SIGN_REG-with_new_data.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Introduction\n",
    "\n",
    "The aim of this code is to develop a model recognizing signs from one to ten, with a simple CNN graphic training model. Our final form of data are plots with 6 lines, representing movements of five fingers and the palm from fist to the sign.\n",
    "\n",
    "The original data is from https://github.com/MEICLabFZU/Gesture-Recognition-based-on-IMU"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torchvision\n",
    "import sys,os\n",
    "sys.path.append(os.pardir)\n",
    "import glob as glob\n",
    "import numpy as np\n",
    "import re"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Convert all data into plots\n",
    "\n",
    "Our initial data are text files with numbers only. A final graph of the sign process are made from 12 text files that two of them generates the x-value and y-value of one line.\n",
    "\n",
    "Thus we need:\n",
    "\n",
    "- a function to delete all empty rows at the last of text files\n",
    "\n",
    "- a function to convert numbers in the text file to a numpy array"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "def remove_last_line(l=[\"\"]):\n",
    "    while l[-1]==\"\":\n",
    "        l.remove(l[-1])\n",
    "        \n",
    "    return l"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "def txt_to_array(xn=\"\"):\n",
    "    x1=open(xn,\"rb\")\n",
    "    lx = x1.read().splitlines()\n",
    "    str_lx=[x.decode('utf-8-sig') for x in lx]\n",
    "    str_lx=remove_last_line(str_lx)\n",
    "    lx_=np.float_(str_lx)\n",
    "    return(lx_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Each 12 files representing one movement are saved in the same directory, and then in the directory with \"sign_pack\" which sign is what the data standing for. \n",
    "\n",
    "Then the process to convert 12 texts in one folder into a simgle graph is simple. All we need to do is to iterate each folder under one directory until we finished plotting all data for all sign."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "path=\"C:\\\\Users\\\\yexue\\\\Gesture-Recognition-based-on-IMU\\\\data\\\\supplementarydata\"\n",
    "fig_path=\"C:\\\\Users\\\\yexue\\\\Gesture-Recognition-based-on-IMU\\\\data\\\\figs_supp\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Peep look into the original and the final data\n",
    "\n",
    "As shown below, the original data is in the form of a text file represent data on one axis of a line in a single plot. For example, following code printed the first 20 lines of data on x-axis of fore finger movement when showing eight."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1200\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "['188.5836623692839',\n",
       " '188.49997627509987',\n",
       " '188.26995488234772',\n",
       " '188.00490746933494',\n",
       " '187.82819793030413',\n",
       " '187.876425085561',\n",
       " '188.31601159932808',\n",
       " '189.25016596842872',\n",
       " '190.26262523163038',\n",
       " '190.03506711893377',\n",
       " '188.2202994280498',\n",
       " '186.43506575248705',\n",
       " '184.58211721041843',\n",
       " '179.84250656893965',\n",
       " '186.18673410610037',\n",
       " '190.49862677723414',\n",
       " '196.0562241467781',\n",
       " '198.3283803931128',\n",
       " '200.28871246493242',\n",
       " '198.54261366044238']"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# only show first 20 lines of the text file\n",
    "x=open(os.path.join(path,'1.data\\\\one-1.txt'),\"r\")\n",
    "list_x=x.read().splitlines()\n",
    "print(len(list_x))\n",
    "list_x[:20]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1200\n",
      "[198.28846887 198.13431381 197.88804743 197.50995118 197.03608453\n",
      " 196.66207286 196.4992172  196.55308216 196.82901418 197.24323546\n",
      " 197.68827239 198.12465381 198.40609804 198.25494977 197.57420124\n",
      " 196.55164467 195.52567173 195.12324117 195.71763849 196.937023  ]\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": [
    "lx = txt_to_array(os.path.join(path,'8.data\\\\eight-1.txt'))\n",
    "print(len(lx))\n",
    "print(lx[:20])\n",
    "width = len(lx)//12\n",
    "\n",
    "for i in range(6):\n",
    "    x1 = lx[i * width:(i+1) * width]\n",
    "    y1 = lx[(i+1) * width:(i+2) * width]\n",
    "    plt.plot(x1,y1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['ten', '100']"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "signList=os.listdir(path)\n",
    "signList=[int(x.split(\".\")[0]) for x in signList]\n",
    "signList\n",
    "sign=signList[1]\n",
    "sign_path=os.path.join(path,str(sign) + \".data\")\n",
    "sign_path\n",
    "packs=os.listdir(sign_path)\n",
    "pack=packs[2]\n",
    "re.findall(r'[0-9]+|[a-z]+',pack)[:-1]\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1 done!\n",
      "10 done!\n",
      "2 done!\n",
      "3 done!\n",
      "4 done!\n",
      "5 done!\n",
      "6 done!\n",
      "7 done!\n",
      "8 done!\n",
      "9 done!\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<Figure size 432x288 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# create a directory\n",
    "if not os.path.isdir(fig_path):\n",
    "    os.mkdir(fig_path)\n",
    "    fig_path=os.path.join(fig_path,\"train\")\n",
    "    os.mkdir(fig_path)\n",
    "    \n",
    "# the list to store all folder for different signs under the initial path\n",
    "signList=os.listdir(path)\n",
    "signList=[int(x.split(\".\")[0]) for x in signList]\n",
    "\n",
    "for sign in signList: # look for sign 1-10\n",
    "    sp=os.path.join(fig_path,str(sign))\n",
    "    if not os.path.isdir(sp):\n",
    "        os.mkdir(sp) # make directory for each sign to save plots\n",
    "    sign_path=os.path.join(path,str(sign) + \".data\")\n",
    "    \n",
    "    packs=os.listdir(sign_path)\n",
    "    for data in packs: # for each pack of one sign, generate the plot respectively\n",
    "        file_name=os.path.join(sign_path, data)\n",
    "        lx = txt_to_array(file_name)\n",
    "        for i in range(6):\n",
    "            x1 = lx[i * width:(i+1) * width]\n",
    "            y1 = lx[(i+1) * width:(i+2) * width]\n",
    "            plt.plot(x1,y1)\n",
    "        #plt.savefig(os.path.join(fig_path,sign,str(pack))+\".jpg\")\n",
    "        lis=re.findall(r'[0-9]+|[a-z]+',data)[:-1]\n",
    "        if len(lis)==1:\n",
    "            lis.append(\"1\")\n",
    "        fig=\".\".join(lis)\n",
    "        plt.savefig(os.path.join(fig_path,str(sign),fig)+\".jpg\")\n",
    "        plt.clf() # clear matplotlib space\n",
    "        #print(pack)\n",
    "    print(str(sign)+\" done!\")\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The generated plots looks as follows that each color represents movement of one part of the hand. This is the an example of sign eight."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "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 cv2\n",
    "packs=os.listdir(os.path.join(fig_path,\"8\"))\n",
    "import matplotlib.image as mpimg\n",
    "img=mpimg.imread(os.path.join(fig_path,\"8\\\\eight.1.jpg\"))\n",
    "imgplot = plt.imshow(img)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Using pytorch to analyze the plots\n",
    "\n",
    "We have generated the plots of sign movement. In order to make and train the model, this step is to devide the data into \"train\" and \"valid\" parts. Since our data size is quite small (about 22 plots for each sign), train size:valid size is approximately 2:1, which is not a usual ratio for such training.\n",
    "\n",
    "First we import all packages."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "#pytorch\n",
    "\n",
    "import glob as glob\n",
    "import os\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import shutil\n",
    "from torchvision import transforms\n",
    "from torchvision import models\n",
    "import torch\n",
    "from torch.autograd import Variable\n",
    "import torch.nn as nn\n",
    "import torch.nn.functional as F\n",
    "from torch.optim import lr_scheduler\n",
    "from torch import optim\n",
    "from torchvision.datasets import ImageFolder\n",
    "from torchvision.utils import make_grid\n",
    "from torch.utils.data import Dataset,DataLoader\n",
    "import time\n",
    "%matplotlib inline\n",
    "\n",
    "import math"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To split the data into train data and valid data, we shuffle first and then take the first one third of all."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "path = \"C:\\\\Users\\\\yexue\\\\Gesture-Recognition-based-on-IMU\\\\data\\\\figs_supp\\\\\"\n",
    "train_path=os.path.join(path,\"train\")\n",
    "packs=os.listdir(train_path)\n",
    "if not os.path.isdir(os.path.join(path,'valid')):\n",
    "    os.mkdir(os.path.join(path,'valid'))\n",
    "valid_path=os.path.join(path,'valid')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "for pack in packs:\n",
    "#     print(pack)\n",
    "    files=glob.glob(os.path.join(train_path,pack,\"*.jpg\"))\n",
    "#     print(os.path.join(train_path,pack,\"*.jpg\"))\n",
    "    num = len(files)\n",
    "    valid_size = math.floor(num/3)\n",
    "    shuffle = np.random.permutation(num)\n",
    "#     print(shuffle)\n",
    "    \n",
    "    valid_path=os.path.join(path,'valid')\n",
    "    if not os.path.isdir(valid_path):\n",
    "        os.mkdir(valid_path)\n",
    "    pack_path = os.path.join(valid_path,pack)\n",
    "    if not os.path.isdir(pack_path):\n",
    "        os.mkdir(os.path.join(valid_path,pack))\n",
    "    \n",
    "    for i in shuffle[0:valid_size]:\n",
    "#         folder = files[i].split('\\\\')[-1].split('.')[0]\n",
    "\n",
    "        image = files[i].split('\\\\')[-1]\n",
    "#         print(image)\n",
    "        os.rename(files[i],os.path.join(path,'valid',pack,image))\n",
    "    "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Model construction and training part. (Refer from an example to classify dogs and cats.)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "simple_transform = transforms.Compose([transforms.Resize((224,224))\n",
    "                                       ,transforms.ToTensor()\n",
    "                                       ,transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])])\n",
    "train = ImageFolder(train_path,simple_transform)\n",
    "valid = ImageFolder(valid_path,simple_transform)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "torch.Size([3, 224, 224])\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": [
    "def imshow(inp,cmap=None):\n",
    "    \"\"\"Imshow for Tensor.\"\"\"\n",
    "    inp = inp.numpy().transpose((1, 2, 0))\n",
    "    mean = np.array([0.485, 0.456, 0.406])\n",
    "    std = np.array([0.229, 0.224, 0.225])\n",
    "    inp = std * inp + mean\n",
    "    inp = np.clip(inp, 0, 1)\n",
    "    plt.imshow(inp,cmap)\n",
    "   \n",
    "imshow(valid[50][0])\n",
    "print(valid[50][0].shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_data_loader = torch.utils.data.DataLoader(train,batch_size=64,shuffle=True)\n",
    "valid_data_loader = torch.utils.data.DataLoader(valid,batch_size=64,shuffle=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Net(nn.Module):\n",
    "    def __init__(self):\n",
    "        super().__init__()\n",
    "        self.conv1 = nn.Conv2d(3, 10, kernel_size=5)\n",
    "        self.conv2 = nn.Conv2d(10, 20, kernel_size=5)\n",
    "#         self.conv2_drop = nn.Dropout2d()\n",
    "#         self.fc1 = nn.Linear(320, 500)\n",
    "        self.fc1 = nn.Linear(56180,500)\n",
    "        self.fc2 = nn.Linear(500,50)\n",
    "        self.fc3 = nn.Linear(50, 10)\n",
    "\n",
    "    def forward(self, x):\n",
    "#         print(x.size())  # [5,3,28,28]\n",
    "        x = F.relu(F.max_pool2d(self.conv1(x), 2)) # [3,224,224]\n",
    "#         print(x.size())  # [5,10,12,12]\n",
    "        x = F.relu(F.max_pool2d(self.conv2(x), 2))\n",
    "#         print(x.size())  # [5,20,4,4]\n",
    "        x = x.view(x.size(0), -1)\n",
    "        x = F.relu(self.fc1(x))\n",
    "#         print(x.size())  # [5,40]\n",
    "        #x = F.dropout(x, training=self.training)\n",
    "        x = F.relu(self.fc2(x))\n",
    "#         print(x.size())  # [5,10]\n",
    "        #x = F.dropout(x,training=self.training)\n",
    "        x = self.fc3(x)\n",
    "#         print(x.size())  # [5,10]\n",
    "\n",
    "        return F.log_softmax(x,dim=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# class Net(nn.Module):\n",
    "#     def __init__(self):\n",
    "#         super(Net, self).__init__()\n",
    "#         self.conv1 = nn.Conv2d(3, 10, kernel_size=5)\n",
    "#         self.conv2 = nn.Conv2d(10, 20, kernel_size=5)\n",
    "#         self.conv2_drop = nn.Dropout2d()\n",
    "#         self.fc1 = nn.Linear(320, 50)\n",
    "#         self.fc2 = nn.Linear(50, 10)\n",
    "\n",
    "#     def forward(self, x):\n",
    "#         x = F.relu(F.max_pool2d(self.conv1(x), 2))\n",
    "#         x = F.relu(F.max_pool2d(self.conv2_drop(self.conv2(x)), 2))\n",
    "#         x = x.view(x.size(0),-1)\n",
    "#         x = F.relu(self.fc1(x))\n",
    "#         x = F.dropout(x, training=self.training)\n",
    "#         x = self.fc2(x)\n",
    "#         return F.log_softmax(x)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = Net()\n",
    "optimizer = optim.SGD(model.parameters(),lr=0.01,momentum=0.5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "def fit(epoch,model,data_loader,phase='training',volatile=False):\n",
    "    if phase == 'training':\n",
    "        model.train()\n",
    "    if phase == 'validation':\n",
    "        model.eval()\n",
    "        volatile=True\n",
    "    running_loss = 0.0\n",
    "    running_correct = 0\n",
    "    for batch_idx , (data,target) in enumerate(data_loader):\n",
    "#         if is_cuda:\n",
    "#             data,target = data.cuda(),target.cuda()\n",
    "        data , target = Variable(data,volatile),Variable(target)\n",
    "        if phase == 'training':\n",
    "            optimizer.zero_grad()\n",
    "        output = model(data)\n",
    "        loss = F.nll_loss(output,target)\n",
    "        \n",
    "        running_loss += F.nll_loss(output,target,size_average=False).item()\n",
    "        preds = output.data.max(dim=1,keepdim=True)[1]\n",
    "        running_correct += preds.eq(target.data.view_as(preds)).cpu().sum()\n",
    "        if phase == 'training':\n",
    "            loss.backward()\n",
    "            optimizer.step()\n",
    "    \n",
    "    loss = running_loss/len(data_loader.dataset)\n",
    "    accuracy = 100. * running_correct/len(data_loader.dataset)\n",
    "    \n",
    "    print(f'{phase} loss is {loss:{5}.{2}} and {phase} accuracy is {running_correct}/{len(data_loader.dataset)}{accuracy:{10}.{4}}')\n",
    "    return loss,accuracy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\yexue\\Anaconda3\\lib\\site-packages\\torch\\nn\\_reduction.py:43: UserWarning: size_average and reduce args will be deprecated, please use reduction='sum' instead.\n",
      "  warnings.warn(warning.format(ret))\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "training loss is   2.3 and training accuracy is 97/1040     9.327\n",
      "validation loss is   2.3 and validation accuracy is 51/510      10.0\n",
      "training loss is   2.3 and training accuracy is 125/1040     12.02\n",
      "validation loss is   2.3 and validation accuracy is 51/510      10.0\n",
      "training loss is   2.3 and training accuracy is 121/1040     11.63\n",
      "validation loss is   2.3 and validation accuracy is 79/510     15.49\n",
      "training loss is   2.3 and training accuracy is 216/1040     20.77\n",
      "validation loss is   2.2 and validation accuracy is 153/510      30.0\n",
      "training loss is   2.2 and training accuracy is 342/1040     32.88\n",
      "validation loss is   2.1 and validation accuracy is 265/510     51.96\n",
      "training loss is   2.0 and training accuracy is 494/1040      47.5\n",
      "validation loss is   1.8 and validation accuracy is 272/510     53.33\n",
      "training loss is   1.6 and training accuracy is 552/1040     53.08\n",
      "validation loss is   1.5 and validation accuracy is 205/510      40.2\n",
      "training loss is  0.98 and training accuracy is 713/1040     68.56\n",
      "validation loss is   1.3 and validation accuracy is 241/510     47.25\n",
      "training loss is  0.73 and training accuracy is 810/1040     77.88\n",
      "validation loss is  0.79 and validation accuracy is 396/510     77.65\n"
     ]
    }
   ],
   "source": [
    "train_losses , train_accuracy = [],[]\n",
    "val_losses , val_accuracy = [],[]\n",
    "for epoch in range(1,10):\n",
    "    epoch_loss, epoch_accuracy = fit(epoch,model,train_data_loader,phase='training')\n",
    "    val_epoch_loss , val_epoch_accuracy = fit(epoch,model,valid_data_loader,phase='validation')\n",
    "    train_losses.append(epoch_loss)\n",
    "    train_accuracy.append(epoch_accuracy)\n",
    "    val_losses.append(val_epoch_loss)\n",
    "    val_accuracy.append(val_epoch_accuracy)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "training loss is  0.36 and training accuracy is 934/1040     89.81\n",
      "validation loss is  0.41 and validation accuracy is 449/510     88.04\n",
      "training loss is  0.17 and training accuracy is 995/1040     95.67\n",
      "validation loss is  0.22 and validation accuracy is 475/510     93.14\n",
      "training loss is 0.097 and training accuracy is 1015/1040      97.6\n",
      "validation loss is  0.11 and validation accuracy is 489/510     95.88\n",
      "training loss is 0.069 and training accuracy is 1023/1040     98.37\n",
      "validation loss is 0.092 and validation accuracy is 495/510     97.06\n",
      "training loss is 0.031 and training accuracy is 1038/1040     99.81\n",
      "validation loss is 0.057 and validation accuracy is 502/510     98.43\n",
      "training loss is 0.023 and training accuracy is 1038/1040     99.81\n",
      "validation loss is 0.041 and validation accuracy is 501/510     98.24\n",
      "training loss is 0.018 and training accuracy is 1038/1040     99.81\n",
      "validation loss is  0.03 and validation accuracy is 510/510     100.0\n",
      "training loss is 0.016 and training accuracy is 1039/1040      99.9\n",
      "validation loss is 0.034 and validation accuracy is 505/510     99.02\n",
      "training loss is 0.013 and training accuracy is 1039/1040      99.9\n",
      "validation loss is 0.024 and validation accuracy is 510/510     100.0\n"
     ]
    }
   ],
   "source": [
    "for epoch in range(1,10):\n",
    "    epoch_loss, epoch_accuracy = fit(epoch,model,train_data_loader,phase='training')\n",
    "    val_epoch_loss , val_epoch_accuracy = fit(epoch,model,valid_data_loader,phase='validation')\n",
    "    train_losses.append(epoch_loss)\n",
    "    train_accuracy.append(epoch_accuracy)\n",
    "    val_losses.append(val_epoch_loss)\n",
    "    val_accuracy.append(val_epoch_accuracy)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "training loss is 0.011 and training accuracy is 1039/1040      99.9\n",
      "validation loss is 0.023 and validation accuracy is 505/510     99.02\n",
      "training loss is 0.011 and training accuracy is 1039/1040      99.9\n",
      "validation loss is 0.021 and validation accuracy is 505/510     99.02\n",
      "training loss is 0.0086 and training accuracy is 1039/1040      99.9\n",
      "validation loss is 0.032 and validation accuracy is 505/510     99.02\n",
      "training loss is 0.0099 and training accuracy is 1039/1040      99.9\n",
      "validation loss is  0.02 and validation accuracy is 505/510     99.02\n"
     ]
    }
   ],
   "source": [
    "for epoch in range(1,5):\n",
    "    epoch_loss, epoch_accuracy = fit(epoch,model,train_data_loader,phase='training')\n",
    "    val_epoch_loss , val_epoch_accuracy = fit(epoch,model,valid_data_loader,phase='validation')\n",
    "    train_losses.append(epoch_loss)\n",
    "    train_accuracy.append(epoch_accuracy)\n",
    "    val_losses.append(val_epoch_loss)\n",
    "    val_accuracy.append(val_epoch_accuracy)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "for epoch in range(1,10):\n",
    "    epoch_loss, epoch_accuracy = fit(epoch,model,train_data_loader,phase='training')\n",
    "    val_epoch_loss , val_epoch_accuracy = fit(epoch,model,valid_data_loader,phase='validation')\n",
    "    train_losses.append(epoch_loss)\n",
    "    train_accuracy.append(epoch_accuracy)\n",
    "    val_losses.append(val_epoch_loss)\n",
    "    val_accuracy.append(val_epoch_accuracy)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "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": [
    "path = \"C:\\\\Users\\\\yexue\\\\Gesture-Recognition-based-on-IMU\"\n",
    "\n",
    "plt.plot(range(1,len(train_losses)+1),train_losses,'bo',label = 'training loss')\n",
    "plt.plot(range(1,len(val_losses)+1),val_losses,'r',label = 'validation loss')\n",
    "plt.legend()\n",
    "plt.savefig(os.path.join(path,\"loss.jpg\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAD4CAYAAAAXUaZHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8QZhcZAAAgAElEQVR4nO3deXxU1fnH8c9h0TQssqoskqCFsoclKIILi1K3ulPll6q4oVVR3Be0YC2tilKLrWhcCmpcKIqAdYXi0mrYRFxABAViBCHsYAQEnt8fZxICTLbJTO5k8n2/XvO6M2fuufdhMjw5Offcc5yZISIiiaVG0AGIiEj0KbmLiCQgJXcRkQSk5C4ikoCU3EVEElCtoAMAaNKkiaWmpgYdhohIlTJ//vx1ZtY03HtxkdxTU1OZN29e0GGIiFQpzrmVxb2nbhkRkQSk5C4ikoCU3EVEElBc9LmH8/PPP5Obm8v27duDDkVKkZSURMuWLaldu3bQoYhISNwm99zcXOrVq0dqairOuaDDkWKYGevXryc3N5fWrVsHHY6IhJTaLeOce8Y5t9Y590WRskbOuXedc0tD24ahcuecG+ecW+ac+8w51z3SwLZv307jxo2V2OOcc47GjRvrLyzZR1YWpKZCjRp+m5UV+7rV5ZxlZmYlPoATgO7AF0XKHgTuCD2/A3gg9Pw04E3AAb2A2aUd38zo0aOH7W/RokUHlEn80s8r9p5/3iwlxcw5v33++djWi7Tu88+bJSebwd5HcnJs61aXc+4PmGfF5e7i3thnJ0jdL7kvAZqFnjcDloSePwEMDrdfSQ8l96pPP6/YCiIBZU382U44ONtuZoxdyRPWl//YUUm59vxze0qsl5Ky7/kKHikppZxwyxY77fD5dgEv2p2Mtit5wk5kljXje0tpFaNzVqBuEOfcX0nJ3VkZ5nN3zqUCr5tZp9DrTWbWoMj7G82soXPudeB+M/tvqHwmcLuZHXCHknNuKDAUoFWrVj1Wrtx3LP7ixYtp3759qbHFyqZNm3jhhRe45ppryl33tNNO44UXXqBBgwal75wggv55JbrUVFgZ5naVlBRYsSJK9cxgyRKYMQNmzGDLtPeob5sPqPujq0OdtDbQti20CW0LHo0aUaOGP9T+nIM9P+2Ab76Br7/2j6VL9z7/4Ydi/x3bqEPdrvudq+D8pZ1zT/GfDxBx3SDOeeD+br6ZpYd7L9oXVMN1kIf97WFmmUAmQHp6eoVXDMnKghEjICcHWrWC0aMhIyPy423atInHHnssbHLfvXs3NWvWLLbuG2+8EfmJY6jgN3qNGhoBG5RIv6c5OeUrL3O91ath5szChM733/vy1q152X7Lu5zEe/Qlie205Wva8jW/sq+5oflS+OQTeOUV2L1774EbNWJe7bZ8vrMtS2nDVurxS5bRlq/pUONr+MXKfbPaoYf6JH3aadCmDVeNbcv/8tqynNY0Zn3hOdPrfc1lzb4Of87GjZl7UFs+3+H3XkVz9oQuJzZpBDxb8mc0vBGsW39geWl1I623f93ZHMPX/Arw34moKa5JX/RBnHfLRKv/qqgLLrjAkpKSLC0tzW655RabNWuW9e3b1wYPHmzt27c3M7OzzjrLunfvbh06dLAnnniisG5KSorl5eXZ8uXLrV27dnbFFVdYhw4d7OSTT7b8/PwDzjVt2jQ7+uijrWvXrjZgwAD74YcfzMxs69atNmTIEOvUqZN17tzZJk+ebGZmb775pnXr1s26dOli/fv3NzOzkSNH2pgxYwqP2bFjR1u+fHlhDL///e+ta9eutmLFCrv66qutR48e1qFDB/vDH/5QWGfOnDl27LHHWpcuXaxnz562ZcsWO+6442zBggWF+/Tu3dsWLlx4wL+hunXLVHZfdMGf8XXZYg3YYEnkm2N3ubsO6rLFTme6PV3vBrOOHfe+0bix2W9/a5aZafbNN2Hrhu062LHD7KuvzKZPN3v4YbOrrrLVHfrbd65lYYVN1Ld5NdJt+bGDzUaONMvKMps712zTpsg+o4JzTptW5Jz9LNe1CB9wnD+uYnxc9bmPYd8Lqg+Gnp/OvhdU55Tl+BVN7tHqvypq+fLl1rFjx8LXs2bNsuTkZPv2228Ly9avX29mZvn5+daxY0dbt25dKJ69yb1mzZqFyXHQoEH23HPPHXCuDRs22J49vk/xySeftJtuusnMzG677Ta74YYb9tlv7dq11rJly8I4CmIoKbk75+zjjz8+IO5du3bZiSeeaAsXLrQdO3ZY69atbc6cOWZmtnnzZvv5559twoQJhTEsWbLEwv2szKpXco80SZf7e7ptm9kHH5g9/LAtP3awLXW/PKDyrpq1zerWNWvSxKxlS7OjjvJJu3t3s969bXWHfvZWjVPtFc6xD+ljO6llBvZz7SSzgQPNHnzQ7JNPzHbvjtq/s6BuuyO22aGssZRWe2J+EbfoOY/kGzuu+Tf22thv/C+qMjxeG+vrlLdupPWK1q3P5nJf6C5QUnIvtVvGOfci0Bdo4pzLBUYC9wOTnHOXAznAoNDub+BHzCwD8oFLy/2nRAQi/ZO1vI4++uh9xnKPGzeOKVOmAPDdd9+xdOlSGjduvE+d1q1b07VrVwB69OjBijAdpLm5uVxwwQWsXr2anTt3Fp5jxowZvPTSS4X7NWzYkOnTp3PCCScU7tOoUaNS405JSaFXr16FrydNmkRmZia7du1i9erVLFq0COcczZo1o2fPngDUr18fgEGDBnHfffcxZswYnnnmGYYMGVLq+RLdiBGQn79vWX6+Ly+pi6XE7+mOHfDZZzBvHsyd6x+LFhV2wKYecQQ56T15cOmlrNpUh8MbbOfUATtIa7vd192+3T8Knoe2h9fcQa3UdeR9t52NP9flifq30e66kzjpnmMhKanEf2fBvyWSbqSMDMjIqAPUKX3nsHXLXa3IOY8sd92zbvSPyqpX0bplUWpyN7PBxbw1IMy+Blxb0aDKq1Wr8BeNotp/BdSps/eL+t577zFjxgw+/vhjkpOT6du3b9ix3gcffHDh85o1a/LTTz8dsM+wYcO46aabOPPMM3nvvfcYNWoU4P+q2n+cf7gygFq1arGnyJWYorEUjXv58uU89NBDzJ07l4YNGzJkyBC2b99e7HGTk5M5+eSTmTp1KpMmTdLsnUTemCj4ntZkF+34ip7MpSdz6VN7LtT/DHbu9Ds2bQo9e8K55/ptz55w2GG0Am6LIN4moQdA73LWjTTRSvAS4sra6NGQnLxvWXKyL49UvXr12Lp1a7Hvb968mYYNG5KcnMxXX31FdnZ2xOfavHkzLVq0AGDixImF5QMHDuTvf/974euNGzdy7LHH8v7777N8+XIANmzYAPhpkz/55BMAPvnkk8L397dlyxbq1KnDIYccwpo1a3jzzTcBaNeuHatWrWLu3LkAbN26lV27dgFwxRVXcP3119OzZ88y/aWQ6IprNBxQbuYvWL79Njz0EP854hI+dd3YRl2+oDP/5DIyyOKwX9aD4cPhX//yQ1jWrIF//xvuvRfOOAMOOyzW/yRJQAmR3DMyIDPTD+9yzm8zMyvW4mjcuDF9+vShU6dO3HrrrQe8f8opp7Br1y66dOnCPffcs0+3R3mNGjWKQYMGcfzxx9OkSZPC8rvvvpuNGzfSqVMn0tLSmDVrFk2bNiUzM5Nzzz2XtLQ0LrjgAgDOO+88NmzYQNeuXRk/fjxt27YNe660tDS6detGx44dueyyy+jTpw8ABx10EC+//DLDhg0jLS2Nk08+ubD136NHD+rXr8+ll1ZKL1ulifQOwXCNiaa/2Mbjl86Gp56C66+Hfv18C7x5czjlFLj1Vo78dgZNOx3GxPrDuIjn6N/8K15/diOHf/kfeOABOP/8vV9ikQoq0zj3WEtPT7f9/9zXuOn4sWrVKvr27ctXX31V7DDKqvbzysqCoUP37TtPTi57o+DVcbksG/kcbTbNoVutz0nd9c3eN+vUgU6doHNn6NLFbzt3hv2ux4hUVGWOc5cE8+yzzzJixAjGjh2bUOPjI7ooaubHhI8fz7lTp/qx1r/6FXTuBl0u2ZvEW7f2fw6IBEjJXUp08cUXc/HFFwcdRtSV66Loxo0wcSKMH+/vpGzcGG6+Ga66Co4s/8gMkcqg5C7VUplGWM2fD489Bi++CD/9BMceC8895/vGSxlGKBI0/e0o1VJxI6zuH/kTTJgAxxwD6enw0ktw0UWwYAF89BH87ndK7FIlqOUu1dL+N+gc32wZ47s+Toebn/HdMO3bw7hxcPHFcMghwQYrEgEld6m2MjIgo+1cuPtueOcdWFsLzjkHrrkGTjxRQxKlSlNyj6K6deuybdu2oMOQsjDz/ek33ugvkP7xj3DFFdCsWdCRiUSFknsC2bVrF7Vq6Udaqm3b4MorfX/66afDs8+C7ryVBKMLqsW4/fbbeeyxxwpfjxo1iocffpht27YxYMAAunfvTufOnZk6dWqpxzr77LPp0aMHHTt2JDMzs7D8rbfeonv37qSlpTFggJ+qZ9u2bVx66aV07tyZLl268MorrwD+r4ICkydPLpzAa8iQIdx0003069eP22+/nTlz5tC7d2+6detG7969WbJkCeDnoL/lllsKj/voo48yc+ZMzjnnnMLjvvvuu5x77rmRf2hVwaJFfq6WSZPgz3+GadOU2CUhVY1m3vDh8Omn0T1m167wyCPFvn3hhRcyfPjwwsU6Jk2axFtvvUVSUhJTpkyhfv36rFu3jl69enHmmWeWuJD3M888Q6NGjfjpp5/o2bMn5513Hnv27OHKK6/kgw8+oHXr1oVzxNx3330ccsghfP7554CfT6Y0X3/9NTNmzKBmzZps2bKFDz74gFq1ajFjxgzuuusuXnnlFTIzM1m+fDkLFiygVq1abNiwgYYNG3LttdeSl5dH06ZN+ec//5lwUwzs44UXfIu9bl2/MEW/fkFHJBIzarkXo1u3bqxdu5ZVq1axcOFCGjZsSKtWrTAz7rrrLrp06cJJJ53E999/z5o1a0o81rhx40hLS6NXr16FUwNnZ2eHnbp3xowZXHvt3ok1GzZsWGqsgwYNKlwZavPmzQwaNIhOnTpx44038uWXXxYe9+qrry7stmnUqBHOOS666CKef/55Nm3axMcff8ypp55a/g8rYKXOEbNjh79ImpEBPXr4YY1K7JLgqkbLvYQWdiydf/75TJ48mR9++IELL7wQgKysLPLy8pg/fz61a9cmNTU17FS/BYqbGri4KXaLKy9atv/5ik7pe88999CvXz+mTJnCihUr6Nu3b4nHvfTSS/nNb35DUlISgwYNqnJ99vvPEbNypX8NoeGOK1bAoEF+nvRbb/VdMVXs3ygSCbXcS3DhhRfy0ksvMXnyZM4//3zAt4wPPfRQateuzaxZs9h/Ye/9FTc1cHFT94ab5hfgsMMOY/HixezZs6dwgZDizlcwffCECRMKywcOHMjjjz9eOI1vwfmaN29O8+bN+dOf/lQlF+IoaY4Y/v1v6N7dL8I8ZQo8+KASu1QbSu4l6NixI1u3bqVFixY0Cw2Ry8jIYN68eaSnp5OVlUW7du1KPEZxUwMXN3VvuGl+Ae6//37OOOMM+vfvXxhLOLfddht33nknffr0YXeRRYSvuOIKWrVqRZcuXUhLS+OFF14ofC8jI4MjjjiCDh06RPZBBSjcXDA12M2VK+/2c6GnpPhpBM4+u/KDEwmQpvwVrrvuOrp168bll18e8TGC+nmlpu47R8yhrOEF/o8B/MePWx83Dn7xi0qPS6QylDTlr1ru1VyPHj347LPP+N3vfhd0KBEpOkfMcXzIArrRm4/4eOg/4cknldil2lIHZDU3f/78oEOokII5Yr4Z9gh3bbyF72q15j9/fIvT7+wSbGAiAYvr5F7cCA+JL0F37WU0eRs23ghnn03rCRNorYm+ROK3WyYpKYn169cHnjikZGbG+vXrSQpqGtxt2/yiGe3a+ekElNhFgDhuubds2ZLc3Fzy8vKCDkVKkZSURMuWLYM5+R/+4K+ofvghHHxwMDGIxKG4Te61a9cuvHtTJKw5c+Bvf4Orr4bjjgs6GpG4ErfdMiIl+vlnP9Tx8MPh/vuDjkYk7sRty12kRA89BJ9/Dq+9pn52kTDUcpeq5+uv4d57/ULVZ50VdDQicUnJXaqWPXv8zGBJSf7uUxEJS90yUrU8/TS8/76/+1RL4okUSy13qTpWr/bT9vbtCxWYB0ekOlByl6pj2DDYvh0yM0F3LouUSN0yUjVMmQKvvAJ/+Qu0aRN0NCJxTy13iX+bN8O110JaGtx8c9DRiFQJarlL/LvjDlizBqZOhdq1g45GpEpQy13i24cfwuOPw/Dh0LNn0NGIVBkVSu7OuRudc186575wzr3onEtyzrV2zs12zi11zr3snDsoWsFKNbN9O1x5pV9u6Y9/DDoakSol4uTunGsBXA+km1knoCZwIfAA8FczawNsBDRmTSIzejQsWQJPPAF16gQdjUiVUtFumVrAL5xztYBkYDXQH5gcen8ioJWJpfw+/9xPCHbRRTBwYNDRiFQ5ESd3M/seeAjIwSf1zcB8YJOZ7Qrtlgu0CFffOTfUOTfPOTdPc7bLPnbv9t0xDRrA2LFBRyNSJVWkW6YhcBbQGmgO1AFODbNr2KWUzCzTzNLNLL1p06aRhiEJIivLd63XqAGjmv4DZs/2c7U3aRJ0aCJVUkWGQp4ELDezPADn3KtAb6CBc65WqPXeElhV8TAlkWVl+bnA8vOhFSu5ZeNdvFPjFPL2DCYj6OBEqqiK9LnnAL2cc8nOr2I9AFgEzALOD+1zCTC1YiFKohsxwid2MB7jGgCu3PM4I+7WFAMikapIn/ts/IXTT4DPQ8fKBG4HbnLOLQMaA09HIU5JYDk5fnsur3I6bzCC0eSQUlguIuVXoTtUzWwkMHK/4m+BoytyXKleWrXya1yfy6usohl/57rCchGJjO5QlcCNHg3JydCLbD6iN3uoSXKyLxeRyCi5S+AyMmDimLUcxbfMphcpKX5W3wxdTRWJmCYOk7hw/hGzARjzYS/GHBdwMCIJQC13iQ/Z2VCrFnTvHnQkIglByV3iQ3a2n689OTnoSEQSgpK7BG/3bpgzB3r1CjoSkYSh5C7BW7QItm1TcheJIiV3CV52tt8quYtEjZK7BC87Gxo3hqOOCjoSkYSh5C7By872rXanuWREokXJXYK1aZPvc1eXjEhUKblLsObO9Vsld5GoUnKXYGVn++6Ynj2DjkQkoSi5S7Cys6FDBzjkkKAjEUkoSu4SHLO9F1NFJKqU3CU4y5bBhg1K7iIxoOQuUVV0oevUVP+6WLp5SSRmNOWvRE3Rha7Br640dKh/HnZu9uxsqFcP2revtBhFqgu13CVq9i50vVd+vi8PKzsbjj4aataMeWwi1Y2Su0RNcQtahy3Pz4eFC9UlIxIjSu4SNcUtaB22fP58P9WvkrtITCi5S9QULHRdVLELXRdcTD3mmJjHJVIdKblL1GRk+IWtU1L8TaclLnSdne1ngWzatNLjFKkONFpGoiojo5hkXpQZfPwx9O9fKTGJVEdquUvly82F1avV3y4SQ0ruUvl085JIzCm5S+XLzoakJOjSJehIRBKWkrtUvuxs6NEDDjoo6EhEEpaSu1SunTv9GHd1yYjElJK7VK6FC2HHDiV3kRhTcpfKpYupIpVCyV0qV3Y2tGgBLVsGHYlIQlNyl8qllZdEKoWSu1SetWvh22+V3EUqQYWSu3OugXNusnPuK+fcYufcsc65Rs65d51zS0PbhtEKVqq42bP9VsldJOYq2nL/G/CWmbUD0oDFwB3ATDNrA8wMvRbxXTK1akH37kFHIpLwIk7uzrn6wAnA0wBmttPMNgFnARNDu00Ezq5okJIgsrMhLe3AeYFFJOoq0nI/EsgD/umcW+Cce8o5Vwc4zMxWA4S2h4ar7Jwb6pyb55ybl5eXV4EwpErYvRvmzFGXjEglqUhyrwV0B8abWTfgR8rRBWNmmWaWbmbpTTWnd+JbtAi2bVNyF6kkFUnuuUCumYWukjEZn+zXOOeaAYS2aysWoiQE3bwkUqkiTu5m9gPwnXPuV6GiAcAiYBpwSajsEmBqhSKUxJCdDY0b+9WXRCTmKroS0zAgyzl3EPAtcCn+F8Yk59zlQA4wqILnkERQcPOSc0FHIlItVCi5m9mnQHqYtwZU5LiSYDZt8n3ugwcHHYlItaE7VCX25s71W/W3i1QaJXeJvexs3x3Ts2fQkYhUG0ruEnvZ2dChAxxySNCRiFQbSu4SW2aaCVIkAEruElvLlsGGDUruIpVMyV1iSzcviQRCyV1iKzsb6tWD9u2DjkSkWlFylwNkZUFqKtSo4bdZWRU4WHY2HH001KwZpehEpCyU3BONGeTkwKuvwp13wumnw9tvl7l6VhYMHQorV/pDrVzpX0eU4PPzYeFCdcmIBKCi0w9I0PLy/E1CRR9rQ3O11aoFSUmwdKm/Q7RW6T/uESN8Ti4qP9+XZ2SUM7b58/1Uv0ruIpVOyb0q2bzZJ8yiiTwnx7/nnO/XPvVUf7NQerpfGOOtt+Ccc3zT+5JLSj4+ew9X1vISFVxMPeaYCCqLSEUouVcFs2fDkCHw1Vd7y448Eo49FoYN88m8e3d/4XJ/Z50F3brBfffB//0f1K5d4qlatfJdMeHKyy07288Cqfn6RSqdkntV8MADvvvlvvv2tsobNy5bXefg3nvhzDPhuefgsstK3H30aN/HXrRrJjnZl5eLGXz8MfTvX86KIhINuqAa77Zvh3fegd/+Fu6+G37967In9gJnnOF/Idx3H/z8c4m7ZmRAZiakpPjfCykp/nW5+9tzc2H1avW3iwREyT3ezZoFP/4Iv/lN5McoaL2vWAETJpS6e0aG33XPHr8td2IH3bwkEjAl93g3fTrUqQP9+lXsOKee6i9s/ulPsHNndGIrSXa2H6nTpUvszyUiB1Byj2dmPrkPHOgTZUUUtN5zcuCZZ6ITX0mys6FHDzjooNifS0QOoOQezz791Pddn3lmdI43cCD07u2vju7YEZ1jhrNzpx+yqS4ZkcAoucezadN8i/u006JzvILWe24uPPVUdI4ZzsKF/peHkrtIYJTc49n06X4s+6GHRlQ97BwxAwbA8cfDn//sR+LEgi6migROyT1eff+979qIcJRMsXPEvBBqva9a5cc4xkJ2NrRoAS1bxub4IlIqJfd49frrfhthf3tJc8TQrx/07Qt/+Qv89FOFwjxAwc1LarWLBErJPV5Nm+anGIhwHvRS54i591744Qd4/PHI4gvHDG67DZYv90MvRSQwSu7x6McfYeZM32p3LqJDFDcXTGH5CSf4/vf77/fni4bRo+Ghh+Daa0ud5kBEYkvJPR69+64fbVKBu1JHj/ZzwhR1wBwx997rpwcePz7i8xQaNw7uuQcuusg/j/CXkohEh5J7PJo+HQ45xI9qiVCZ5ojp08ePfX/gAdi2LfJ4J0yAG26As8/2N0jV0NdKJGj6Xxhv9uzxF1NPPbXU6XlLU6Y5Yu69F9atg3/8I7KTvPIKXH45nHQSvPRSmRYEEZHYU3KPN3Pm+K6SaN2VWppevfwvkgcfhC1bylf37bdh8GA/Z81rr8HBB8cmRhEpNyX3eDNtml9M+pRTKu+c994LGzbAo4+Wvc5//+tXeOrQAd54w09uJiJxQ8k93kyf7keyNGxYeefs2dNfvH34Yb+UX2k++cQvvN2qlZ9rvkGD2McoIuWi5B5Pli+HL76o2NztkRo1CjZuhL/9reT9Fi/2C4Y0aOBH9UQ4NYKIxJaSezyZPt1vK6u/vaju3f1ol7FjYdOm8PusWAEnn+y7jWbMgCOOqNQQRaTslNzjybRp/o7Uo44K5vyjRvlumb/+9cD3Vq/2I2Ly832LvU2bSg9PRMpOyT1ebN4M778fTKu9QFoanHcePPKIv8BaYP1632L/4Qd4803o3Dm4GEWkTJTc48Vbb8GuXcH0txc1cqQfEjl2rH+9dasfKrlsme82OuaYYOMTkTKpcHJ3ztV0zi1wzr0eet3aOTfbObfUOfeyc07rrJXF9OnQpEnwsyl27gy//a2/sJqb63/ZLFgA//pXxddxFZFKE42W+w3A4iKvHwD+amZtgI3A5VE4R2LbtcuPFT/9dH+xMmgjR/rJxLp2hQ8+gGefDf4vChEplwold+dcS+B04KnQawf0ByaHdpkInF2Rc1QL//ufH4YYLwm0Qwe48ELf1/744/4uVBGpUiracn8EuA3YE3rdGNhkZrtCr3OBFuEqOueGOufmOefm5eXlVTCMKm7aNDjoID+JV7zIzPSLbgwdGnQkIhKBiJO7c+4MYK2ZzS9aHGZXC1ffzDLNLN3M0ps2bRppGFWfmU/u/fpBvXpBR7NX3brB9/+LSMQqMoVfH+BM59xpQBJQH9+Sb+CcqxVqvbcEVlU8zAS2ZIkfiXLjjUFHIiIJJOKWu5ndaWYtzSwVuBD4j5llALOA80O7XQJMrXCUiazgrtQzzgg2DhFJKLEY5347cJNzbhm+D/7pGJwjcUyb5kelFLcunohIBKKysoKZvQe8F3r+LXB0NI6b8Natg48+ghEjgo5ERBKM7lAN0htv+GWSgpxyQEQSkpJ7kKZPh2bN/IyMIiJRpOQelB07/Hwyv/mNFpQWkahTVgnK++/Dtm3xc1eqiCQUJfegTJsGv/gFDBgQdCQikoCU3INg5vvbTz7ZJ3gRkShTcg/CZ59BTo5GyYhIzCi5B6HgrtTTTy9116wsSE3111xTU/1rEZHSROUmJimnadP8ikaHH17ibllZflLG/Hz/euXKvZM0ZmTEOEYRqdLUcq9sq1fD3LllGiUzYsTexF4gP183tIpI6ZTcK9vrr/ttGfrbc3LKVy4iUkDJvbJNnw4pKdCpU6m7FjeXmOYYE5HSKLlXpvx8ePdd32p34dY12dfo0ZCcvG9ZcrIvFxEpiZJ7ZZo5E7ZvL/NdqRkZfrW7lBT/uyAlxb/WxVQRKY1Gy1SmadP8UnonnljmKhkZSuYiUn5quVeWn3/2/e2nnOIXwxYRiSEl98oyaRKsWQMXXRR0JCJSDSi5VwYzGDMG2rUr012pIkdif70AAAh/SURBVCIVpT73yjBjBixcCE8/rbnbRaRSKNNUhjFj/IpLujIqIpVEyT3WFizwY9uvvx4OPjjoaESkmlByj7WHHoK6deHqq4OORESqESX3WFq5El5+2U/l2KBB0NGISDWi5B5Ljzziby0dPjzoSESkmlFyj5WNG+HJJ2HwYDjiiKCjEZFqRsk9VsaPhx9/hFtuCToSEamGlNxjYft2GDcOfv1r6NIl6GhEpBpSco+F557zUw3cemvQkYhINaXkHm179sDDD0P37tC/f9DRiEg1pekHom36dFiyBF58sUwLcoiIxIJa7tE2ZgykpsL55wcdiYhUY2q5R9NHH8H//ucvptbSRysiwVHLPZrGjIFGjeCyy4KORESqOSX3aFmyBKZOhWuugTp1Dng7K8v31tSo4bdZWZUeoYhUIxEnd+fcEc65Wc65xc65L51zN4TKGznn3nXOLQ1tG0Yv3Dj28MN++bxhww54KyvLTy+zcqVft2PlSv9aCV5EYqUiLfddwM1m1h7oBVzrnOsA3AHMNLM2wMzQ68S2Zg08+ywMGQKHHnrA2yNGQH7+vmX5+b5cRCQWIk7uZrbazD4JPd8KLAZaAGcBE0O7TQTOrmiQce/RR2HnTrj55rBv5+SEr1ZcuYhIRUWlz905lwp0A2YDh5nZavC/AIADm7KJZNs2eOwxOOccaNMm7C6tWoWvWly5iEhFVTi5O+fqAq8Aw81sSznqDXXOzXPOzcvLy6toGMF5+mk/A2QJUw2MHg3JyfuWJSf7chGRWKhQcnfO1cYn9iwzezVUvMY51yz0fjNgbbi6ZpZpZulmlt60adOKhBGcXbvgr3+F446DXr2K3S0jAzIzISXF37SakuJfa0lVEYmViO+0cc454GlgsZmNLfLWNOAS4P7QdmqFIoxn//qXH/ry6KOl7pqRoWQuIpXHmVlkFZ07DvgQ+BzYEyq+C9/vPgloBeQAg8xsQ0nHSk9Pt3nz5kUUR2DMoEcP+Okn+PJLP4BdRKQSOefmm1l6uPcibrmb2X+B4mbGGhDpcauMmTNhwQJ46ikldhGJO8pKkRozBg4/HH73u6AjERE5gJJ7JBYuhHfegRtugIMPDjoaEZEDKLlHYswYqFsXrr466EhERMJSci+vnBx46SW48kpo0CDoaEREwlJyL4+cHDjrLH8BdfjwoKMRESmWkntZffQR9OwJ337rp/bV3AEiEseU3MtiwgTo1w/q14fsbDj11KAjEhEpkZJ7SXbvhltugUsvheOPh9mzoX37oKMSESmVFvoszubNMHgwvPkmXHcdjB0LtWsHHZWISJlUy5Z7qUveLV3qJwJ79114/HE/d4wSu4hUIdUuuZe65N3MmXDMMZCX55P7VVftU1froIpIVVDtknuxS97dZfCPf8Cvfw3Nm8OcOdC3b+E+WgdVRKqSapfcwy1tV5ud3Jnze9+3ftppftjjkUfus4/WQRWRqqTaJff9h6c3Zh3vMJCreALuuAOmTPFDHvejdVBFpCqpdsm96JJ3HfmCORxNL7L539XPwV/+AjVrhq2ndVBFpCqpskMhs7Ig87ZlHLJqMU2bwkUX7dNFXqyM+tDianh7wmru2nAzP9Wsy3v3vM8pI48psd7o0b6PvWjXjNZBFZF4FfFKTNFU3pWYCi5uXpM/hjHcFvmJe/SA116Dli3LfN4RI3xXTKtWPrFr6TwRCUpJKzFVyeSemupHqxzGD7Tg+8LyZofD66+X8SA1akDHjnDQQeULVkQkTsRkmb0gFVzEXMPhrOHwwnK3BugRTEwiIvGkSl5Q1cVNEZGSVcnkXnTESwFd3BQR2atKJveMDMjMhJQUcM5vMzN1cVNEpECV7HMHn8iVzEVEwquSLXcRESmZkruISAJSchcRSUBK7iIiCUjJXUQkAcXF9APOuTzgR2Bd0LHEsSbo8ymNPqOS6fMpXVX7jFLMrGm4N+IiuQM45+YVN0eC6PMpC31GJdPnU7pE+ozULSMikoCU3EVEElA8JffMoAOIc/p8SqfPqGT6fEqXMJ9R3PS5i4hI9MRTy11ERKJEyV1EJAEFntydc6c455Y455Y55+4IOp545Jxb4Zz73Dn3qXOu7OsRJjDn3DPOubXOuS+KlDVyzr3rnFsa2jYMMsYgFfP5jHLOfR/6Hn3qnDstyBiD5Jw7wjk3yzm32Dn3pXPuhlB5wnyHAk3uzrmawD+AU4EOwGDnXIcgY4pj/cysa6KMwY2CCcAp+5XdAcw0szbAzNDr6moCB34+AH8NfY+6mtkblRxTPNkF3Gxm7YFewLWh3JMw36GgW+5HA8vM7Fsz2wm8BJwVcExSBZjZB8CG/YrPAiaGnk8Ezq7UoOJIMZ+PhJjZajP7JPR8K7AYaEECfYeCTu4tgO+KvM4Nlcm+DHjHOTffOTc06GDi2GFmthr8f17g0IDjiUfXOec+C3XbVNkuh2hyzqUC3YDZJNB3KOjk7sKUaWzmgfqYWXd899W1zrkTgg5IqqTxwFFAV2A18HCw4QTPOVcXeAUYbmZbgo4nmoJO7rnAEUVetwRWBRRL3DKzVaHtWmAKvjtLDrTGOdcMILRdG3A8ccXM1pjZbjPbAzxJNf8eOedq4xN7lpm9GipOmO9Q0Ml9LtDGOdfaOXcQcCEwLeCY4opzro5zrl7Bc2Ag8EXJtaqtacAloeeXAFMDjCXuFCStkHOoxt8j55wDngYWm9nYIm8lzHco8DtUQ8OxHgFqAs+Y2ehAA4ozzrkj8a118Auav6DPCJxzLwJ98VO0rgFGAq8Bk4BWQA4wyMyq5UXFYj6fvvguGQNWAFcV9C9XN86544APgc+BPaHiu/D97gnxHQo8uYuISPQF3S0jIiIxoOQuIpKAlNxFRBKQkruISAJSchcRSUBK7iIiCUjJXUQkAf0/iIkspv1n19sAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(range(1,len(train_accuracy)+1),train_accuracy,'bo',label = 'train accuracy')\n",
    "plt.plot(range(1,len(val_accuracy)+1),val_accuracy,'r',label = 'val accuracy')\n",
    "plt.legend()\n",
    "plt.savefig(os.path.join(path,\"accuracy.jpg\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 166,
   "metadata": {},
   "outputs": [],
   "source": [
    "# def train(epoch):\n",
    "#     model.train()\n",
    "#     for batch_idx, (data, target) in enumerate(train_data_loader):\n",
    "#         optimizer.zero_grad()\n",
    "#         output = model(data)\n",
    "#         loss = F.nll_loss(output, target)\n",
    "#         loss.backward()\n",
    "#         optimizer.step()\n",
    "#         if batch_idx % log_interval == 0:\n",
    "#             print('Train Epoch: {} [{}/{} ({:.0f}%)]\\tLoss: {:.6f}'.format(\n",
    "#                 epoch, batch_idx * len(data), len(train_data_loader.dataset),\n",
    "#                 100. * batch_idx / len(train_data_loader), loss.item()))\n",
    "#             train_losses.append(loss.item())\n",
    "#             train_counter.append(\n",
    "#                 (batch_idx*64) + ((epoch-1)*len(train_data_loader.dataset)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 167,
   "metadata": {},
   "outputs": [],
   "source": [
    "# def test():\n",
    "#     model.eval()\n",
    "#     test_loss = 0\n",
    "#     correct = 0\n",
    "#     with torch.no_grad():\n",
    "#         for data, target in valid_data_loader:\n",
    "#             output = model(data)\n",
    "#             test_loss += F.nll_loss(output, target, size_average=False).item()\n",
    "#             pred = output.data.max(1, keepdim=True)[1]\n",
    "#             correct += pred.eq(target.data.view_as(pred)).sum()\n",
    "#     test_loss /= len(valid_data_loader.dataset)\n",
    "#     test_losses.append(test_loss)\n",
    "#     print('\\nTest set: Avg. loss: {:.4f}, Accuracy: {}/{} ({:.0f}%)\\n'.format(\n",
    "#         test_loss, correct, len(valid_data_loader.dataset),\n",
    "#         100. * correct / len(valid_data_loader.dataset)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 169,
   "metadata": {
    "collapsed": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Test set: Avg. loss: 2.3004, Accuracy: 71/510 (14%)\n",
      "\n",
      "Train Epoch: 1 [0/1040 (0%)]\tLoss: 2.298346\n",
      "Train Epoch: 1 [640/1040 (59%)]\tLoss: 2.301919\n",
      "\n",
      "Test set: Avg. loss: 2.2992, Accuracy: 75/510 (15%)\n",
      "\n",
      "Train Epoch: 2 [0/1040 (0%)]\tLoss: 2.294892\n",
      "Train Epoch: 2 [640/1040 (59%)]\tLoss: 2.301924\n",
      "\n",
      "Test set: Avg. loss: 2.2985, Accuracy: 74/510 (15%)\n",
      "\n",
      "Train Epoch: 3 [0/1040 (0%)]\tLoss: 2.294682\n",
      "Train Epoch: 3 [640/1040 (59%)]\tLoss: 2.296587\n",
      "\n",
      "Test set: Avg. loss: 2.2971, Accuracy: 88/510 (17%)\n",
      "\n",
      "Train Epoch: 4 [0/1040 (0%)]\tLoss: 2.292691\n",
      "Train Epoch: 4 [640/1040 (59%)]\tLoss: 2.301257\n",
      "\n",
      "Test set: Avg. loss: 2.2957, Accuracy: 51/510 (10%)\n",
      "\n",
      "Train Epoch: 5 [0/1040 (0%)]\tLoss: 2.298717\n",
      "Train Epoch: 5 [640/1040 (59%)]\tLoss: 2.296419\n",
      "\n",
      "Test set: Avg. loss: 2.2948, Accuracy: 51/510 (10%)\n",
      "\n",
      "Train Epoch: 6 [0/1040 (0%)]\tLoss: 2.296418\n",
      "Train Epoch: 6 [640/1040 (59%)]\tLoss: 2.291975\n",
      "\n",
      "Test set: Avg. loss: 2.2927, Accuracy: 58/510 (11%)\n",
      "\n",
      "Train Epoch: 7 [0/1040 (0%)]\tLoss: 2.296333\n",
      "Train Epoch: 7 [640/1040 (59%)]\tLoss: 2.292885\n",
      "\n",
      "Test set: Avg. loss: 2.2904, Accuracy: 90/510 (18%)\n",
      "\n",
      "Train Epoch: 8 [0/1040 (0%)]\tLoss: 2.291479\n",
      "Train Epoch: 8 [640/1040 (59%)]\tLoss: 2.292890\n",
      "\n",
      "Test set: Avg. loss: 2.2883, Accuracy: 61/510 (12%)\n",
      "\n",
      "Train Epoch: 9 [0/1040 (0%)]\tLoss: 2.287119\n",
      "Train Epoch: 9 [640/1040 (59%)]\tLoss: 2.296059\n",
      "\n",
      "Test set: Avg. loss: 2.2853, Accuracy: 51/510 (10%)\n",
      "\n",
      "Train Epoch: 10 [0/1040 (0%)]\tLoss: 2.291048\n",
      "Train Epoch: 10 [640/1040 (59%)]\tLoss: 2.276034\n",
      "\n",
      "Test set: Avg. loss: 2.2815, Accuracy: 166/510 (33%)\n",
      "\n",
      "Train Epoch: 11 [0/1040 (0%)]\tLoss: 2.279611\n",
      "Train Epoch: 11 [640/1040 (59%)]\tLoss: 2.275883\n",
      "\n",
      "Test set: Avg. loss: 2.2779, Accuracy: 64/510 (13%)\n",
      "\n",
      "Train Epoch: 12 [0/1040 (0%)]\tLoss: 2.289625\n",
      "Train Epoch: 12 [640/1040 (59%)]\tLoss: 2.267743\n",
      "\n",
      "Test set: Avg. loss: 2.2703, Accuracy: 185/510 (36%)\n",
      "\n",
      "Train Epoch: 13 [0/1040 (0%)]\tLoss: 2.267324\n",
      "Train Epoch: 13 [640/1040 (59%)]\tLoss: 2.272429\n",
      "\n",
      "Test set: Avg. loss: 2.2609, Accuracy: 170/510 (33%)\n",
      "\n",
      "Train Epoch: 14 [0/1040 (0%)]\tLoss: 2.253457\n",
      "Train Epoch: 14 [640/1040 (59%)]\tLoss: 2.259839\n",
      "\n",
      "Test set: Avg. loss: 2.2511, Accuracy: 212/510 (42%)\n",
      "\n",
      "Train Epoch: 15 [0/1040 (0%)]\tLoss: 2.255338\n",
      "Train Epoch: 15 [640/1040 (59%)]\tLoss: 2.229632\n",
      "\n",
      "Test set: Avg. loss: 2.2354, Accuracy: 55/510 (11%)\n",
      "\n",
      "Train Epoch: 16 [0/1040 (0%)]\tLoss: 2.242553\n",
      "Train Epoch: 16 [640/1040 (59%)]\tLoss: 2.230123\n",
      "\n",
      "Test set: Avg. loss: 2.1993, Accuracy: 162/510 (32%)\n",
      "\n",
      "Train Epoch: 17 [0/1040 (0%)]\tLoss: 2.193167\n",
      "Train Epoch: 17 [640/1040 (59%)]\tLoss: 2.202038\n",
      "\n",
      "Test set: Avg. loss: 2.1497, Accuracy: 174/510 (34%)\n",
      "\n",
      "Train Epoch: 18 [0/1040 (0%)]\tLoss: 2.167885\n",
      "Train Epoch: 18 [640/1040 (59%)]\tLoss: 2.107428\n",
      "\n",
      "Test set: Avg. loss: 2.1543, Accuracy: 100/510 (20%)\n",
      "\n",
      "Train Epoch: 19 [0/1040 (0%)]\tLoss: 2.104742\n",
      "Train Epoch: 19 [640/1040 (59%)]\tLoss: 1.958378\n",
      "\n",
      "Test set: Avg. loss: 1.9082, Accuracy: 155/510 (30%)\n",
      "\n",
      "Train Epoch: 20 [0/1040 (0%)]\tLoss: 1.960652\n",
      "Train Epoch: 20 [640/1040 (59%)]\tLoss: 1.717282\n",
      "\n",
      "Test set: Avg. loss: 1.7977, Accuracy: 262/510 (51%)\n",
      "\n",
      "Train Epoch: 21 [0/1040 (0%)]\tLoss: 1.806412\n",
      "Train Epoch: 21 [640/1040 (59%)]\tLoss: 1.687969\n",
      "\n",
      "Test set: Avg. loss: 1.4352, Accuracy: 283/510 (55%)\n",
      "\n",
      "Train Epoch: 22 [0/1040 (0%)]\tLoss: 1.327237\n",
      "Train Epoch: 22 [640/1040 (59%)]\tLoss: 1.391162\n",
      "\n",
      "Test set: Avg. loss: 1.3530, Accuracy: 277/510 (54%)\n",
      "\n",
      "Train Epoch: 23 [0/1040 (0%)]\tLoss: 1.349794\n",
      "Train Epoch: 23 [640/1040 (59%)]\tLoss: 1.146442\n",
      "\n",
      "Test set: Avg. loss: 1.1426, Accuracy: 299/510 (59%)\n",
      "\n",
      "Train Epoch: 24 [0/1040 (0%)]\tLoss: 1.017065\n",
      "Train Epoch: 24 [640/1040 (59%)]\tLoss: 0.848737\n",
      "\n",
      "Test set: Avg. loss: 1.0399, Accuracy: 324/510 (64%)\n",
      "\n",
      "Train Epoch: 25 [0/1040 (0%)]\tLoss: 1.115489\n",
      "Train Epoch: 25 [640/1040 (59%)]\tLoss: 0.862808\n",
      "\n",
      "Test set: Avg. loss: 1.5865, Accuracy: 214/510 (42%)\n",
      "\n",
      "Train Epoch: 26 [0/1040 (0%)]\tLoss: 1.490536\n",
      "Train Epoch: 26 [640/1040 (59%)]\tLoss: 0.943281\n",
      "\n",
      "Test set: Avg. loss: 0.8321, Accuracy: 359/510 (70%)\n",
      "\n",
      "Train Epoch: 27 [0/1040 (0%)]\tLoss: 0.939980\n",
      "Train Epoch: 27 [640/1040 (59%)]\tLoss: 0.624861\n",
      "\n",
      "Test set: Avg. loss: 1.0732, Accuracy: 297/510 (58%)\n",
      "\n",
      "Train Epoch: 28 [0/1040 (0%)]\tLoss: 0.954111\n",
      "Train Epoch: 28 [640/1040 (59%)]\tLoss: 0.829519\n",
      "\n",
      "Test set: Avg. loss: 0.7093, Accuracy: 406/510 (80%)\n",
      "\n",
      "Train Epoch: 29 [0/1040 (0%)]\tLoss: 0.680849\n",
      "Train Epoch: 29 [640/1040 (59%)]\tLoss: 0.599825\n",
      "\n",
      "Test set: Avg. loss: 0.6417, Accuracy: 392/510 (77%)\n",
      "\n",
      "Train Epoch: 30 [0/1040 (0%)]\tLoss: 0.730152\n",
      "Train Epoch: 30 [640/1040 (59%)]\tLoss: 0.648065\n",
      "\n",
      "Test set: Avg. loss: 0.7268, Accuracy: 379/510 (74%)\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# n_epochs = 30\n",
    "# batch_size_train = 64\n",
    "# batch_size_test = 1000\n",
    "# learning_rate = 0.01\n",
    "# momentum = 0.5\n",
    "# log_interval = 10\n",
    "\n",
    "# train_losses = []\n",
    "# train_counter = []\n",
    "# test_losses = []\n",
    "# test_counter = [i*len(train_data_loader.dataset) for i in range(n_epochs + 1)]\n",
    "\n",
    "# test()\n",
    "# for epoch in range(1, n_epochs + 1):\n",
    "#     train(epoch)\n",
    "#     test()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 174,
   "metadata": {
    "collapsed": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0, 0.5, 'negative log likelihood loss')"
      ]
     },
     "execution_count": 174,
     "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": [
    "# fig = plt.figure()\n",
    "# plt.plot(train_counter, train_losses, color='blue')\n",
    "# plt.scatter(test_counter, test_losses, color='red')\n",
    "# plt.legend(['Train Loss', 'Test Loss'], loc='upper right')\n",
    "# plt.xlabel('number of training examples seen')\n",
    "# plt.ylabel('negative log likelihood loss')"
   ]
  }
 ],
 "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": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}