Skip to content

Instantly share code, notes, and snippets.

@mcdlee

mcdlee/u_and_n.ipynb

Last active June 23, 2020 02:29
Show Gist options
  • Save mcdlee/24f691be13e003622e179e97470b6a51 to your computer and use it in GitHub Desktop.
Save mcdlee/24f691be13e003622e179e97470b6a51 to your computer and use it in GitHub Desktop.
Download ZIP
Using PyTorch to classify my writing
Display the source blob
Display the rendered blob
Raw
u_and_n.ipynb
{
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Prepare dependented modules"
]
},
{
"cell_type": "code",
"execution_count": 41,
"metadata": {},
"outputs": [],
"source": [
"import torch\n",
"import torchvision\n",
"from torchvision import transforms\n",
"from torchvision.datasets import folder\n",
"from torchvision.datasets import ImageFolder\n",
"\n",
"from torch.utils.data.sampler import SubsetRandomSampler\n",
"from torch.utils.data import DataLoader, Dataset\n",
"\n",
"# https://stackoverflow.com/questions/50544730/how-do-i-split-a-custom-dataset-into-training-and-test-datasets/50544887\n",
"\n",
"from PIL import Image\n",
"import numpy as np\n",
"\n",
"import matplotlib.pyplot as plt\n",
"import numpy as np\n",
"\n",
"import random"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Loading data"
]
},
{
"cell_type": "code",
"execution_count": 42,
"metadata": {
"scrolled": true
},
"outputs": [],
"source": [
"transform = transforms.Compose([\n",
" #transforms.Resize(32),\n",
" #transforms.RandomCrop(32),\n",
" transforms.Grayscale(),\n",
" transforms.ToTensor(),\n",
" transforms.Normalize(mean=[0.5], std=[0.5])\n",
" \n",
" \n",
"]) #need redefine\n",
"\n",
"dataset1 = ImageFolder(root=\"pics\", loader=Image.open, transform=transform, is_valid_file=None)\n",
"dataloader1 = DataLoader(dataset1, shuffle=True)"
]
},
{
"cell_type": "code",
"execution_count": 43,
"metadata": {
"scrolled": true
},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"tensor([1])\n"
]
},
{
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 432x288 with 1 Axes>"
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
}
],
"source": [
"import matplotlib.pyplot as plt\n",
"import numpy as np\n",
"\n",
"images, labels = next(iter(dataloader1))\n",
"\n",
"img = torchvision.utils.make_grid(images)\n",
"\n",
"img = img.numpy().transpose(1,2,0)\n",
"plt.imshow(img)\n",
"print(labels)"
]
},
{
"cell_type": "markdown",
"metadata": {
"scrolled": true
},
"source": [
"## Divide data into validation and training"
]
},
{
"cell_type": "code",
"execution_count": 44,
"metadata": {},
"outputs": [],
"source": [
"batch_size = 1\n",
"validation_split = 0.2\n",
"shuffle_dataset=True\n",
"random_seed = 42\n",
"\n",
"# Creating data indices for training and validation splits:\n",
"dataset_size = len(dataset1)\n",
"indices = list(range(dataset_size))\n",
"split = int(np.floor(validation_split * dataset_size))\n",
"if shuffle_dataset :\n",
" np.random.seed(random_seed)\n",
" np.random.shuffle(indices)\n",
"train_indices, val_indices = indices[split:], indices[:split]"
]
},
{
"cell_type": "code",
"execution_count": 45,
"metadata": {},
"outputs": [],
"source": [
"# Creating PT data samplers and loaders:\n",
"train_sampler = SubsetRandomSampler(train_indices)\n",
"valid_sampler = SubsetRandomSampler(val_indices)\n",
"\n",
"train_loader = DataLoader(dataset1, batch_size=batch_size, sampler=train_sampler)\n",
"validation_loader = DataLoader(dataset1, batch_size=batch_size, sampler=valid_sampler)\n",
"\n"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Neural network\n",
"\n",
"Needs revised"
]
},
{
"cell_type": "code",
"execution_count": 112,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Net(\n",
" (conv1): Conv2d(1, 8, kernel_size=(3, 3), stride=(1, 1))\n",
" (conv2): Conv2d(8, 16, kernel_size=(3, 3), stride=(1, 1))\n",
" (fc1): Linear(in_features=576, out_features=120, bias=True)\n",
" (fc2): Linear(in_features=120, out_features=84, bias=True)\n",
" (fc3): Linear(in_features=84, out_features=10, bias=True)\n",
")\n"
]
}
],
"source": [
"import torch.nn as nn\n",
"import torch.nn.functional as F\n",
"\n",
"\n",
"class Net(nn.Module):\n",
"\n",
" def __init__(self):\n",
" super(Net, self).__init__()\n",
" # 1 input image channel, 8 output channels, 3x3 square convolution\n",
" # 8 input image channel, 16 output channels, 3x3 square convolution\n",
" # kernel\n",
" self.conv1 = nn.Conv2d(1, 8, 3)\n",
" self.conv2 = nn.Conv2d(8, 16, 3)\n",
" # an affine operation: y = Wx + b\n",
" self.fc1 = nn.Linear(16 * 6 * 6, 120) # 8*8 from image dimension\n",
" self.fc2 = nn.Linear(120, 84)\n",
" self.fc3 = nn.Linear(84, 10)\n",
"\n",
" def forward(self, x):\n",
" # Max pooling over a 3 window\n",
" x = F.max_pool2d(F.relu(self.conv1(x)), 3)\n",
" # If the size is a square you can only specify a single number\n",
" x = F.max_pool2d(F.relu(self.conv2(x)), 3)\n",
" x = x.view(-1, self.num_flat_features(x))\n",
" x = F.relu(self.fc1(x))\n",
" x = F.relu(self.fc2(x))\n",
" x = self.fc3(x)\n",
" return x\n",
"\n",
" def num_flat_features(self, x):\n",
" size = x.size()[1:] # all dimensions except the batch dimension\n",
" num_features = 1\n",
" for s in size:\n",
" num_features *= s\n",
" return num_features\n",
"\n",
"net = Net()\n",
"\n",
"print(net)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Define the loss function"
]
},
{
"cell_type": "code",
"execution_count": 113,
"metadata": {},
"outputs": [],
"source": [
"import torch.optim as optim\n",
"\n",
"criterion = nn.CrossEntropyLoss()\n",
"optimizer = optim.SGD(net.parameters(), lr=0.001, momentum=0.9)\n"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# run the NN"
]
},
{
"cell_type": "code",
"execution_count": 114,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"[1, 5] loss: 11.885\n",
"[1, 10] loss: 11.738\n",
"[1, 15] loss: 11.571\n",
"[1, 20] loss: 11.431\n",
"[1, 25] loss: 11.251\n",
"[1, 30] loss: 11.071\n",
"[1, 35] loss: 10.882\n",
"[1, 40] loss: 10.654\n",
"[1, 45] loss: 10.405\n",
"[1, 50] loss: 10.294\n",
"[1, 55] loss: 9.884\n",
"[1, 60] loss: 9.470\n",
"[1, 65] loss: 8.777\n",
"[1, 70] loss: 8.308\n",
"[1, 75] loss: 6.988\n",
"[1, 80] loss: 5.738\n",
"[1, 85] loss: 4.446\n",
"[1, 90] loss: 3.870\n",
"[2, 5] loss: 3.854\n",
"[2, 10] loss: 4.141\n",
"[2, 15] loss: 1.752\n",
"[2, 20] loss: 6.756\n",
"[2, 25] loss: 5.239\n",
"[2, 30] loss: 3.273\n",
"[2, 35] loss: 3.087\n",
"[2, 40] loss: 3.929\n",
"[2, 45] loss: 3.635\n",
"[2, 50] loss: 3.068\n",
"[2, 55] loss: 3.646\n",
"[2, 60] loss: 4.594\n",
"[2, 65] loss: 3.646\n",
"[2, 70] loss: 4.740\n",
"[2, 75] loss: 3.822\n",
"[2, 80] loss: 4.925\n",
"[2, 85] loss: 4.571\n",
"[2, 90] loss: 3.963\n",
"[3, 5] loss: 4.280\n",
"[3, 10] loss: 7.055\n",
"[3, 15] loss: 3.700\n",
"[3, 20] loss: 4.063\n",
"[3, 25] loss: 2.963\n",
"[3, 30] loss: 3.846\n",
"[3, 35] loss: 3.824\n",
"[3, 40] loss: 4.866\n",
"[3, 45] loss: 3.598\n",
"[3, 50] loss: 5.940\n",
"[3, 55] loss: 3.682\n",
"[3, 60] loss: 3.890\n",
"[3, 65] loss: 3.953\n",
"[3, 70] loss: 3.646\n",
"[3, 75] loss: 4.281\n",
"[3, 80] loss: 3.698\n",
"[3, 85] loss: 3.065\n",
"[3, 90] loss: 3.433\n",
"[4, 5] loss: 5.099\n",
"[4, 10] loss: 3.608\n",
"[4, 15] loss: 3.884\n",
"[4, 20] loss: 3.593\n",
"[4, 25] loss: 4.469\n",
"[4, 30] loss: 3.501\n",
"[4, 35] loss: 3.852\n",
"[4, 40] loss: 4.218\n",
"[4, 45] loss: 3.669\n",
"[4, 50] loss: 4.060\n",
"[4, 55] loss: 3.512\n",
"[4, 60] loss: 3.647\n",
"[4, 65] loss: 3.318\n",
"[4, 70] loss: 3.464\n",
"[4, 75] loss: 3.479\n",
"[4, 80] loss: 3.629\n",
"[4, 85] loss: 3.791\n",
"[4, 90] loss: 4.038\n",
"[5, 5] loss: 3.322\n",
"[5, 10] loss: 2.766\n",
"[5, 15] loss: 5.321\n",
"[5, 20] loss: 2.646\n",
"[5, 25] loss: 4.818\n",
"[5, 30] loss: 3.285\n",
"[5, 35] loss: 2.678\n",
"[5, 40] loss: 3.702\n",
"[5, 45] loss: 6.211\n",
"[5, 50] loss: 3.003\n",
"[5, 55] loss: 3.861\n",
"[5, 60] loss: 3.638\n",
"[5, 65] loss: 3.539\n",
"[5, 70] loss: 3.584\n",
"[5, 75] loss: 4.055\n",
"[5, 80] loss: 3.454\n",
"[5, 85] loss: 3.753\n",
"[5, 90] loss: 2.761\n",
"[6, 5] loss: 4.442\n",
"[6, 10] loss: 2.773\n",
"[6, 15] loss: 4.294\n",
"[6, 20] loss: 3.704\n",
"[6, 25] loss: 3.513\n",
"[6, 30] loss: 3.595\n",
"[6, 35] loss: 3.763\n",
"[6, 40] loss: 4.233\n",
"[6, 45] loss: 3.576\n",
"[6, 50] loss: 3.629\n",
"[6, 55] loss: 3.539\n",
"[6, 60] loss: 3.610\n",
"[6, 65] loss: 3.445\n",
"[6, 70] loss: 4.340\n",
"[6, 75] loss: 3.023\n",
"[6, 80] loss: 5.142\n",
"[6, 85] loss: 6.200\n",
"[6, 90] loss: 3.565\n",
"[7, 5] loss: 3.633\n",
"[7, 10] loss: 4.141\n",
"[7, 15] loss: 3.744\n",
"[7, 20] loss: 3.584\n",
"[7, 25] loss: 2.809\n",
"[7, 30] loss: 4.300\n",
"[7, 35] loss: 3.059\n",
"[7, 40] loss: 4.005\n",
"[7, 45] loss: 4.066\n",
"[7, 50] loss: 3.290\n",
"[7, 55] loss: 4.582\n",
"[7, 60] loss: 2.515\n",
"[7, 65] loss: 3.433\n",
"[7, 70] loss: 2.682\n",
"[7, 75] loss: 4.523\n",
"[7, 80] loss: 3.480\n",
"[7, 85] loss: 3.805\n",
"[7, 90] loss: 3.673\n",
"[8, 5] loss: 3.465\n",
"[8, 10] loss: 4.290\n",
"[8, 15] loss: 3.908\n",
"[8, 20] loss: 3.665\n",
"[8, 25] loss: 3.752\n",
"[8, 30] loss: 3.565\n",
"[8, 35] loss: 3.472\n",
"[8, 40] loss: 3.734\n",
"[8, 45] loss: 3.157\n",
"[8, 50] loss: 3.972\n",
"[8, 55] loss: 3.842\n",
"[8, 60] loss: 3.856\n",
"[8, 65] loss: 3.501\n",
"[8, 70] loss: 3.475\n",
"[8, 75] loss: 2.879\n",
"[8, 80] loss: 3.569\n",
"[8, 85] loss: 4.152\n",
"[8, 90] loss: 3.695\n",
"[9, 5] loss: 3.966\n",
"[9, 10] loss: 3.910\n",
"[9, 15] loss: 3.235\n",
"[9, 20] loss: 3.197\n",
"[9, 25] loss: 3.596\n",
"[9, 30] loss: 4.902\n",
"[9, 35] loss: 3.885\n",
"[9, 40] loss: 3.470\n",
"[9, 45] loss: 3.816\n",
"[9, 50] loss: 3.341\n",
"[9, 55] loss: 3.688\n",
"[9, 60] loss: 4.639\n",
"[9, 65] loss: 3.599\n",
"[9, 70] loss: 4.665\n",
"[9, 75] loss: 4.425\n",
"[9, 80] loss: 3.653\n",
"[9, 85] loss: 3.450\n",
"[9, 90] loss: 4.434\n",
"[10, 5] loss: 4.171\n",
"[10, 10] loss: 3.671\n",
"[10, 15] loss: 3.471\n",
"[10, 20] loss: 3.463\n",
"[10, 25] loss: 3.451\n",
"[10, 30] loss: 4.068\n",
"[10, 35] loss: 3.848\n",
"[10, 40] loss: 3.670\n",
"[10, 45] loss: 4.000\n",
"[10, 50] loss: 3.830\n",
"[10, 55] loss: 2.912\n",
"[10, 60] loss: 3.831\n",
"[10, 65] loss: 2.994\n",
"[10, 70] loss: 2.935\n",
"[10, 75] loss: 3.654\n",
"[10, 80] loss: 5.434\n",
"[10, 85] loss: 3.813\n",
"[10, 90] loss: 3.248\n",
"[11, 5] loss: 3.673\n",
"[11, 10] loss: 2.719\n",
"[11, 15] loss: 2.666\n",
"[11, 20] loss: 4.148\n",
"[11, 25] loss: 4.925\n",
"[11, 30] loss: 2.919\n",
"[11, 35] loss: 3.079\n",
"[11, 40] loss: 4.460\n",
"[11, 45] loss: 3.745\n",
"[11, 50] loss: 3.510\n",
"[11, 55] loss: 3.623\n",
"[11, 60] loss: 3.521\n",
"[11, 65] loss: 3.445\n",
"[11, 70] loss: 3.119\n",
"[11, 75] loss: 2.639\n",
"[11, 80] loss: 4.095\n",
"[11, 85] loss: 2.531\n",
"[11, 90] loss: 3.832\n",
"[12, 5] loss: 3.467\n",
"[12, 10] loss: 3.822\n",
"[12, 15] loss: 3.664\n",
"[12, 20] loss: 3.504\n",
"[12, 25] loss: 4.447\n",
"[12, 30] loss: 3.928\n",
"[12, 35] loss: 3.238\n",
"[12, 40] loss: 4.633\n",
"[12, 45] loss: 4.705\n",
"[12, 50] loss: 5.176\n",
"[12, 55] loss: 3.700\n",
"[12, 60] loss: 3.727\n",
"[12, 65] loss: 3.655\n",
"[12, 70] loss: 3.898\n",
"[12, 75] loss: 3.815\n",
"[12, 80] loss: 3.651\n",
"[12, 85] loss: 3.506\n",
"[12, 90] loss: 3.467\n",
"[13, 5] loss: 3.727\n",
"[13, 10] loss: 3.775\n",
"[13, 15] loss: 3.543\n",
"[13, 20] loss: 3.593\n",
"[13, 25] loss: 3.673\n",
"[13, 30] loss: 3.563\n",
"[13, 35] loss: 3.433\n",
"[13, 40] loss: 3.417\n",
"[13, 45] loss: 2.840\n",
"[13, 50] loss: 4.294\n",
"[13, 55] loss: 3.416\n",
"[13, 60] loss: 4.316\n",
"[13, 65] loss: 3.437\n",
"[13, 70] loss: 3.650\n",
"[13, 75] loss: 3.508\n",
"[13, 80] loss: 3.438\n",
"[13, 85] loss: 3.695\n",
"[13, 90] loss: 3.174\n",
"[14, 5] loss: 3.633\n",
"[14, 10] loss: 3.423\n",
"[14, 15] loss: 2.771\n",
"[14, 20] loss: 4.652\n",
"[14, 25] loss: 3.521\n",
"[14, 30] loss: 3.472\n",
"[14, 35] loss: 3.773\n",
"[14, 40] loss: 3.472\n",
"[14, 45] loss: 3.600\n",
"[14, 50] loss: 3.615\n",
"[14, 55] loss: 3.530\n",
"[14, 60] loss: 4.005\n",
"[14, 65] loss: 2.415\n",
"[14, 70] loss: 4.050\n",
"[14, 75] loss: 3.925\n",
"[14, 80] loss: 4.031\n",
"[14, 85] loss: 3.204\n",
"[14, 90] loss: 3.464\n",
"[15, 5] loss: 3.901\n",
"[15, 10] loss: 5.788\n",
"[15, 15] loss: 4.069\n",
"[15, 20] loss: 3.639\n",
"[15, 25] loss: 3.209\n",
"[15, 30] loss: 3.823\n",
"[15, 35] loss: 3.490\n",
"[15, 40] loss: 3.477\n",
"[15, 45] loss: 3.566\n",
"[15, 50] loss: 4.852\n",
"[15, 55] loss: 3.700\n",
"[15, 60] loss: 3.450\n",
"[15, 65] loss: 4.195\n",
"[15, 70] loss: 3.424\n",
"[15, 75] loss: 3.164\n",
"[15, 80] loss: 3.910\n",
"[15, 85] loss: 3.416\n",
"[15, 90] loss: 3.404\n",
"[16, 5] loss: 3.796\n",
"[16, 10] loss: 3.452\n",
"[16, 15] loss: 3.530\n",
"[16, 20] loss: 3.504\n",
"[16, 25] loss: 3.347\n",
"[16, 30] loss: 3.754\n",
"[16, 35] loss: 3.691\n",
"[16, 40] loss: 3.341\n",
"[16, 45] loss: 3.315\n",
"[16, 50] loss: 4.522\n",
"[16, 55] loss: 2.948\n",
"[16, 60] loss: 3.599\n",
"[16, 65] loss: 2.986\n",
"[16, 70] loss: 3.783\n",
"[16, 75] loss: 3.333\n",
"[16, 80] loss: 3.818\n",
"[16, 85] loss: 3.924\n",
"[16, 90] loss: 3.163\n",
"[17, 5] loss: 2.670\n",
"[17, 10] loss: 3.881\n",
"[17, 15] loss: 2.562\n",
"[17, 20] loss: 3.879\n",
"[17, 25] loss: 3.578\n",
"[17, 30] loss: 3.475\n",
"[17, 35] loss: 4.896\n",
"[17, 40] loss: 3.849\n",
"[17, 45] loss: 3.022\n",
"[17, 50] loss: 3.465\n",
"[17, 55] loss: 3.528\n",
"[17, 60] loss: 4.407\n",
"[17, 65] loss: 3.422\n",
"[17, 70] loss: 2.280\n",
"[17, 75] loss: 4.066\n",
"[17, 80] loss: 3.417\n",
"[17, 85] loss: 4.268\n",
"[17, 90] loss: 3.447\n",
"[18, 5] loss: 3.215\n",
"[18, 10] loss: 4.441\n",
"[18, 15] loss: 3.463\n",
"[18, 20] loss: 3.333\n",
"[18, 25] loss: 3.252\n",
"[18, 30] loss: 3.782\n",
"[18, 35] loss: 3.387\n",
"[18, 40] loss: 2.694\n",
"[18, 45] loss: 4.918\n",
"[18, 50] loss: 3.861\n",
"[18, 55] loss: 3.417\n",
"[18, 60] loss: 3.686\n",
"[18, 65] loss: 3.150\n",
"[18, 70] loss: 3.096\n",
"[18, 75] loss: 4.325\n",
"[18, 80] loss: 3.559\n",
"[18, 85] loss: 3.210\n",
"[18, 90] loss: 3.515\n",
"[19, 5] loss: 3.634\n",
"[19, 10] loss: 3.305\n",
"[19, 15] loss: 4.360\n",
"[19, 20] loss: 3.458\n",
"[19, 25] loss: 2.803\n",
"[19, 30] loss: 3.382\n",
"[19, 35] loss: 2.959\n",
"[19, 40] loss: 3.332\n",
"[19, 45] loss: 3.199\n",
"[19, 50] loss: 2.527\n",
"[19, 55] loss: 3.065\n",
"[19, 60] loss: 3.467\n",
"[19, 65] loss: 2.701\n",
"[19, 70] loss: 2.881\n",
"[19, 75] loss: 2.503\n",
"[19, 80] loss: 3.360\n",
"[19, 85] loss: 3.134\n",
"[19, 90] loss: 3.703\n",
"[20, 5] loss: 1.662\n",
"[20, 10] loss: 2.492\n",
"[20, 15] loss: 2.734\n",
"[20, 20] loss: 2.559\n",
"[20, 25] loss: 2.153\n",
"[20, 30] loss: 1.706\n",
"[20, 35] loss: 1.811\n",
"[20, 40] loss: 2.031\n",
"[20, 45] loss: 2.914\n",
"[20, 50] loss: 1.196\n",
"[20, 55] loss: 1.422\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"[20, 60] loss: 2.780\n",
"[20, 65] loss: 1.334\n",
"[20, 70] loss: 1.905\n",
"[20, 75] loss: 2.814\n",
"[20, 80] loss: 0.926\n",
"[20, 85] loss: 0.957\n",
"[20, 90] loss: 0.521\n",
"[21, 5] loss: 0.789\n",
"[21, 10] loss: 0.444\n",
"[21, 15] loss: 0.273\n",
"[21, 20] loss: 0.677\n",
"[21, 25] loss: 0.575\n",
"[21, 30] loss: 0.106\n",
"[21, 35] loss: 1.909\n",
"[21, 40] loss: 0.726\n",
"[21, 45] loss: 1.585\n",
"[21, 50] loss: 2.381\n",
"[21, 55] loss: 0.941\n",
"[21, 60] loss: 0.181\n",
"[21, 65] loss: 0.256\n",
"[21, 70] loss: 0.511\n",
"[21, 75] loss: 0.078\n",
"[21, 80] loss: 0.201\n",
"[21, 85] loss: 0.841\n",
"[21, 90] loss: 0.057\n",
"[22, 5] loss: 0.059\n",
"[22, 10] loss: 0.064\n",
"[22, 15] loss: 2.874\n",
"[22, 20] loss: 0.843\n",
"[22, 25] loss: 0.312\n",
"[22, 30] loss: 0.071\n",
"[22, 35] loss: 0.319\n",
"[22, 40] loss: 0.156\n",
"[22, 45] loss: 0.175\n",
"[22, 50] loss: 0.192\n",
"[22, 55] loss: 0.054\n",
"[22, 60] loss: 0.545\n",
"[22, 65] loss: 3.544\n",
"[22, 70] loss: 0.655\n",
"[22, 75] loss: 0.219\n",
"[22, 80] loss: 0.140\n",
"[22, 85] loss: 0.052\n",
"[22, 90] loss: 0.144\n",
"[23, 5] loss: 0.104\n",
"[23, 10] loss: 1.556\n",
"[23, 15] loss: 0.469\n",
"[23, 20] loss: 0.188\n",
"[23, 25] loss: 0.232\n",
"[23, 30] loss: 0.086\n",
"[23, 35] loss: 1.009\n",
"[23, 40] loss: 0.510\n",
"[23, 45] loss: 0.228\n",
"[23, 50] loss: 0.064\n",
"[23, 55] loss: 0.652\n",
"[23, 60] loss: 3.653\n",
"[23, 65] loss: 0.042\n",
"[23, 70] loss: 0.034\n",
"[23, 75] loss: 0.137\n",
"[23, 80] loss: 0.159\n",
"[23, 85] loss: 0.069\n",
"[23, 90] loss: 0.438\n",
"[24, 5] loss: 1.107\n",
"[24, 10] loss: 1.414\n",
"[24, 15] loss: 0.592\n",
"[24, 20] loss: 0.061\n",
"[24, 25] loss: 0.045\n",
"[24, 30] loss: 0.060\n",
"[24, 35] loss: 0.066\n",
"[24, 40] loss: 1.211\n",
"[24, 45] loss: 0.031\n",
"[24, 50] loss: 0.728\n",
"[24, 55] loss: 0.041\n",
"[24, 60] loss: 0.028\n",
"[24, 65] loss: 0.212\n",
"[24, 70] loss: 0.163\n",
"[24, 75] loss: 0.920\n",
"[24, 80] loss: 1.959\n",
"[24, 85] loss: 0.246\n",
"[24, 90] loss: 0.133\n",
"[25, 5] loss: 0.087\n",
"[25, 10] loss: 0.183\n",
"[25, 15] loss: 0.804\n",
"[25, 20] loss: 0.047\n",
"[25, 25] loss: 0.052\n",
"[25, 30] loss: 0.054\n",
"[25, 35] loss: 0.271\n",
"[25, 40] loss: 0.025\n",
"[25, 45] loss: 0.076\n",
"[25, 50] loss: 0.024\n",
"[25, 55] loss: 0.048\n",
"[25, 60] loss: 0.875\n",
"[25, 65] loss: 0.094\n",
"[25, 70] loss: 0.308\n",
"[25, 75] loss: 0.114\n",
"[25, 80] loss: 0.537\n",
"[25, 85] loss: 0.007\n",
"[25, 90] loss: 0.021\n",
"[26, 5] loss: 0.966\n",
"[26, 10] loss: 0.105\n",
"[26, 15] loss: 0.345\n",
"[26, 20] loss: 0.122\n",
"[26, 25] loss: 0.028\n",
"[26, 30] loss: 0.597\n",
"[26, 35] loss: 0.018\n",
"[26, 40] loss: 0.021\n",
"[26, 45] loss: 0.050\n",
"[26, 50] loss: 0.803\n",
"[26, 55] loss: 0.109\n",
"[26, 60] loss: 0.421\n",
"[26, 65] loss: 0.035\n",
"[26, 70] loss: 0.109\n",
"[26, 75] loss: 0.034\n",
"[26, 80] loss: 0.019\n",
"[26, 85] loss: 0.030\n",
"[26, 90] loss: 0.011\n",
"[27, 5] loss: 0.020\n",
"[27, 10] loss: 0.008\n",
"[27, 15] loss: 0.010\n",
"[27, 20] loss: 0.248\n",
"[27, 25] loss: 0.018\n",
"[27, 30] loss: 0.005\n",
"[27, 35] loss: 0.019\n",
"[27, 40] loss: 0.015\n",
"[27, 45] loss: 0.023\n",
"[27, 50] loss: 0.029\n",
"[27, 55] loss: 0.013\n",
"[27, 60] loss: 0.024\n",
"[27, 65] loss: 0.083\n",
"[27, 70] loss: 0.008\n",
"[27, 75] loss: 0.328\n",
"[27, 80] loss: 0.003\n",
"[27, 85] loss: 1.344\n",
"[27, 90] loss: 0.101\n",
"[28, 5] loss: 0.340\n",
"[28, 10] loss: 0.554\n",
"[28, 15] loss: 0.034\n",
"[28, 20] loss: 0.031\n",
"[28, 25] loss: 0.007\n",
"[28, 30] loss: 0.018\n",
"[28, 35] loss: 0.014\n",
"[28, 40] loss: 0.058\n",
"[28, 45] loss: 0.006\n",
"[28, 50] loss: 0.004\n",
"[28, 55] loss: 0.059\n",
"[28, 60] loss: 0.021\n",
"[28, 65] loss: 0.004\n",
"[28, 70] loss: 0.004\n",
"[28, 75] loss: 0.211\n",
"[28, 80] loss: 1.346\n",
"[28, 85] loss: 0.334\n",
"[28, 90] loss: 0.004\n",
"[29, 5] loss: 0.028\n",
"[29, 10] loss: 0.003\n",
"[29, 15] loss: 0.014\n",
"[29, 20] loss: 0.195\n",
"[29, 25] loss: 1.267\n",
"[29, 30] loss: 0.015\n",
"[29, 35] loss: 0.027\n",
"[29, 40] loss: 0.018\n",
"[29, 45] loss: 0.010\n",
"[29, 50] loss: 1.731\n",
"[29, 55] loss: 0.348\n",
"[29, 60] loss: 0.005\n",
"[29, 65] loss: 0.010\n",
"[29, 70] loss: 0.100\n",
"[29, 75] loss: 0.061\n",
"[29, 80] loss: 0.049\n",
"[29, 85] loss: 1.052\n",
"[29, 90] loss: 0.105\n",
"[30, 5] loss: 0.022\n",
"[30, 10] loss: 0.045\n",
"[30, 15] loss: 0.283\n",
"[30, 20] loss: 0.321\n",
"[30, 25] loss: 0.463\n",
"[30, 30] loss: 0.050\n",
"[30, 35] loss: 0.040\n",
"[30, 40] loss: 0.007\n",
"[30, 45] loss: 0.024\n",
"[30, 50] loss: 0.016\n",
"[30, 55] loss: 0.073\n",
"[30, 60] loss: 0.037\n",
"[30, 65] loss: 0.009\n",
"[30, 70] loss: 0.017\n",
"[30, 75] loss: 0.041\n",
"[30, 80] loss: 0.136\n",
"[30, 85] loss: 0.002\n",
"[30, 90] loss: 0.009\n",
"[31, 5] loss: 0.066\n",
"[31, 10] loss: 0.007\n",
"[31, 15] loss: 0.038\n",
"[31, 20] loss: 0.024\n",
"[31, 25] loss: 0.043\n",
"[31, 30] loss: 0.093\n",
"[31, 35] loss: 0.018\n",
"[31, 40] loss: 0.003\n",
"[31, 45] loss: 0.053\n",
"[31, 50] loss: 0.005\n",
"[31, 55] loss: 0.125\n",
"[31, 60] loss: 0.007\n",
"[31, 65] loss: 0.005\n",
"[31, 70] loss: 0.036\n",
"[31, 75] loss: 0.002\n",
"[31, 80] loss: 0.070\n",
"[31, 85] loss: 0.004\n",
"[31, 90] loss: 0.004\n",
"[32, 5] loss: 0.101\n",
"[32, 10] loss: 0.003\n",
"[32, 15] loss: 0.009\n",
"[32, 20] loss: 0.047\n",
"[32, 25] loss: 0.006\n",
"[32, 30] loss: 0.004\n",
"[32, 35] loss: 0.004\n",
"[32, 40] loss: 0.061\n",
"[32, 45] loss: 0.005\n",
"[32, 50] loss: 0.068\n",
"[32, 55] loss: 0.003\n",
"[32, 60] loss: 0.002\n",
"[32, 65] loss: 0.000\n",
"[32, 70] loss: 0.054\n",
"[32, 75] loss: 0.059\n",
"[32, 80] loss: 0.008\n",
"[32, 85] loss: 0.002\n",
"[32, 90] loss: 0.014\n",
"[33, 5] loss: 0.075\n",
"[33, 10] loss: 0.002\n",
"[33, 15] loss: 0.003\n",
"[33, 20] loss: 0.002\n",
"[33, 25] loss: 0.014\n",
"[33, 30] loss: 0.025\n",
"[33, 35] loss: 0.011\n",
"[33, 40] loss: 0.002\n",
"[33, 45] loss: 0.058\n",
"[33, 50] loss: 0.003\n",
"[33, 55] loss: 0.062\n",
"[33, 60] loss: 0.052\n",
"[33, 65] loss: 0.001\n",
"[33, 70] loss: 0.007\n",
"[33, 75] loss: 0.002\n",
"[33, 80] loss: 0.004\n",
"[33, 85] loss: 0.006\n",
"[33, 90] loss: 0.036\n",
"[34, 5] loss: 0.006\n",
"[34, 10] loss: 0.001\n",
"[34, 15] loss: 0.056\n",
"[34, 20] loss: 0.001\n",
"[34, 25] loss: 0.021\n",
"[34, 30] loss: 0.019\n",
"[34, 35] loss: 0.000\n",
"[34, 40] loss: 0.021\n",
"[34, 45] loss: 0.070\n",
"[34, 50] loss: 0.007\n",
"[34, 55] loss: 0.008\n",
"[34, 60] loss: 0.002\n",
"[34, 65] loss: 0.006\n",
"[34, 70] loss: 0.001\n",
"[34, 75] loss: 0.003\n",
"[34, 80] loss: 0.004\n",
"[34, 85] loss: 0.059\n",
"[34, 90] loss: 0.024\n",
"[35, 5] loss: 0.001\n",
"[35, 10] loss: 0.001\n",
"[35, 15] loss: 0.011\n",
"[35, 20] loss: 0.001\n",
"[35, 25] loss: 0.014\n",
"[35, 30] loss: 0.034\n",
"[35, 35] loss: 0.024\n",
"[35, 40] loss: 0.002\n",
"[35, 45] loss: 0.053\n",
"[35, 50] loss: 0.002\n",
"[35, 55] loss: 0.001\n",
"[35, 60] loss: 0.031\n",
"[35, 65] loss: 0.036\n",
"[35, 70] loss: 0.004\n",
"[35, 75] loss: 0.000\n",
"[35, 80] loss: 0.049\n",
"[35, 85] loss: 0.003\n",
"[35, 90] loss: 0.001\n",
"[36, 5] loss: 0.002\n",
"[36, 10] loss: 0.020\n",
"[36, 15] loss: 0.038\n",
"[36, 20] loss: 0.001\n",
"[36, 25] loss: 0.031\n",
"[36, 30] loss: 0.044\n",
"[36, 35] loss: 0.003\n",
"[36, 40] loss: 0.033\n",
"[36, 45] loss: 0.001\n",
"[36, 50] loss: 0.003\n",
"[36, 55] loss: 0.015\n",
"[36, 60] loss: 0.000\n",
"[36, 65] loss: 0.006\n",
"[36, 70] loss: 0.008\n",
"[36, 75] loss: 0.001\n",
"[36, 80] loss: 0.028\n",
"[36, 85] loss: 0.001\n",
"[36, 90] loss: 0.000\n",
"[37, 5] loss: 0.002\n",
"[37, 10] loss: 0.007\n",
"[37, 15] loss: 0.039\n",
"[37, 20] loss: 0.001\n",
"[37, 25] loss: 0.038\n",
"[37, 30] loss: 0.000\n",
"[37, 35] loss: 0.002\n",
"[37, 40] loss: 0.000\n",
"[37, 45] loss: 0.005\n",
"[37, 50] loss: 0.028\n",
"[37, 55] loss: 0.001\n",
"[37, 60] loss: 0.006\n",
"[37, 65] loss: 0.026\n",
"[37, 70] loss: 0.018\n",
"[37, 75] loss: 0.000\n",
"[37, 80] loss: 0.036\n",
"[37, 85] loss: 0.000\n",
"[37, 90] loss: 0.003\n",
"[38, 5] loss: 0.006\n",
"[38, 10] loss: 0.003\n",
"[38, 15] loss: 0.001\n",
"[38, 20] loss: 0.000\n",
"[38, 25] loss: 0.002\n",
"[38, 30] loss: 0.048\n",
"[38, 35] loss: 0.021\n",
"[38, 40] loss: 0.009\n",
"[38, 45] loss: 0.016\n",
"[38, 50] loss: 0.018\n",
"[38, 55] loss: 0.004\n",
"[38, 60] loss: 0.001\n",
"[38, 65] loss: 0.001\n",
"[38, 70] loss: 0.000\n",
"[38, 75] loss: 0.018\n",
"[38, 80] loss: 0.035\n",
"[38, 85] loss: 0.000\n",
"[38, 90] loss: 0.000\n",
"[39, 5] loss: 0.020\n",
"[39, 10] loss: 0.002\n",
"[39, 15] loss: 0.001\n",
"[39, 20] loss: 0.000\n",
"[39, 25] loss: 0.017\n",
"[39, 30] loss: 0.033\n",
"[39, 35] loss: 0.027\n",
"[39, 40] loss: 0.005\n",
"[39, 45] loss: 0.021\n",
"[39, 50] loss: 0.000\n",
"[39, 55] loss: 0.000\n",
"[39, 60] loss: 0.000\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"[39, 65] loss: 0.004\n",
"[39, 70] loss: 0.014\n",
"[39, 75] loss: 0.024\n",
"[39, 80] loss: 0.001\n",
"[39, 85] loss: 0.001\n",
"[39, 90] loss: 0.000\n",
"[40, 5] loss: 0.000\n",
"[40, 10] loss: 0.001\n",
"[40, 15] loss: 0.031\n",
"[40, 20] loss: 0.006\n",
"[40, 25] loss: 0.001\n",
"[40, 30] loss: 0.004\n",
"[40, 35] loss: 0.048\n",
"[40, 40] loss: 0.000\n",
"[40, 45] loss: 0.000\n",
"[40, 50] loss: 0.000\n",
"[40, 55] loss: 0.020\n",
"[40, 60] loss: 0.000\n",
"[40, 65] loss: 0.000\n",
"[40, 70] loss: 0.000\n",
"[40, 75] loss: 0.005\n",
"[40, 80] loss: 0.016\n",
"[40, 85] loss: 0.017\n",
"[40, 90] loss: 0.001\n",
"[41, 5] loss: 0.000\n",
"[41, 10] loss: 0.001\n",
"[41, 15] loss: 0.001\n",
"[41, 20] loss: 0.014\n",
"[41, 25] loss: 0.015\n",
"[41, 30] loss: 0.000\n",
"[41, 35] loss: 0.001\n",
"[41, 40] loss: 0.001\n",
"[41, 45] loss: 0.000\n",
"[41, 50] loss: 0.001\n",
"[41, 55] loss: 0.002\n",
"[41, 60] loss: 0.001\n",
"[41, 65] loss: 0.000\n",
"[41, 70] loss: 0.017\n",
"[41, 75] loss: 0.041\n",
"[41, 80] loss: 0.032\n",
"[41, 85] loss: 0.000\n",
"[41, 90] loss: 0.018\n",
"[42, 5] loss: 0.000\n",
"[42, 10] loss: 0.000\n",
"[42, 15] loss: 0.000\n",
"[42, 20] loss: 0.003\n",
"[42, 25] loss: 0.001\n",
"[42, 30] loss: 0.003\n",
"[42, 35] loss: 0.024\n",
"[42, 40] loss: 0.008\n",
"[42, 45] loss: 0.017\n",
"[42, 50] loss: 0.000\n",
"[42, 55] loss: 0.018\n",
"[42, 60] loss: 0.014\n",
"[42, 65] loss: 0.003\n",
"[42, 70] loss: 0.015\n",
"[42, 75] loss: 0.021\n",
"[42, 80] loss: 0.000\n",
"[42, 85] loss: 0.001\n",
"[42, 90] loss: 0.000\n",
"[43, 5] loss: 0.019\n",
"[43, 10] loss: 0.000\n",
"[43, 15] loss: 0.025\n",
"[43, 20] loss: 0.000\n",
"[43, 25] loss: 0.000\n",
"[43, 30] loss: 0.003\n",
"[43, 35] loss: 0.002\n",
"[43, 40] loss: 0.001\n",
"[43, 45] loss: 0.014\n",
"[43, 50] loss: 0.008\n",
"[43, 55] loss: 0.001\n",
"[43, 60] loss: 0.019\n",
"[43, 65] loss: 0.002\n",
"[43, 70] loss: 0.000\n",
"[43, 75] loss: 0.016\n",
"[43, 80] loss: 0.000\n",
"[43, 85] loss: 0.004\n",
"[43, 90] loss: 0.005\n",
"[44, 5] loss: 0.002\n",
"[44, 10] loss: 0.013\n",
"[44, 15] loss: 0.001\n",
"[44, 20] loss: 0.000\n",
"[44, 25] loss: 0.000\n",
"[44, 30] loss: 0.002\n",
"[44, 35] loss: 0.001\n",
"[44, 40] loss: 0.000\n",
"[44, 45] loss: 0.025\n",
"[44, 50] loss: 0.001\n",
"[44, 55] loss: 0.001\n",
"[44, 60] loss: 0.023\n",
"[44, 65] loss: 0.020\n",
"[44, 70] loss: 0.017\n",
"[44, 75] loss: 0.000\n",
"[44, 80] loss: 0.000\n",
"[44, 85] loss: 0.000\n",
"[44, 90] loss: 0.010\n",
"[45, 5] loss: 0.000\n",
"[45, 10] loss: 0.014\n",
"[45, 15] loss: 0.013\n",
"[45, 20] loss: 0.000\n",
"[45, 25] loss: 0.004\n",
"[45, 30] loss: 0.013\n",
"[45, 35] loss: 0.001\n",
"[45, 40] loss: 0.000\n",
"[45, 45] loss: 0.004\n",
"[45, 50] loss: 0.014\n",
"[45, 55] loss: 0.002\n",
"[45, 60] loss: 0.011\n",
"[45, 65] loss: 0.001\n",
"[45, 70] loss: 0.000\n",
"[45, 75] loss: 0.021\n",
"[45, 80] loss: 0.000\n",
"[45, 85] loss: 0.008\n",
"[45, 90] loss: 0.000\n",
"[46, 5] loss: 0.000\n",
"[46, 10] loss: 0.013\n",
"[46, 15] loss: 0.029\n",
"[46, 20] loss: 0.013\n",
"[46, 25] loss: 0.002\n",
"[46, 30] loss: 0.004\n",
"[46, 35] loss: 0.001\n",
"[46, 40] loss: 0.000\n",
"[46, 45] loss: 0.000\n",
"[46, 50] loss: 0.010\n",
"[46, 55] loss: 0.004\n",
"[46, 60] loss: 0.000\n",
"[46, 65] loss: 0.012\n",
"[46, 70] loss: 0.000\n",
"[46, 75] loss: 0.000\n",
"[46, 80] loss: 0.000\n",
"[46, 85] loss: 0.007\n",
"[46, 90] loss: 0.000\n",
"[47, 5] loss: 0.000\n",
"[47, 10] loss: 0.000\n",
"[47, 15] loss: 0.021\n",
"[47, 20] loss: 0.013\n",
"[47, 25] loss: 0.013\n",
"[47, 30] loss: 0.001\n",
"[47, 35] loss: 0.015\n",
"[47, 40] loss: 0.014\n",
"[47, 45] loss: 0.000\n",
"[47, 50] loss: 0.000\n",
"[47, 55] loss: 0.001\n",
"[47, 60] loss: 0.001\n",
"[47, 65] loss: 0.000\n",
"[47, 70] loss: 0.002\n",
"[47, 75] loss: 0.000\n",
"[47, 80] loss: 0.000\n",
"[47, 85] loss: 0.007\n",
"[47, 90] loss: 0.005\n",
"[48, 5] loss: 0.001\n",
"[48, 10] loss: 0.015\n",
"[48, 15] loss: 0.003\n",
"[48, 20] loss: 0.010\n",
"[48, 25] loss: 0.007\n",
"[48, 30] loss: 0.002\n",
"[48, 35] loss: 0.009\n",
"[48, 40] loss: 0.000\n",
"[48, 45] loss: 0.013\n",
"[48, 50] loss: 0.001\n",
"[48, 55] loss: 0.000\n",
"[48, 60] loss: 0.000\n",
"[48, 65] loss: 0.001\n",
"[48, 70] loss: 0.000\n",
"[48, 75] loss: 0.000\n",
"[48, 80] loss: 0.017\n",
"[48, 85] loss: 0.000\n",
"[48, 90] loss: 0.001\n",
"[49, 5] loss: 0.000\n",
"[49, 10] loss: 0.000\n",
"[49, 15] loss: 0.013\n",
"[49, 20] loss: 0.000\n",
"[49, 25] loss: 0.006\n",
"[49, 30] loss: 0.000\n",
"[49, 35] loss: 0.007\n",
"[49, 40] loss: 0.002\n",
"[49, 45] loss: 0.001\n",
"[49, 50] loss: 0.012\n",
"[49, 55] loss: 0.012\n",
"[49, 60] loss: 0.004\n",
"[49, 65] loss: 0.001\n",
"[49, 70] loss: 0.002\n",
"[49, 75] loss: 0.010\n",
"[49, 80] loss: 0.000\n",
"[49, 85] loss: 0.000\n",
"[49, 90] loss: 0.015\n",
"[50, 5] loss: 0.006\n",
"[50, 10] loss: 0.000\n",
"[50, 15] loss: 0.002\n",
"[50, 20] loss: 0.005\n",
"[50, 25] loss: 0.012\n",
"[50, 30] loss: 0.000\n",
"[50, 35] loss: 0.000\n",
"[50, 40] loss: 0.000\n",
"[50, 45] loss: 0.000\n",
"[50, 50] loss: 0.000\n",
"[50, 55] loss: 0.000\n",
"[50, 60] loss: 0.000\n",
"[50, 65] loss: 0.012\n",
"[50, 70] loss: 0.010\n",
"[50, 75] loss: 0.006\n",
"[50, 80] loss: 0.009\n",
"[50, 85] loss: 0.000\n",
"[50, 90] loss: 0.015\n",
"Finished Training\n"
]
}
],
"source": [
"for epoch in range(50): # loop over the dataset multiple times\n",
"\n",
" running_loss = 0.0\n",
" for i, data in enumerate(train_loader, 0):\n",
" # get the inputs; data is a list of [inputs, labels]\n",
" inputs, labels = data\n",
" # zero the parameter gradients\n",
" optimizer.zero_grad()\n",
"\n",
" # forward + backward + optimize\n",
" outputs = net(inputs)\n",
" loss = criterion(outputs, labels)\n",
" loss.backward()\n",
" optimizer.step()\n",
"\n",
" # print statistics\n",
" running_loss += loss.item()\n",
" if i % 5 == 4: # print every 5 mini-batches\n",
" print('[%d, %5d] loss: %.3f' %(epoch + 1, i + 1, running_loss))\n",
" running_loss = 0.0\n",
"\n",
"print('Finished Training')"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Validation"
]
},
{
"cell_type": "code",
"execution_count": 115,
"metadata": {},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
]
},
{
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 432x288 with 1 Axes>"
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"ground true: n\n",
"predicted: n\n"
]
},
{
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 432x288 with 1 Axes>"
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"ground true: n\n",
"predicted: n\n"
]
},
{
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 432x288 with 1 Axes>"
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"ground true: u\n",
"predicted: u\n"
]
},
{
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 432x288 with 1 Axes>"
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"ground true: u\n",
"predicted: u\n"
]
},
{
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 432x288 with 1 Axes>"
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"ground true: u\n",
"predicted: u\n"
]
},
{
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 432x288 with 1 Axes>"
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"ground true: n\n",
"predicted: n\n"
]
},
{
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 432x288 with 1 Axes>"
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"ground true: u\n",
"predicted: u\n"
]
},
{
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 432x288 with 1 Axes>"
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"ground true: u\n",
"predicted: u\n"
]
},
{
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 432x288 with 1 Axes>"
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"ground true: n\n",
"predicted: n\n"
]
},
{
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 432x288 with 1 Axes>"
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"ground true: n\n",
"predicted: n\n"
]
}
],
"source": [
"for i in range(10):\n",
" dataiter = iter(validation_loader)\n",
" images, labels = dataiter.next()\n",
" \n",
" plt.imshow(torchvision.utils.make_grid(images).numpy().transpose(1,2,0))\n",
" plt.show()\n",
" print('ground true: %s'% dataset1.classes[labels])\n",
"\n",
" outputs = net(images)\n",
" _, predicted=torch.max(outputs, 1)\n",
" print('predicted: %s'% dataset1.classes[predicted])"
]
},
{
"cell_type": "code",
"execution_count": 57,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"['__call__',\n",
" '__class__',\n",
" '__constants__',\n",
" '__delattr__',\n",
" '__dict__',\n",
" '__dir__',\n",
" '__doc__',\n",
" '__eq__',\n",
" '__format__',\n",
" '__ge__',\n",
" '__getattr__',\n",
" '__getattribute__',\n",
" '__gt__',\n",
" '__hash__',\n",
" '__init__',\n",
" '__init_subclass__',\n",
" '__le__',\n",
" '__lt__',\n",
" '__module__',\n",
" '__ne__',\n",
" '__new__',\n",
" '__reduce__',\n",
" '__reduce_ex__',\n",
" '__repr__',\n",
" '__setattr__',\n",
" '__setstate__',\n",
" '__sizeof__',\n",
" '__str__',\n",
" '__subclasshook__',\n",
" '__weakref__',\n",
" '_apply',\n",
" '_backend',\n",
" '_backward_hooks',\n",
" '_buffers',\n",
" '_forward_hooks',\n",
" '_forward_pre_hooks',\n",
" '_get_name',\n",
" '_load_from_state_dict',\n",
" '_load_state_dict_pre_hooks',\n",
" '_modules',\n",
" '_named_members',\n",
" '_parameters',\n",
" '_register_load_state_dict_pre_hook',\n",
" '_register_state_dict_hook',\n",
" '_slow_forward',\n",
" '_state_dict_hooks',\n",
" '_tracing_name',\n",
" '_version',\n",
" 'add_module',\n",
" 'apply',\n",
" 'bias',\n",
" 'buffers',\n",
" 'children',\n",
" 'cpu',\n",
" 'cuda',\n",
" 'dilation',\n",
" 'double',\n",
" 'dump_patches',\n",
" 'eval',\n",
" 'extra_repr',\n",
" 'float',\n",
" 'forward',\n",
" 'groups',\n",
" 'half',\n",
" 'in_channels',\n",
" 'kernel_size',\n",
" 'load_state_dict',\n",
" 'modules',\n",
" 'named_buffers',\n",
" 'named_children',\n",
" 'named_modules',\n",
" 'named_parameters',\n",
" 'out_channels',\n",
" 'output_padding',\n",
" 'padding',\n",
" 'padding_mode',\n",
" 'parameters',\n",
" 'register_backward_hook',\n",
" 'register_buffer',\n",
" 'register_forward_hook',\n",
" 'register_forward_pre_hook',\n",
" 'register_parameter',\n",
" 'reset_parameters',\n",
" 'share_memory',\n",
" 'state_dict',\n",
" 'stride',\n",
" 'to',\n",
" 'train',\n",
" 'training',\n",
" 'transposed',\n",
" 'type',\n",
" 'weight',\n",
" 'zero_grad']"
]
},
"execution_count": 57,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"dir(net.conv1)"
]
},
{
"cell_type": "code",
"execution_count": 109,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"Parameter containing:\n",
"tensor([[[[-0.2916, 0.3298, -0.0198],\n",
" [ 0.1312, 0.0062, -0.2484],\n",
" [ 0.0352, -0.1803, -0.1542]]],\n",
"\n",
"\n",
" [[[ 0.0039, -0.0981, -0.1542],\n",
" [-0.2288, -0.2566, 0.0525],\n",
" [ 0.2249, -0.0858, 0.0459]]],\n",
"\n",
"\n",
" [[[ 0.3393, 0.1017, -0.1836],\n",
" [ 0.2088, -0.1313, 0.2788],\n",
" [ 0.0818, 0.3329, -0.2731]]],\n",
"\n",
"\n",
" [[[-0.0754, 0.1212, 0.0227],\n",
" [ 0.1096, -0.3160, 0.2126],\n",
" [ 0.2864, 0.0977, -0.2449]]],\n",
"\n",
"\n",
" [[[ 0.0898, -0.0923, -0.2198],\n",
" [ 0.1969, 0.0460, -0.0275],\n",
" [-0.2288, 0.0851, -0.2544]]],\n",
"\n",
"\n",
" [[[-0.1093, 0.2823, 0.3064],\n",
" [ 0.2368, -0.1769, 0.2854],\n",
" [ 0.3115, -0.1341, 0.3057]]],\n",
"\n",
"\n",
" [[[-0.2234, 0.0800, 0.3144],\n",
" [ 0.1521, -0.2729, -0.0126],\n",
" [-0.3270, -0.2522, -0.1532]]],\n",
"\n",
"\n",
" [[[ 0.2311, -0.1927, 0.3400],\n",
" [ 0.0208, -0.3216, -0.1702],\n",
" [ 0.1410, 0.1508, 0.2295]]]], requires_grad=True)"
]
},
"execution_count": 109,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"net.conv1.weight"
]
},
{
"cell_type": "code",
"execution_count": 98,
"metadata": {},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
]
},
{
"data": {
"text/plain": [
"<matplotlib.image.AxesImage at 0x1aa3b1ebc50>"
]
},
"execution_count": 98,
"metadata": {},
"output_type": "execute_result"
},
{
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 432x288 with 1 Axes>"
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
}
],
"source": []
},
{
"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": "3.6.8"
}
},
"nbformat": 4,
"nbformat_minor": 2
}
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment