106904091"}}},"cell_type":"code","source":["# ====== Random Seed Initialization ====== #\n","seed = 123\n","np.random.seed(seed)\n","torch.manual_seed(seed)\n","\n","parser = argparse.ArgumentParser()\n","args = parser.parse_args(\"\")\n","args.exp_name = \"exp1_lr_model_code\"\n","\n","# ====== Model ====== #\n","args.model_code = 'VGG11'\n","args.in_channels = 3\n","args.out_dim = 10\n","args.act = 'relu'\n","\n","# ====== Regularization ======= #\n","args.l2 = 0.00001\n","args.use_bn = True\n","\n","# ====== Optimizer & Training ====== #\n","args.optim = 'RMSprop' #'RMSprop' #SGD, RMSprop, ADAM...\n","args.lr = 0.0015\n","args.epoch = 10\n","\n","args.train_batch_size = 256\n","args.test_batch_size = 1024\n","\n","# ====== Experiment Variable ====== #\n","name_var1 = 'lr'\n","name_var2 = 'model_code'\n","list_var1 = [0.0001, 0.00001]\n","list_var2 = ['VGG11', 'VGG13']\n","\n","\n","for var1 in list_var1:\n","    for var2 in list_var2:\n","        setattr(args, name_var1, var1)\n","        setattr(args, name_var2, var2)\n","        print(args)\n","                \n","        setting, result = experiment(partition, deepcopy(args))\n","        save_exp_result(setting, result)"],"execution_count":46,"outputs":[{"output_type":"stream","text":["Namespace(act='relu', epoch=10, exp_name='exp1_lr_model_code', in_channels=3, l2=1e-05, lr=0.0001, model_code='VGG11', optim='RMSprop', out_dim=10, test_batch_size=1024, train_batch_size=256, use_bn=True)\n","Epoch 0, Acc(train/val): 41.33/49.75, Loss(train/val) 1.50/1.39. Took 88.23 sec\n","Epoch 1, Acc(train/val): 57.09/59.87, Loss(train/val) 1.16/1.12. Took 88.28 sec\n","Epoch 2, Acc(train/val): 65.28/65.71, Loss(train/val) 0.95/0.98. Took 88.16 sec\n","Epoch 3, Acc(train/val): 69.96/66.86, Loss(train/val) 0.84/0.98. Took 88.32 sec\n","Epoch 4, Acc(train/val): 73.04/71.81, Loss(train/val) 0.76/0.82. Took 88.27 sec\n","Epoch 5, Acc(train/val): 76.50/72.27, Loss(train/val) 0.67/0.81. Took 88.32 sec\n","Epoch 6, Acc(train/val): 79.83/74.21, Loss(train/val) 0.58/0.74. Took 88.30 sec\n","Epoch 7, Acc(train/val): 82.42/74.69, Loss(train/val) 0.51/0.75. Took 88.19 sec\n","Epoch 8, Acc(train/val): 84.65/76.17, Loss(train/val) 0.44/0.72. Took 88.29 sec\n","Epoch 9, Acc(train/val): 86.77/73.72, Loss(train/val) 0.39/0.87. Took 88.27 sec\n","Namespace(act='relu', epoch=10, exp_name='exp1_lr_model_code', in_channels=3, l2=1e-05, lr=0.0001, model_code='VGG13', optim='RMSprop', out_dim=10, test_batch_size=1024, train_batch_size=256, use_bn=True)\n"],"name":"stdout"},{"output_type":"stream","text":["Traceback (most recent call last):\n","Traceback (most recent call last):\n","  File \"/usr/lib/python3.6/multiprocessing/queues.py\", line 240, in _feed\n","    send_bytes(obj)\n","  File \"/usr/lib/python3.6/multiprocessing/queues.py\", line 240, in _feed\n","    send_bytes(obj)\n","  File \"/usr/lib/python3.6/multiprocessing/connection.py\", line 200, in send_bytes\n","    self._send_bytes(m[offset:offset + size])\n","  File \"/usr/lib/python3.6/multiprocessing/connection.py\", line 200, in send_bytes\n","    self._send_bytes(m[offset:offset + size])\n","  File \"/usr/lib/python3.6/multiprocessing/connection.py\", line 404, in _send_bytes\n","    self._send(header + buf)\n","  File \"/usr/lib/python3.6/multiprocessing/connection.py\", line 368, in _send\n","    n = write(self._handle, buf)\n","BrokenPipeError: [Errno 32] Broken pipe\n","  File \"/usr/lib/python3.6/multiprocessing/connection.py\", line 404, in _send_bytes\n","    self._send(header + buf)\n","  File \"/usr/lib/python3.6/multiprocessing/connection.py\", line 368, in _send\n","    n = write(self._handle, buf)\n","BrokenPipeError: [Errno 32] Broken pipe\n"],"name":"stderr"},{"output_type":"error","ename":"KeyboardInterrupt","evalue":"ignored","traceback":["\u001b[0;31m---------------------------------------------------------------------------\u001b[0m","\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)","\u001b[0;32m<ipython-input-46-7c59ccb04616>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m     38\u001b[0m         \u001b[0mprint\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     39\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 40\u001b[0;31m         \u001b[0msetting\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mresult\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mexperiment\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mpartition\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdeepcopy\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     41\u001b[0m         \u001b[0msave_exp_result\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0msetting\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mresult\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n","\u001b[0;32m<ipython-input-45-8666d351e698>\u001b[0m in \u001b[0;36mexperiment\u001b[0;34m(partition, args)\u001b[0m\n\u001b[1;32m     26\u001b[0m         \u001b[0mts\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtime\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtime\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     27\u001b[0m         \u001b[0mnet\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtrain_loss\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtrain_acc\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtrain\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mnet\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mpartition\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0moptimizer\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcriterion\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0margs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 28\u001b[0;31m         \u001b[0mval_loss\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mval_acc\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mvalidate\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mnet\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mpartition\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcriterion\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0margs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     29\u001b[0m         \u001b[0mte\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtime\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtime\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     30\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n","\u001b[0;32m<ipython-input-17-1258369bc981>\u001b[0m in \u001b[0;36mvalidate\u001b[0;34m(net, partition, criterion, args)\u001b[0m\n\u001b[1;32m     17\u001b[0m             \u001b[0mloss\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcriterion\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0moutputs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlabels\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     18\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 19\u001b[0;31m             \u001b[0mval_loss\u001b[0m \u001b[0;34m+=\u001b[0m \u001b[0mloss\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mitem\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     20\u001b[0m             \u001b[0m_\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mpredicted\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtorch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmax\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0moutputs\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdata\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;36m1\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     21\u001b[0m             \u001b[0mtotal\u001b[0m \u001b[0;34m+=\u001b[0m \u001b[0mlabels\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msize\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n","\u001b[0;31mKeyboardInterrupt\u001b[0m: "]}]},{"metadata":{"id":"WTUdQrQYIgFp","colab_type":"code","colab":{"base_uri":"https://localhost:8080/","height":886},"outputId":"d43c983b-8609-487d-96c9-ae2fa4110fe4","executionInfo":{"status":"ok","timestamp":1548912900634,"user_tz":-540,"elapsed":2404,"user":{"displayName":"김승수","photoUrl":"","userId":"00478032631106904091"}}},"cell_type":"code","source":["var1 = 'lr'\n","var2 = 'model_code'\n","df = load_exp_result('exp1')\n","\n","plot_acc(var1, var2, df)\n","plot_loss_variation(var1, var2, df, sharey=False) #sharey를 True로 하면 모둔 subplot의 y축의 스케일이 같아집니다.\n","plot_acc_variation(var1, var2, df, margin_titles=True, sharey=True) #margin_titles를 True로 하면 그래프의 가장자리에 var1과 var2 값이 표시되고 False로 하면 각 subplot 위에 표시됩니다."],"execution_count":47,"outputs":[{"output_type":"stream","text":["/usr/local/lib/python3.6/dist-packages/seaborn/categorical.py:1468: FutureWarning: remove_na is deprecated and is a private function. Do not use.\n","  stat_data = remove_na(group_data[hue_mask])\n"],"name":"stderr"},{"output_type":"display_data","data":{"image/png":"iVBORw0KGgoAAAANSUhEUgAAA3MAAAGCCAYAAABKGJrvAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n/wV5SXXM7ol0rwAAAABJRU5ErkJggg==\n","text/plain":["<Figure size 1080x432 with 3 Axes>"]},"metadata":{"tags":[]}},{"output_type":"display_data","data":{"image/png":"iVBORw0KGgoAAAANSUhEUgAAAP8AAADVCAYAAACYPyaaAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRK5CYII=\n","text/plain":["<Figure size 271.375x216 with 1 Axes>"]},"metadata":{"tags":[]}},{"output_type":"display_data","data":{"image/png":"iVBORw0KGgoAAAANSUhEUgAAAQAAAADVCAYAAAC49i5OAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nBYCIxRQAIhZTAIhYTAEgYjEFgIjFvgAmbnQFNqT7KwAAAABJRU5ErkJggg==\n","text/plain":["<Figure size 271.375x216 with 1 Axes>"]},"metadata":{"tags":[]}}]},{"metadata":{"id":"RS7Tj2WuZs-r","colab_type":"text"},"cell_type":"markdown","source":[]}]}