NN_self_hand.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "「nientzu.ipynb」的副本",
      "provenance": [],
      "collapsed_sections": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/s07362022/CNN_to_train/blob/master/NN_self_hand.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "tDUpvV6Hv1Zs",
        "outputId": "e3356dab-97fc-471b-cb2f-bd0b4af59ea6"
      },
      "source": [
        "from google.colab import drive\n",
        "import os\n",
        "import matplotlib.pyplot as plt\n",
        "drive.mount('/content/drive')\n",
        "project_path='/content/drive/MyDrive/'\n",
        "os.chdir(project_path)"
      ],
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Drive already mounted at /content/drive; to attempt to forcibly remount, call drive.mount(\"/content/drive\", force_remount=True).\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OYYddWRrze2p"
      },
      "source": [
        "import numpy as np\n",
        "def shuffle(x,y):\n",
        "    x = x.T\n",
        "    y = y.T\n",
        "    combine = np.c_[x.reshape(len(x), -1), y.reshape(len(y), -1)]\n",
        "    np.random.shuffle(combine)\n",
        "    return combine[:, :x.size//len(x)].reshape(x.shape).T,combine[:, x.size//len(x):].reshape(y.shape).T\n",
        "class One_layer:    \n",
        "    def __init__(self, in_dim, out_dim, act,lr,apt_type,val):\n",
        "        self.in_dim = in_dim\n",
        "        self.out_dim = out_dim\n",
        "        self.weight = np.random.randn(out_dim,in_dim+1)/10\n",
        "        \n",
        "        #self.weight = np.random.rand(out_dim,in_dim+1)/10\n",
        "        \n",
        "        act_list = {\"relu\":self.sigmoid, \"tanh\":self.tanh,\"relu\":self.relu,\"sigmoid\":self.sigmoid,\n",
        "                    \"softmax\":self.softmax,\"linear\":self.linear}\n",
        "        dif_list = {\"relu\":self.dif_sigmoid, \"tanh\":self.dif_tanh,\"relu\":self.dif_relu,\"sigmoid\":self.dif_sigmoid,\n",
        "                    \"softmax\":self.dif_softmax,\"linear\":self.dif_linear}\n",
        "        self.act = act_list[act]\n",
        "        self.dif_act = dif_list[act]\n",
        "        self.lr = lr\n",
        "        self.apt_type=apt_type\n",
        "        if apt_type=='Momentum':\n",
        "          self.Momentum = val\n",
        "          self.weight_f = np.random.randn(out_dim,in_dim+1)/10\n",
        "        elif apt_type=='AdaGrad':\n",
        "          self.EPS = val\n",
        "          self.weight_f = np.zeros([out_dim,in_dim+1])\n",
        "        #self.batch_size = batch_size\n",
        "        self.input = np.array([]) #np.zeros((batch_size,in_dim))\n",
        "        self.local_grad = np.array([]) #np.zeros((batch_size,out_dim))    \n",
        "        self.value = np.array([])\n",
        "        \n",
        "    def forward(self,data):    \n",
        "        data = np.concatenate((np.ones(data.shape[1]).reshape(1,-1),data), axis=0)         \n",
        "        self.input = data\n",
        "        self.value = self.weight.dot(data)\n",
        "        return self.act(self.value) \n",
        "    \n",
        "    def output(self,data): \n",
        "        data = np.concatenate((np.ones(data.shape[1]).reshape(1,-1),data), axis=0) \n",
        "        return self.act(self.weight.dot(data))    \n",
        "    \n",
        "    def backward(self,err):\n",
        "        self.local_grad = self.dif_act(self.value,err)\n",
        "        return self.weight.T.dot(self.local_grad)[1:,:]    \n",
        "    \n",
        "    def weight_renew(self,batch_size):       \n",
        "        delta_w = self.lr*(self.local_grad).dot(self.input.T)#/batch_size#\n",
        "        if self.apt_type=='None':\n",
        "          self.weight = self.weight - delta_w\n",
        "        elif self.apt_type=='Momentum':\n",
        "          #self.weight = self.weight - delta_w + (self.Momentum * self.weight_f)\n",
        "          self.weight = self.weight + self.weight_f\n",
        "          #self.weight_f = (self.local_grad).dot(self.input.T)\n",
        "          self.weight_f = self.Momentum * self.weight_f - delta_w\n",
        "        elif self.apt_type=='AdaGrad':\n",
        "          self.weight = self.weight - delta_w * (1/np.sqrt(self.weight_f+self.EPS))\n",
        "          self.weight_f = self.weight_f + ((self.local_grad).dot(self.input.T))**2\n",
        "        return self.weight    \n",
        "    def sigmoid(self,z):\n",
        "        return 1/(1+np.exp(-z))\n",
        "    def tanh(self,z):\n",
        "        return (np.exp(z)-np.exp(-z))/(np.exp(z)+np.exp(-z))\n",
        "    def relu(self,z):\n",
        "        return np.maximum(0,z)  \n",
        "    \n",
        "    def linear(self,z):\n",
        "        return z \n",
        "    \n",
        "    def dif_linear(self,z,err):\n",
        "        return err\n",
        "    \n",
        "    def softmax(self,x):\n",
        "        shiftx = x - np.max(x)\n",
        "        exps = np.exp(shiftx)\n",
        "        return exps / np.sum(exps,axis=0)    \n",
        "    \n",
        "    def dif_sigmoid(self,z,err):\n",
        "        temp=self.sigmoid(z)        \n",
        "        return (temp*(1-temp))*err    \n",
        "    def dif_tanh(self,z,err):\n",
        "        return (1-self.tanh(self,z)**2)*err    \n",
        "    def dif_relu(self,z,err):\n",
        "        return np.maximum(np.sign(z),0)*err     \n",
        "    def dif_softmax(self,z,err):\n",
        "        local_grad = []\n",
        "        for i,ee in enumerate(z.T):\n",
        "            ee = ee.T.reshape(1,-1)\n",
        "            tt = self.softmax(ee.T)\n",
        "            local_grad.append((np.eye(ee.size)*tt-tt.dot(tt.T)).dot(err[:,i]))\n",
        "        return np.array(local_grad).T#(np.eye(self.out_dim)*S - S.dot(S.T)).dot(err)\n",
        "    def fetch_w(self):\n",
        "        return self.weight\n",
        "\n",
        "class RBF_layer:    \n",
        "    def __init__(self, in_dim, out_dim, cent_lr, wid_lr,data,act=\"gaussian\"):\n",
        "        self.in_dim = in_dim\n",
        "        self.out_dim = out_dim\n",
        "        self.centers = self.plus_plus(data, out_dim)\n",
        "        self.var_sqr_inv = np.ones((out_dim,in_dim))/10\n",
        "        \n",
        "        act_list = {\"gaussian\":self.gaussian, \"inv_quadratic\":self.quadratic}\n",
        "        act_diff_list = {\"gaussian\":self.diff_gaussian, \"inv_quadratic\":self.diff_quadratic}\n",
        "        \n",
        "        self.act = act_list[act]\n",
        "        self.act_diff = act_diff_list[act]\n",
        "        \n",
        "        self.cent_lr = cent_lr\n",
        "        self.wid_lr = wid_lr\n",
        "        self.output_norm1 = []\n",
        "        self.out = []\n",
        "        self.err_diff = []\n",
        "    def gaussian(self,data):\n",
        "        return np.exp(-data)\n",
        "    def quadratic(self,data):\n",
        "        return 1/(1+data)**0.5\n",
        "    \n",
        "    def forward(self,data): \n",
        "        o1 = []\n",
        "        for i,cc in enumerate(self.centers):\n",
        "            o1.append((data.T-cc))            \n",
        "        self.output_norm1 = np.array(o1) \n",
        "        \n",
        "        ot = []          \n",
        "        for i,dc in enumerate(np.array(o1)):\n",
        "            ot.append((dc**2).dot(self.var_sqr_inv[i].reshape(-1,1))[:,0])        \n",
        "        temp =  self.act(np.array(ot))\n",
        "        self.out = temp\n",
        "        return temp    \n",
        "    def output(self,data): \n",
        "        o1 = []\n",
        "        for i,cc in enumerate(self.centers):\n",
        "            o1.append((data.T-cc))             \n",
        "        ot = []           \n",
        "        for i,dc in enumerate(np.array(o1)):\n",
        "            ot.append((dc**2).dot(self.var_sqr_inv[i].reshape(-1,1))[:,0])  \n",
        "        temp =  self.act(np.array(ot))\n",
        "        #self.out = temp\n",
        "        return temp  \n",
        "    \n",
        "    def backward(self,err):\n",
        "        self.err_diff = err\n",
        "        return 0 \n",
        "    \n",
        "    def weight_renew(self,batch_size):   \n",
        "        self.act_diff()\n",
        "    \n",
        "    def diff_quadratic(self):\n",
        "        delta_c = []\n",
        "        for i,dc in enumerate(self.output_norm1):\n",
        "            delta_c.append(-(2*dc.T*self.var_sqr_inv[i].reshape(-1,1)).dot((self.err_diff[i]*-0.5*self.out[i]**3).reshape(-1,1))[:,0])\n",
        "        delta_c = np.array(delta_c)\n",
        "        \n",
        "        delta_wid = []\n",
        "        for i,dc in enumerate(self.output_norm1):\n",
        "            delta_wid.append((self.err_diff[i]*-0.5*self.out[i]**3).dot(dc**2))\n",
        "        delta_wid = np.array(delta_wid)\n",
        "        \n",
        "        self.centers -= self.cent_lr*delta_c\n",
        "        self.var_sqr_inv = np.maximum(0,self.var_sqr_inv- self.wid_lr*delta_wid) \n",
        "        \n",
        "    def diff_gaussian(self):\n",
        "        delta_c = []\n",
        "        for i,dc in enumerate(self.output_norm1):\n",
        "            delta_c.append((2*dc.T*self.var_sqr_inv[i].reshape(-1,1)).dot((self.err_diff[i]*self.out[i]).reshape(-1,1))[:,0])\n",
        "        delta_c = np.array(delta_c)\n",
        "        \n",
        "        delta_wid = []\n",
        "        for i,dc in enumerate(self.output_norm1):\n",
        "            delta_wid.append(-(self.err_diff[i]*self.out[i]).dot(dc**2))\n",
        "        delta_wid = np.array(delta_wid)\n",
        "        \n",
        "        self.centers -= self.cent_lr*delta_c\n",
        "        self.var_sqr_inv = np.maximum(0,self.var_sqr_inv- self.wid_lr*delta_wid) \n",
        "        \n",
        "    def plus_plus(self,ds, k):\n",
        "        centroids = [ds[0]]\n",
        "        for _ in range(1, k):\n",
        "            dist_sq = np.array([min([np.inner(c-x,c-x) for c in centroids]) for x in ds])\n",
        "            probs = dist_sq/dist_sq.sum()\n",
        "            cumulative_probs = probs.cumsum()\n",
        "            r = np.random.rand()\n",
        "            for j, p in enumerate(cumulative_probs):\n",
        "                if r < p:\n",
        "                    i = j\n",
        "                    break\n",
        "            centroids.append(ds[i])\n",
        "        return np.array(centroids)\n",
        "\n",
        "class NN_model:\n",
        "    def __init__(self,layer_list,batch_size,loss):\n",
        "        cost_list = {\"Binary\":self.CE_twoclass, \"Multi_class\":self.CE,\n",
        "                     \"Regression_mse\":self.MSE,\"Regression_mae\":self.MAE}     \n",
        "        cost_dif_list = {\"Binary\":self.dif_ce_twoclass, \"Multi_class\":self.dif_ce,\n",
        "                         \"Regression_mse\":self.dif_mse,\"Regression_mae\":self.dif_mae}\n",
        "        self.layer_list = layer_list\n",
        "        self.batch_size = batch_size\n",
        "        self.loss=cost_list[loss]\n",
        "        self.dif_loss=cost_dif_list[loss]\n",
        "        \n",
        "        acc_type_list = {\"Binary\":self.acc_binary, \"Multi_class\":self.acc_multiclass\n",
        "                         ,\"Regression_mae\":self.acc_regression,\"Regression_mse\":self.acc_regression}\n",
        "        self.loss_func = acc_type_list[loss]\n",
        "        \n",
        "    def output(self,x):\n",
        "        temp = x\n",
        "        for ll in self.layer_list:\n",
        "            temp = ll.output(temp)\n",
        "        return temp\n",
        "    \n",
        "    def forward(self,x):\n",
        "        temp = x\n",
        "        for ll in self.layer_list:\n",
        "            temp = ll.forward(temp)            \n",
        "        return temp\n",
        "    \n",
        "    def back_propagation(self,x):\n",
        "        temp = x\n",
        "        for ll in self.layer_list[::-1]:\n",
        "            temp = ll.backward(temp)\n",
        "        return temp\n",
        "    \n",
        "    def loss_dif(self,y,y_hat):  \n",
        "        return self.dif_loss(y,y_hat)    \n",
        "    \n",
        "    def MSE(self,y,y_hat,scale):\n",
        "        err_avg = np.sum(((y_hat-y)*scale[1])**2,axis=0)\n",
        "        return err_avg,sum(err_avg)/y_hat.shape[1]\n",
        "    \n",
        "    def MAE(self,y,y_hat,scale):\n",
        "        err_avg = np.sum(np.abs((y_hat-y)*scale[1]),axis=0)\n",
        "        return err_avg,sum(err_avg)/y_hat.shape[1]\n",
        "    \n",
        "    def CE_twoclass(self,p,y):\n",
        "        err = -(y*np.log(p)+(1-y)*np.log(1-p))\n",
        "        each_batch_avg_err = np.sum(err,axis=1)/err.shape[1]\n",
        "        total_avg_err = sum(each_batch_avg_err)\n",
        "        return err,total_avg_err\n",
        "    \n",
        "    def CE(self,p,y):\n",
        "        err = -np.log(np.sum(p*y,axis=0))\n",
        "        return err,sum(err)/p.shape[1]\n",
        "    \n",
        "    def dif_ce_twoclass(self,p,y):            \n",
        "        #return (np.sum(-(np.divide(y, p) - np.divide(1 - y, 1 - p)),axis=1)/y.shape[1]).reshape(-1,1) \n",
        "        return (-(np.divide(y, p) - np.divide(1 - y, 1 - p)))\n",
        "    \n",
        "    def dif_ce(self,p,y):\n",
        "        return -np.divide(1,p)*y \n",
        "    \n",
        "    def dif_mae(self,y,y_hat):\n",
        "        #return -(np.sum(y_hat-y,axis=1)/y.shape[1]).reshape(-1,1)\n",
        "        return -np.sign((y_hat-y))\n",
        "    \n",
        "    def dif_mse(self,y,y_hat):\n",
        "        #return -(np.sum(y_hat-y,axis=1)/y.shape[1]).reshape(-1,1)\n",
        "        return -(y_hat-y)\n",
        "    \n",
        "    def update_weight(self):\n",
        "        for ll in self.layer_list:\n",
        "            ll.weight_renew(self.batch_size)\n",
        "        return self.layer_list[len(self.layer_list)-1].weight\n",
        "    def shuffle(self,x,y):\n",
        "        x = x.T\n",
        "        y = y.T\n",
        "        combine = np.c_[x.reshape(len(x), -1), y.reshape(len(y), -1)]\n",
        "        np.random.shuffle(combine)\n",
        "        return combine[:, :x.size//len(x)].reshape(x.shape).T,combine[:, x.size//len(x):].reshape(y.shape).T\n",
        "    \n",
        "    # -------------------------------------- Acc --------------------------------------------------#\n",
        "    def acc_binary(self,ee,epochs,x_train,x_valid,y_train,y_valid,scale=[]):\n",
        "        y_va = self.output(x_valid)\n",
        "        vl = self.loss(y_va,y_valid)[1]\n",
        "        y_tr = self.output(x_train)\n",
        "        tl = self.loss(y_tr,y_train)[1]        \n",
        "        ta = self.bb_acc(y_tr,y_train)\n",
        "        va = self.bb_acc(y_va,y_valid)        \n",
        "        print(\"Epoch\",ee,\"/\",epochs,\"- loss: %.4f\" %tl,\n",
        "            \"- accuracy: %.4f\" %ta,\n",
        "            \"- val_loss: %.4f\" %vl,\n",
        "            \"- val_accuracy: %.4f\" %va)            \n",
        "        return [tl,vl,ta,va]\n",
        "    \n",
        "        out =(yhat> 0.5).astype(int)\n",
        "        return np.sum(np.equal(y.reshape(-1,1),out.reshape(-1,1)))/y.size\n",
        "    \n",
        "    def bb_acc(self,yhat,y):\n",
        "        out =(yhat> 0.5).astype(int)\n",
        "        return np.sum(np.equal(y.reshape(-1,1),out.reshape(-1,1)))/y.size\n",
        "    \n",
        "    def acc_multiclass(self,ee,epochs,x_train,x_valid,y_train,y_valid,scale=[]):\n",
        "        y_va = self.output(x_valid)\n",
        "        vl = self.loss(y_va,y_valid)[1]\n",
        "        y_tr = self.output(x_train)\n",
        "        tl = self.loss(y_tr,y_train)[1]         \n",
        "        ta = self.mc_acc(y_tr,y_train)\n",
        "        va = self.mc_acc(y_va,y_valid)        \n",
        "        print(\"Epoch\",ee,\"/\",epochs,\"- loss: %.4f\" %tl,\n",
        "            \"- accuracy: %.4f\" %ta,\n",
        "            \"- val_loss: %.4f\" %vl,\n",
        "            \"- val_accuracy: %.4f\" %va)            \n",
        "        return [tl,vl,ta,va]\n",
        "    \n",
        "    def mc_acc(self,yhat,y):\n",
        "        y_hat = np.argmax(yhat,axis=0)\n",
        "        yy = np.argmax(y,axis=0)\n",
        "        return np.sum(np.equal(y_hat,yy))/len(yy)\n",
        "    \n",
        "    def acc_regression(self,ee,epochs,x_train,x_valid,y_train,y_valid,scale):\n",
        "        y_va = self.output(x_valid)\n",
        "        vl = self.loss(y_va,y_valid,scale)[1]\n",
        "        y_tr = self.output(x_train)\n",
        "        tl = self.loss(y_tr,y_train,scale)[1] \n",
        "        print(\"Epoch\",ee,\"/\",epochs,\"- loss: %.4f\" %tl,\n",
        "              \"- val_loss: %.2f \" %vl) \n",
        "        return [tl,vl]\n",
        "    \n",
        "    def training(self,x_train,x_valid,y_train,y_valid,epochs,scale=[]):        \n",
        "        batches = int(x_train.shape[1]/self.batch_size)\n",
        "        max_index = x_train.shape[1]    \n",
        "        history = []\n",
        "        w_list = []\n",
        "        for ee in range(epochs):\n",
        "            x_train,y_train = self.shuffle(x_train,y_train) ## 每次training data 的順序打亂\n",
        "            for batch in range(batches):\n",
        "                st = batch*self.batch_size\n",
        "                end = min((st+self.batch_size),max_index)                  \n",
        "                out = self.forward(x_train[:,st:end])\n",
        "                err = self.loss_dif(out,y_train[:,st:end])\n",
        "                self.back_propagation(err)\n",
        "                w_list.append(self.update_weight())                \n",
        "            history.append(self.loss_func(ee,epochs,x_train,x_valid,y_train,y_valid,scale))\n",
        "        return np.array(history),w_list\n",
        "\n"
      ],
      "execution_count": 2,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XD1MbUDKxd5O"
      },
      "source": [
        "import pandas as pd\n",
        "import numpy as np\n",
        "import os\n",
        "\n",
        "\n",
        "class DataLoader:\n",
        "    def __init__(self, file_path=None):\n",
        "        self.__file_df = None\n",
        "\n",
        "        if file_path is not None:\n",
        "            try:\n",
        "                self.load_csv_file(file_path)\n",
        "            except Exception as e:\n",
        "                print(e)\n",
        "\n",
        "    def load_csv_file(self, file_path):\n",
        "        self.__file_df = pd.read_csv(file_path, )\n",
        "        print(\"load_csv_file:ok\")\n",
        "\n",
        "    def process_data(self, normalize='max', label='probability', validation_rate=0.2, balance='mix'):\n",
        "        \"\"\"\n",
        "        data processing\n",
        "        :param normalize:   'mean': (x-u)/std;\n",
        "                            'min_max': (x-min)/(max-min);\n",
        "                            'max': x/max;\n",
        "                            None: do not execute normalize;\n",
        "        :param label:   'binary': shape(1, #), 0: N, 1: Y;\n",
        "                        'probability': shape(2, #);\n",
        "        :param validation_rate: rate of validation data\n",
        "        :param balance:     'over_sample': copy minor data twice\n",
        "                            'down_sample': remove major data until the amount of major data is as same as minor data,\n",
        "                            'mix': mix the above methods\n",
        "                            None: do not execute balance;\n",
        "        \"\"\"\n",
        "        __data = None\n",
        "        __label = None\n",
        "\n",
        "        if self.__file_df is not None:\n",
        "            # try:\n",
        "                # delete 'duration'\n",
        "                #self.__file_df=self.__file_df.drop(columns=['duration'])\n",
        "\n",
        "                # process 'pdays' data\n",
        "                #self.__file_df['pdays']=(self.__file_df['pdays']!=999).astype(int)\n",
        "\n",
        "                # Process data\n",
        "                #print(df.columns.values)\n",
        "                print(self.__file_df)\n",
        "                \n",
        "                for col in self.__file_df.columns.values:\n",
        "                    #print(col)\n",
        "                    \n",
        "                    # Separate label from data\n",
        "                    if col == 'Drug':\n",
        "                        #print(\"OK2\")\n",
        "                        #__label = self.__file_df[col].replace(to_replace=['drugA','drugB','drugC','drugX','DrugY'], value=[0,1,2,3,4])\n",
        "                        if label == 'binary':\n",
        "                            __label = self.__file_df[col].replace(to_replace=['drugA','drugB','drugC','drugX','DrugY'], value=[0,1,2,3,4])#.map({'drugA':0,'drugB':1,'drugC':2,'drugX':3,'DrugY':4})\n",
        "                            #one_hot = pd.get_dummies(__label)\n",
        "                            print(\"__label:OK2\")\n",
        "\n",
        "                        elif label == 'probability':\n",
        "                            __label = pd.get_dummies(self.__file_df[col])\n",
        "                            __category = __label.columns.values\n",
        "                        else:\n",
        "                            # raise print('[ERROR] process_data: error input in label')\n",
        "                            raise Exception(label)\n",
        "                        self.__file_df = self.__file_df.drop(col, axis=1)\n",
        "\n",
        "                    # Process category attribute\n",
        "                    elif self.__file_df[col].dtype == 'object':\n",
        "                        # one-hot encoding\n",
        "                        one_hot = pd.get_dummies(self.__file_df[col], prefix=col)\n",
        "                        self.__file_df = self.__file_df.drop(col, axis=1)\n",
        "                        self.__file_df = self.__file_df.join(one_hot)\n",
        "                        print(\"onehot\")\n",
        "                    # Process numerical attribute\n",
        "                    else:\n",
        "                        self.__file_df[col] = self.normalize(self.__file_df[col], normalize)\n",
        "\n",
        "                __data = self.__file_df.to_numpy()\n",
        "                __label = __label.to_numpy()\n",
        "\n",
        "                # Split data\n",
        "                train_data, train_label, valid_data, valid_label = self.split_data(__data, __label, validation_rate=validation_rate)\n",
        "\n",
        "                # Balance data 如果要使用請擇一使用 !!!!!!!!!!!!!!!!!!!!!!!!! !!!!!!!!!!!!!!!!!!!!\n",
        "                train_data, train_label = self.balance_data(train_data, train_label, balance)\n",
        "\n",
        "                return train_data, train_label, valid_data, valid_label\n",
        "            # except Exception as e:\n",
        "            #     print(e)\n",
        "        else:\n",
        "            print('[ERROR] No file loaded')\n",
        "\n",
        "    @staticmethod\n",
        "    def normalize(col, normalize_type):\n",
        "        if normalize_type == 'max':\n",
        "            return col / col.max()\n",
        "        elif normalize_type == 'mean':\n",
        "            return (col - col.mean()) / col.std()\n",
        "        elif normalize_type == 'min_max':\n",
        "            return (col - col.min()) / (col.max() - col.min())\n",
        "        else:\n",
        "            return col\n",
        "\n",
        "    @staticmethod\n",
        "    def split_data(data, label, validation_rate):\n",
        "        #print(len(label.shape))\n",
        "        if len(label.shape) == 1:\n",
        "            no_indices = np.where(label == 0)[0]\n",
        "            yes_indices = np.where(label == 1)[0]\n",
        "            b =  np.where(label == 2)[0]\n",
        "            c = np.where(label == 3)[0]\n",
        "            d = np.where(label == 4)[0]\n",
        "            #print(d)類別需要改\n",
        "        else:\n",
        "            no_indices = np.where(label[:, 0] == 1)[0]\n",
        "            yes_indices = np.where(label[:, 1] == 1)[0]\n",
        "            b = np.where(label[:, 2] == 1)[0]\n",
        "            c = np.where(label[:, 3] == 1)[0]\n",
        "            d = np.where(label[:, 4] == 1)[0]\n",
        "            print(len(no_indices),len(yes_indices),len(b),len(c),len(d))\n",
        "            #print(\"ok4:\",b) 類別需要改\n",
        "\n",
        "        no_indices = np.random.permutation(no_indices)\n",
        "        yes_indices = np.random.permutation(yes_indices)\n",
        "        b = np.random.permutation(b)\n",
        "        c = np.random.permutation(c)\n",
        "        d = np.random.permutation(d)\n",
        "\n",
        "        valid_indices = np.concatenate((no_indices[:int(len(no_indices) * validation_rate)],\n",
        "                                        yes_indices[:int(len(yes_indices) * validation_rate)],b[:int(len(b) * validation_rate)],c[:int(len(c) * validation_rate)],d[:int(len(d) * validation_rate)]), axis=0)\n",
        "\n",
        "        train_indices = np.concatenate((no_indices[int(len(no_indices) * validation_rate):],\n",
        "                                        yes_indices[int(len(yes_indices) * validation_rate):],b[int(len(b) * validation_rate):],c[int(len(c) * validation_rate):],d[int(len(d) * validation_rate):]), axis=0)\n",
        "\n",
        "        train_indices = np.random.permutation(train_indices)\n",
        "        valid_indices = np.random.permutation(valid_indices)\n",
        "\n",
        "        train_data = data[train_indices]\n",
        "        valid_data = data[valid_indices]\n",
        "\n",
        "        if len(label.shape) == 1:\n",
        "            train_label = label[train_indices, np.newaxis]\n",
        "            valid_label = label[valid_indices, np.newaxis]\n",
        "        else:\n",
        "            train_label = label[train_indices]\n",
        "            valid_label = label[valid_indices]\n",
        "\n",
        "        return train_data, train_label, valid_data, valid_label\n",
        "\n",
        "    @staticmethod\n",
        "    #######第一個增加原始資料量比較少的資料,使其各個資料量相差不多!!!!!!!!!!!!!!!!\n",
        "    ######第二個降低原始資料量比較多的資料,使其各個資料量相差不多###########\n",
        "    def balance_data(data, label, method):\n",
        "        if method is not None:\n",
        "            if label.shape[1] == 1: # for binary label\n",
        "                no_indices = np.where(label[:, 0] == 0)[0]\n",
        "                yes_indices = np.where(label[:, 0] == 1)[0]\n",
        "                print(\"labe=1\")\n",
        "                \n",
        "            else: # for probability label\n",
        "                no_indices = np.where(label[:, 0] == 1)[0]\n",
        "                yes_indices = np.where(label[:, 1] == 1)[0]\n",
        "                b = np.where(label[:, 2] == 1)[0]\n",
        "                c = np.where(label[:, 3] == 1)[0]\n",
        "                d = np.where(label[:, 4] == 1)[0]\n",
        "                #print(\"labe!=1\")\n",
        "                #print(len(no_indices),len(yes_indices),len(b),len(c),len(d))\n",
        "\n",
        "            if method == 'over_sample' or method == 'mix':\n",
        "                if len(d) > len(yes_indices):\n",
        "                    yes_indices = np.concatenate((yes_indices, yes_indices), axis=0)\n",
        "                    yes_indices = np.concatenate((yes_indices, yes_indices), axis=0)\n",
        "                    b = np.concatenate((b, b), axis=0)\n",
        "                    b = np.concatenate((b, b), axis=0)\n",
        "                    c = np.concatenate((c, c), axis=0)\n",
        "                    c = np.concatenate((c, c), axis=0)\n",
        "\n",
        "                    #print(\"YES\")\n",
        "                #else:\n",
        "                    #d = np.concatenate((d, d), axis=0)\n",
        "            #if method == 'down_sample' or method == 'mix':\n",
        "                #if len(no_indices) > len(yes_indices):\n",
        "                \n",
        "                    #no_indices = no_indices[: len(yes_indices)]\n",
        "                    #d = d[: len(yes_indices)]\n",
        "                    #print(\"len(no_indices)\",len(no_indices))\n",
        "                    \n",
        "                #else:\n",
        "                    #yes_indices = yes_indices[: len(d)]\n",
        "\n",
        "            indices = np.concatenate((d, yes_indices,c,no_indices,b), axis=0)\n",
        "            indices = np.random.permutation(indices)\n",
        "            print(len(no_indices),len(yes_indices),len(b),len(c),len(d))\n",
        "\n",
        "            data = data[indices]\n",
        "            label = label[indices]\n",
        "\n",
        "        return data, label\n"
      ],
      "execution_count": 3,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "tvRgF_OwwhuU",
        "outputId": "0511bfbc-3fa1-4197-87c0-91b45c202c2a"
      },
      "source": [
        "# from data_loader import DataLoader\n",
        "\n",
        "dl = DataLoader(project_path+\"drug200.csv\")\n",
        "train_data, train_label, valid_data, valid_label = dl.process_data(label='probability', balance='mix')\n",
        "train_data=train_data.T\n",
        "valid_data=valid_data.T\n",
        "train_label=train_label.T\n",
        "valid_label=valid_label.T\n",
        "print(train_data.shape)\n",
        "print(train_label.shape)\n",
        "print(valid_data.shape)\n",
        "print(valid_label.shape)"
      ],
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "load_csv_file:ok\n",
            "     Age Sex      BP Cholesterol  Na_to_K   Drug\n",
            "0     23   F    HIGH        HIGH   25.355  DrugY\n",
            "1     47   M     LOW        HIGH   13.093  drugC\n",
            "2     47   M     LOW        HIGH   10.114  drugC\n",
            "3     28   F  NORMAL        HIGH    7.798  drugX\n",
            "4     61   F     LOW        HIGH   18.043  DrugY\n",
            "..   ...  ..     ...         ...      ...    ...\n",
            "195   56   F     LOW        HIGH   11.567  drugC\n",
            "196   16   M     LOW        HIGH   12.006  drugC\n",
            "197   52   M  NORMAL        HIGH    9.894  drugX\n",
            "198   23   M  NORMAL      NORMAL   14.020  drugX\n",
            "199   40   F     LOW      NORMAL   11.349  drugX\n",
            "\n",
            "[200 rows x 6 columns]\n",
            "onehot\n",
            "onehot\n",
            "onehot\n",
            "91 23 16 16 54\n",
            "73 76 52 52 44\n",
            "(9, 297)\n",
            "(5, 297)\n",
            "(9, 38)\n",
            "(5, 38)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "U8i7EzIiKg-d",
        "outputId": "0b623e72-3a7a-4e2a-c7d5-2ee55158d296"
      },
      "source": [
        "train_label[0]"
      ],
      "execution_count": 201,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([1, 1, 1, 0, 0, 1, 1, 1, 1, 1, 0, 1, 0, 1, 0, 1, 1, 1, 0, 1, 0, 0,\n",
              "       0, 1, 0, 1, 0, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1,\n",
              "       0, 1, 1, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 1, 0, 1, 0, 0,\n",
              "       1, 1, 0, 0, 0, 1, 1, 1, 0, 0, 0, 0, 1, 0, 0, 1, 0, 1, 0, 1, 1, 0,\n",
              "       1, 0, 0, 1, 0, 1, 0, 0, 0, 1, 0, 1, 0, 0, 1, 0, 0, 0, 1, 0, 1, 1,\n",
              "       1, 0, 0, 0, 1, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 1, 0, 0, 1, 0, 0, 1,\n",
              "       1, 0, 1, 1, 1, 1, 1, 1, 0, 1, 0, 1, 0, 0, 1, 0, 1, 1, 0, 0, 0, 1,\n",
              "       1, 1, 0, 0, 0, 1, 1, 1], dtype=uint8)"
            ]
          },
          "metadata": {},
          "execution_count": 201
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jmdPgVuL77lL"
      },
      "source": [
        "df = pd.read_csv(project_path+\"drug200.csv\")"
      ],
      "execution_count": 16,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 423
        },
        "id": "BxBWpQ_I7kMR",
        "outputId": "3d9f6656-f7c6-4171-82df-29b4f75b2a64"
      },
      "source": [
        "df"
      ],
      "execution_count": 202,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>Age</th>\n",
              "      <th>Sex</th>\n",
              "      <th>BP</th>\n",
              "      <th>Cholesterol</th>\n",
              "      <th>Na_to_K</th>\n",
              "      <th>Drug</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>23</td>\n",
              "      <td>F</td>\n",
              "      <td>HIGH</td>\n",
              "      <td>HIGH</td>\n",
              "      <td>25.355</td>\n",
              "      <td>DrugY</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>47</td>\n",
              "      <td>M</td>\n",
              "      <td>LOW</td>\n",
              "      <td>HIGH</td>\n",
              "      <td>13.093</td>\n",
              "      <td>drugC</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>47</td>\n",
              "      <td>M</td>\n",
              "      <td>LOW</td>\n",
              "      <td>HIGH</td>\n",
              "      <td>10.114</td>\n",
              "      <td>drugC</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>28</td>\n",
              "      <td>F</td>\n",
              "      <td>NORMAL</td>\n",
              "      <td>HIGH</td>\n",
              "      <td>7.798</td>\n",
              "      <td>drugX</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>61</td>\n",
              "      <td>F</td>\n",
              "      <td>LOW</td>\n",
              "      <td>HIGH</td>\n",
              "      <td>18.043</td>\n",
              "      <td>DrugY</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>195</th>\n",
              "      <td>56</td>\n",
              "      <td>F</td>\n",
              "      <td>LOW</td>\n",
              "      <td>HIGH</td>\n",
              "      <td>11.567</td>\n",
              "      <td>drugC</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>196</th>\n",
              "      <td>16</td>\n",
              "      <td>M</td>\n",
              "      <td>LOW</td>\n",
              "      <td>HIGH</td>\n",
              "      <td>12.006</td>\n",
              "      <td>drugC</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>197</th>\n",
              "      <td>52</td>\n",
              "      <td>M</td>\n",
              "      <td>NORMAL</td>\n",
              "      <td>HIGH</td>\n",
              "      <td>9.894</td>\n",
              "      <td>drugX</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>198</th>\n",
              "      <td>23</td>\n",
              "      <td>M</td>\n",
              "      <td>NORMAL</td>\n",
              "      <td>NORMAL</td>\n",
              "      <td>14.020</td>\n",
              "      <td>drugX</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>199</th>\n",
              "      <td>40</td>\n",
              "      <td>F</td>\n",
              "      <td>LOW</td>\n",
              "      <td>NORMAL</td>\n",
              "      <td>11.349</td>\n",
              "      <td>drugX</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>200 rows × 6 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "     Age Sex      BP Cholesterol  Na_to_K   Drug\n",
              "0     23   F    HIGH        HIGH   25.355  DrugY\n",
              "1     47   M     LOW        HIGH   13.093  drugC\n",
              "2     47   M     LOW        HIGH   10.114  drugC\n",
              "3     28   F  NORMAL        HIGH    7.798  drugX\n",
              "4     61   F     LOW        HIGH   18.043  DrugY\n",
              "..   ...  ..     ...         ...      ...    ...\n",
              "195   56   F     LOW        HIGH   11.567  drugC\n",
              "196   16   M     LOW        HIGH   12.006  drugC\n",
              "197   52   M  NORMAL        HIGH    9.894  drugX\n",
              "198   23   M  NORMAL      NORMAL   14.020  drugX\n",
              "199   40   F     LOW      NORMAL   11.349  drugX\n",
              "\n",
              "[200 rows x 6 columns]"
            ]
          },
          "metadata": {},
          "execution_count": 202
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "9NGEF7-U26Za",
        "outputId": "766b94f3-0c8d-4b2e-a362-b3178f754469"
      },
      "source": [
        "out_history=[]\n",
        "all_w=[]\n",
        "eta=0.02\n",
        "model2 = NN_model([\n",
        "            One_layer(in_dim=train_data.shape[0],out_dim=32,act=\"sigmoid\",lr=eta,apt_type='AdaGrad',val=0.2),\n",
        "            One_layer(in_dim=32,out_dim=5,act=\"softmax\",lr=eta,apt_type='AdaGrad',val=0.1)],\n",
        "            batch_size = 4,loss='Multi_class'\n",
        "    ) \n",
        "history,all_w = model2.training(train_data,valid_data,train_label,valid_label,epochs=1000)\n",
        "out_history.append(history)\n",
        "\n",
        "def draw(hh,title):\n",
        "    fig = plt.figure(figsize=(16,10))\n",
        "    fig.subplots_adjust(hspace=0.25, wspace=0.2)\n",
        "    plt.subplot(2,2,1)\n",
        "    for i,h in enumerate(hh):\n",
        "        plt.plot(h.T[2])\n",
        "    plt.title('Training accuracy')\n",
        "    plt.ylabel('Accuracy')\n",
        "    plt.xlabel('Epoch')\n",
        "\n",
        "    plt.subplot(2,2,2)\n",
        "    for i,h in enumerate(hh):\n",
        "        plt.plot(h.T[3])\n",
        "    plt.title('Validation accuracy')\n",
        "    plt.ylabel('Accuracy')\n",
        "    plt.xlabel('Epoch')\n",
        "    plt.subplot(2,2,3)\n",
        "    for i,h in enumerate(hh):\n",
        "        plt.plot(h.T[0])\n",
        "    plt.title('Training loss')\n",
        "    plt.ylabel('Loss')\n",
        "    plt.xlabel('Epoch')\n",
        "\n",
        "    plt.subplot(2,2,4)\n",
        "    for i,h in enumerate(hh):\n",
        "        plt.plot(h.T[1])\n",
        "    plt.title('Validation loss')\n",
        "    plt.ylabel('Loss')\n",
        "    plt.xlabel('Epoch')\n",
        "\n",
        "    fig.suptitle(title, fontsize=16)\n",
        "    plt.show()\n",
        "draw(out_history,\"AdaGrad\")"
      ],
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 0 / 1000 - loss: 1.5734 - accuracy: 0.2424 - val_loss: 1.5351 - val_accuracy: 0.4737\n",
            "Epoch 1 / 1000 - loss: 1.5512 - accuracy: 0.2559 - val_loss: 1.5851 - val_accuracy: 0.1053\n",
            "Epoch 2 / 1000 - loss: 1.5292 - accuracy: 0.2694 - val_loss: 1.5701 - val_accuracy: 0.1316\n",
            "Epoch 3 / 1000 - loss: 1.5057 - accuracy: 0.4040 - val_loss: 1.5285 - val_accuracy: 0.3684\n",
            "Epoch 4 / 1000 - loss: 1.4822 - accuracy: 0.4007 - val_loss: 1.5275 - val_accuracy: 0.3684\n",
            "Epoch 5 / 1000 - loss: 1.4570 - accuracy: 0.4680 - val_loss: 1.5188 - val_accuracy: 0.3947\n",
            "Epoch 6 / 1000 - loss: 1.4300 - accuracy: 0.5455 - val_loss: 1.5016 - val_accuracy: 0.3684\n",
            "Epoch 7 / 1000 - loss: 1.4026 - accuracy: 0.4276 - val_loss: 1.4577 - val_accuracy: 0.3947\n",
            "Epoch 8 / 1000 - loss: 1.3748 - accuracy: 0.5455 - val_loss: 1.4402 - val_accuracy: 0.3684\n",
            "Epoch 9 / 1000 - loss: 1.3472 - accuracy: 0.5387 - val_loss: 1.4300 - val_accuracy: 0.3421\n",
            "Epoch 10 / 1000 - loss: 1.3201 - accuracy: 0.5387 - val_loss: 1.4165 - val_accuracy: 0.3421\n",
            "Epoch 11 / 1000 - loss: 1.2937 - accuracy: 0.5387 - val_loss: 1.4055 - val_accuracy: 0.3421\n",
            "Epoch 12 / 1000 - loss: 1.2681 - accuracy: 0.5387 - val_loss: 1.3886 - val_accuracy: 0.3421\n",
            "Epoch 13 / 1000 - loss: 1.2440 - accuracy: 0.5387 - val_loss: 1.3675 - val_accuracy: 0.3421\n",
            "Epoch 14 / 1000 - loss: 1.2208 - accuracy: 0.5387 - val_loss: 1.3504 - val_accuracy: 0.3421\n",
            "Epoch 15 / 1000 - loss: 1.1990 - accuracy: 0.5387 - val_loss: 1.3365 - val_accuracy: 0.3421\n",
            "Epoch 16 / 1000 - loss: 1.1784 - accuracy: 0.5387 - val_loss: 1.3213 - val_accuracy: 0.3421\n",
            "Epoch 17 / 1000 - loss: 1.1591 - accuracy: 0.5387 - val_loss: 1.3041 - val_accuracy: 0.3421\n",
            "Epoch 18 / 1000 - loss: 1.1408 - accuracy: 0.5387 - val_loss: 1.2895 - val_accuracy: 0.3421\n",
            "Epoch 19 / 1000 - loss: 1.1237 - accuracy: 0.5387 - val_loss: 1.2865 - val_accuracy: 0.3421\n",
            "Epoch 20 / 1000 - loss: 1.1077 - accuracy: 0.5387 - val_loss: 1.2698 - val_accuracy: 0.3421\n",
            "Epoch 21 / 1000 - loss: 1.0926 - accuracy: 0.5387 - val_loss: 1.2625 - val_accuracy: 0.3421\n",
            "Epoch 22 / 1000 - loss: 1.0783 - accuracy: 0.5387 - val_loss: 1.2424 - val_accuracy: 0.3421\n",
            "Epoch 23 / 1000 - loss: 1.0650 - accuracy: 0.5387 - val_loss: 1.2353 - val_accuracy: 0.3421\n",
            "Epoch 24 / 1000 - loss: 1.0524 - accuracy: 0.5387 - val_loss: 1.2193 - val_accuracy: 0.3421\n",
            "Epoch 25 / 1000 - loss: 1.0405 - accuracy: 0.5387 - val_loss: 1.2064 - val_accuracy: 0.3421\n",
            "Epoch 26 / 1000 - loss: 1.0293 - accuracy: 0.5387 - val_loss: 1.1973 - val_accuracy: 0.3421\n",
            "Epoch 27 / 1000 - loss: 1.0187 - accuracy: 0.5387 - val_loss: 1.1960 - val_accuracy: 0.3684\n",
            "Epoch 28 / 1000 - loss: 1.0087 - accuracy: 0.5387 - val_loss: 1.1859 - val_accuracy: 0.3421\n",
            "Epoch 29 / 1000 - loss: 0.9992 - accuracy: 0.5556 - val_loss: 1.1797 - val_accuracy: 0.3684\n",
            "Epoch 30 / 1000 - loss: 0.9902 - accuracy: 0.5657 - val_loss: 1.1703 - val_accuracy: 0.3684\n",
            "Epoch 31 / 1000 - loss: 0.9816 - accuracy: 0.5623 - val_loss: 1.1577 - val_accuracy: 0.3684\n",
            "Epoch 32 / 1000 - loss: 0.9734 - accuracy: 0.5657 - val_loss: 1.1503 - val_accuracy: 0.3684\n",
            "Epoch 33 / 1000 - loss: 0.9657 - accuracy: 0.5690 - val_loss: 1.1414 - val_accuracy: 0.3684\n",
            "Epoch 34 / 1000 - loss: 0.9583 - accuracy: 0.5960 - val_loss: 1.1375 - val_accuracy: 0.3684\n",
            "Epoch 35 / 1000 - loss: 0.9512 - accuracy: 0.5960 - val_loss: 1.1319 - val_accuracy: 0.3947\n",
            "Epoch 36 / 1000 - loss: 0.9445 - accuracy: 0.5960 - val_loss: 1.1217 - val_accuracy: 0.3947\n",
            "Epoch 37 / 1000 - loss: 0.9381 - accuracy: 0.5993 - val_loss: 1.1211 - val_accuracy: 0.3947\n",
            "Epoch 38 / 1000 - loss: 0.9320 - accuracy: 0.6094 - val_loss: 1.1112 - val_accuracy: 0.3947\n",
            "Epoch 39 / 1000 - loss: 0.9260 - accuracy: 0.6027 - val_loss: 1.1030 - val_accuracy: 0.3947\n",
            "Epoch 40 / 1000 - loss: 0.9203 - accuracy: 0.5960 - val_loss: 1.0995 - val_accuracy: 0.3684\n",
            "Epoch 41 / 1000 - loss: 0.9147 - accuracy: 0.5960 - val_loss: 1.0949 - val_accuracy: 0.3684\n",
            "Epoch 42 / 1000 - loss: 0.9095 - accuracy: 0.5993 - val_loss: 1.0878 - val_accuracy: 0.3684\n",
            "Epoch 43 / 1000 - loss: 0.9044 - accuracy: 0.5960 - val_loss: 1.0834 - val_accuracy: 0.3947\n",
            "Epoch 44 / 1000 - loss: 0.8995 - accuracy: 0.5960 - val_loss: 1.0783 - val_accuracy: 0.4474\n",
            "Epoch 45 / 1000 - loss: 0.8948 - accuracy: 0.5960 - val_loss: 1.0724 - val_accuracy: 0.4211\n",
            "Epoch 46 / 1000 - loss: 0.8903 - accuracy: 0.5960 - val_loss: 1.0662 - val_accuracy: 0.4211\n",
            "Epoch 47 / 1000 - loss: 0.8859 - accuracy: 0.5960 - val_loss: 1.0613 - val_accuracy: 0.4211\n",
            "Epoch 48 / 1000 - loss: 0.8817 - accuracy: 0.5960 - val_loss: 1.0574 - val_accuracy: 0.4211\n",
            "Epoch 49 / 1000 - loss: 0.8776 - accuracy: 0.5993 - val_loss: 1.0536 - val_accuracy: 0.4474\n",
            "Epoch 50 / 1000 - loss: 0.8737 - accuracy: 0.6027 - val_loss: 1.0506 - val_accuracy: 0.4474\n",
            "Epoch 51 / 1000 - loss: 0.8698 - accuracy: 0.5993 - val_loss: 1.0467 - val_accuracy: 0.4474\n",
            "Epoch 52 / 1000 - loss: 0.8661 - accuracy: 0.5993 - val_loss: 1.0415 - val_accuracy: 0.4474\n",
            "Epoch 53 / 1000 - loss: 0.8625 - accuracy: 0.5993 - val_loss: 1.0367 - val_accuracy: 0.4474\n",
            "Epoch 54 / 1000 - loss: 0.8589 - accuracy: 0.6027 - val_loss: 1.0299 - val_accuracy: 0.4474\n",
            "Epoch 55 / 1000 - loss: 0.8555 - accuracy: 0.6128 - val_loss: 1.0256 - val_accuracy: 0.4474\n",
            "Epoch 56 / 1000 - loss: 0.8522 - accuracy: 0.6027 - val_loss: 1.0238 - val_accuracy: 0.4474\n",
            "Epoch 57 / 1000 - loss: 0.8490 - accuracy: 0.6128 - val_loss: 1.0229 - val_accuracy: 0.4737\n",
            "Epoch 58 / 1000 - loss: 0.8458 - accuracy: 0.6128 - val_loss: 1.0197 - val_accuracy: 0.4737\n",
            "Epoch 59 / 1000 - loss: 0.8427 - accuracy: 0.6128 - val_loss: 1.0176 - val_accuracy: 0.4737\n",
            "Epoch 60 / 1000 - loss: 0.8397 - accuracy: 0.6128 - val_loss: 1.0138 - val_accuracy: 0.4737\n",
            "Epoch 61 / 1000 - loss: 0.8368 - accuracy: 0.6128 - val_loss: 1.0111 - val_accuracy: 0.4737\n",
            "Epoch 62 / 1000 - loss: 0.8339 - accuracy: 0.6162 - val_loss: 1.0073 - val_accuracy: 0.4737\n",
            "Epoch 63 / 1000 - loss: 0.8311 - accuracy: 0.6094 - val_loss: 1.0036 - val_accuracy: 0.4737\n",
            "Epoch 64 / 1000 - loss: 0.8283 - accuracy: 0.6094 - val_loss: 1.0009 - val_accuracy: 0.4737\n",
            "Epoch 65 / 1000 - loss: 0.8256 - accuracy: 0.6162 - val_loss: 0.9961 - val_accuracy: 0.4737\n",
            "Epoch 66 / 1000 - loss: 0.8229 - accuracy: 0.6162 - val_loss: 0.9908 - val_accuracy: 0.4737\n",
            "Epoch 67 / 1000 - loss: 0.8203 - accuracy: 0.6128 - val_loss: 0.9871 - val_accuracy: 0.4737\n",
            "Epoch 68 / 1000 - loss: 0.8178 - accuracy: 0.6094 - val_loss: 0.9853 - val_accuracy: 0.4737\n",
            "Epoch 69 / 1000 - loss: 0.8153 - accuracy: 0.6128 - val_loss: 0.9815 - val_accuracy: 0.4737\n",
            "Epoch 70 / 1000 - loss: 0.8128 - accuracy: 0.6094 - val_loss: 0.9781 - val_accuracy: 0.4737\n",
            "Epoch 71 / 1000 - loss: 0.8104 - accuracy: 0.6162 - val_loss: 0.9752 - val_accuracy: 0.4737\n",
            "Epoch 72 / 1000 - loss: 0.8081 - accuracy: 0.6195 - val_loss: 0.9719 - val_accuracy: 0.4737\n",
            "Epoch 73 / 1000 - loss: 0.8058 - accuracy: 0.6162 - val_loss: 0.9687 - val_accuracy: 0.4737\n",
            "Epoch 74 / 1000 - loss: 0.8035 - accuracy: 0.6162 - val_loss: 0.9670 - val_accuracy: 0.4737\n",
            "Epoch 75 / 1000 - loss: 0.8012 - accuracy: 0.6162 - val_loss: 0.9651 - val_accuracy: 0.4737\n",
            "Epoch 76 / 1000 - loss: 0.7989 - accuracy: 0.6094 - val_loss: 0.9628 - val_accuracy: 0.4737\n",
            "Epoch 77 / 1000 - loss: 0.7967 - accuracy: 0.6195 - val_loss: 0.9588 - val_accuracy: 0.4737\n",
            "Epoch 78 / 1000 - loss: 0.7946 - accuracy: 0.6195 - val_loss: 0.9564 - val_accuracy: 0.4737\n",
            "Epoch 79 / 1000 - loss: 0.7924 - accuracy: 0.6195 - val_loss: 0.9526 - val_accuracy: 0.4737\n",
            "Epoch 80 / 1000 - loss: 0.7903 - accuracy: 0.6195 - val_loss: 0.9519 - val_accuracy: 0.4737\n",
            "Epoch 81 / 1000 - loss: 0.7882 - accuracy: 0.6195 - val_loss: 0.9480 - val_accuracy: 0.4737\n",
            "Epoch 82 / 1000 - loss: 0.7861 - accuracy: 0.6162 - val_loss: 0.9467 - val_accuracy: 0.4737\n",
            "Epoch 83 / 1000 - loss: 0.7841 - accuracy: 0.6162 - val_loss: 0.9451 - val_accuracy: 0.4737\n",
            "Epoch 84 / 1000 - loss: 0.7821 - accuracy: 0.6195 - val_loss: 0.9419 - val_accuracy: 0.4737\n",
            "Epoch 85 / 1000 - loss: 0.7801 - accuracy: 0.6195 - val_loss: 0.9393 - val_accuracy: 0.4737\n",
            "Epoch 86 / 1000 - loss: 0.7782 - accuracy: 0.6229 - val_loss: 0.9375 - val_accuracy: 0.4737\n",
            "Epoch 87 / 1000 - loss: 0.7762 - accuracy: 0.6229 - val_loss: 0.9365 - val_accuracy: 0.4737\n",
            "Epoch 88 / 1000 - loss: 0.7743 - accuracy: 0.6229 - val_loss: 0.9323 - val_accuracy: 0.4737\n",
            "Epoch 89 / 1000 - loss: 0.7724 - accuracy: 0.6229 - val_loss: 0.9296 - val_accuracy: 0.4737\n",
            "Epoch 90 / 1000 - loss: 0.7705 - accuracy: 0.6296 - val_loss: 0.9260 - val_accuracy: 0.4737\n",
            "Epoch 91 / 1000 - loss: 0.7686 - accuracy: 0.6263 - val_loss: 0.9238 - val_accuracy: 0.4737\n",
            "Epoch 92 / 1000 - loss: 0.7667 - accuracy: 0.6330 - val_loss: 0.9208 - val_accuracy: 0.4737\n",
            "Epoch 93 / 1000 - loss: 0.7649 - accuracy: 0.6364 - val_loss: 0.9195 - val_accuracy: 0.4737\n",
            "Epoch 94 / 1000 - loss: 0.7631 - accuracy: 0.6330 - val_loss: 0.9192 - val_accuracy: 0.4737\n",
            "Epoch 95 / 1000 - loss: 0.7613 - accuracy: 0.6330 - val_loss: 0.9167 - val_accuracy: 0.4737\n",
            "Epoch 96 / 1000 - loss: 0.7595 - accuracy: 0.6364 - val_loss: 0.9137 - val_accuracy: 0.4737\n",
            "Epoch 97 / 1000 - loss: 0.7577 - accuracy: 0.6364 - val_loss: 0.9124 - val_accuracy: 0.4737\n",
            "Epoch 98 / 1000 - loss: 0.7559 - accuracy: 0.6364 - val_loss: 0.9073 - val_accuracy: 0.4737\n",
            "Epoch 99 / 1000 - loss: 0.7542 - accuracy: 0.6364 - val_loss: 0.9068 - val_accuracy: 0.4737\n",
            "Epoch 100 / 1000 - loss: 0.7524 - accuracy: 0.6397 - val_loss: 0.9054 - val_accuracy: 0.4737\n",
            "Epoch 101 / 1000 - loss: 0.7507 - accuracy: 0.6397 - val_loss: 0.9042 - val_accuracy: 0.4737\n",
            "Epoch 102 / 1000 - loss: 0.7490 - accuracy: 0.6397 - val_loss: 0.9019 - val_accuracy: 0.4737\n",
            "Epoch 103 / 1000 - loss: 0.7473 - accuracy: 0.6431 - val_loss: 0.8993 - val_accuracy: 0.4737\n",
            "Epoch 104 / 1000 - loss: 0.7456 - accuracy: 0.6431 - val_loss: 0.8977 - val_accuracy: 0.4737\n",
            "Epoch 105 / 1000 - loss: 0.7439 - accuracy: 0.6465 - val_loss: 0.8942 - val_accuracy: 0.4737\n",
            "Epoch 106 / 1000 - loss: 0.7422 - accuracy: 0.6465 - val_loss: 0.8926 - val_accuracy: 0.5000\n",
            "Epoch 107 / 1000 - loss: 0.7406 - accuracy: 0.6465 - val_loss: 0.8901 - val_accuracy: 0.5000\n",
            "Epoch 108 / 1000 - loss: 0.7389 - accuracy: 0.6431 - val_loss: 0.8899 - val_accuracy: 0.5000\n",
            "Epoch 109 / 1000 - loss: 0.7373 - accuracy: 0.6465 - val_loss: 0.8863 - val_accuracy: 0.5000\n",
            "Epoch 110 / 1000 - loss: 0.7356 - accuracy: 0.6532 - val_loss: 0.8844 - val_accuracy: 0.5000\n",
            "Epoch 111 / 1000 - loss: 0.7340 - accuracy: 0.6532 - val_loss: 0.8810 - val_accuracy: 0.5000\n",
            "Epoch 112 / 1000 - loss: 0.7324 - accuracy: 0.6532 - val_loss: 0.8801 - val_accuracy: 0.5000\n",
            "Epoch 113 / 1000 - loss: 0.7308 - accuracy: 0.6566 - val_loss: 0.8774 - val_accuracy: 0.5000\n",
            "Epoch 114 / 1000 - loss: 0.7292 - accuracy: 0.6566 - val_loss: 0.8746 - val_accuracy: 0.5000\n",
            "Epoch 115 / 1000 - loss: 0.7276 - accuracy: 0.6599 - val_loss: 0.8723 - val_accuracy: 0.5000\n",
            "Epoch 116 / 1000 - loss: 0.7260 - accuracy: 0.6566 - val_loss: 0.8696 - val_accuracy: 0.5000\n",
            "Epoch 117 / 1000 - loss: 0.7244 - accuracy: 0.6599 - val_loss: 0.8684 - val_accuracy: 0.5000\n",
            "Epoch 118 / 1000 - loss: 0.7229 - accuracy: 0.6633 - val_loss: 0.8672 - val_accuracy: 0.5000\n",
            "Epoch 119 / 1000 - loss: 0.7213 - accuracy: 0.6633 - val_loss: 0.8652 - val_accuracy: 0.5000\n",
            "Epoch 120 / 1000 - loss: 0.7198 - accuracy: 0.6599 - val_loss: 0.8640 - val_accuracy: 0.5000\n",
            "Epoch 121 / 1000 - loss: 0.7182 - accuracy: 0.6633 - val_loss: 0.8621 - val_accuracy: 0.5000\n",
            "Epoch 122 / 1000 - loss: 0.7167 - accuracy: 0.6633 - val_loss: 0.8586 - val_accuracy: 0.5263\n",
            "Epoch 123 / 1000 - loss: 0.7151 - accuracy: 0.6667 - val_loss: 0.8562 - val_accuracy: 0.5526\n",
            "Epoch 124 / 1000 - loss: 0.7136 - accuracy: 0.6700 - val_loss: 0.8552 - val_accuracy: 0.5526\n",
            "Epoch 125 / 1000 - loss: 0.7121 - accuracy: 0.6700 - val_loss: 0.8539 - val_accuracy: 0.5526\n",
            "Epoch 126 / 1000 - loss: 0.7106 - accuracy: 0.6700 - val_loss: 0.8516 - val_accuracy: 0.5526\n",
            "Epoch 127 / 1000 - loss: 0.7091 - accuracy: 0.6700 - val_loss: 0.8481 - val_accuracy: 0.5526\n",
            "Epoch 128 / 1000 - loss: 0.7076 - accuracy: 0.6700 - val_loss: 0.8459 - val_accuracy: 0.5526\n",
            "Epoch 129 / 1000 - loss: 0.7061 - accuracy: 0.6700 - val_loss: 0.8444 - val_accuracy: 0.5526\n",
            "Epoch 130 / 1000 - loss: 0.7046 - accuracy: 0.6734 - val_loss: 0.8431 - val_accuracy: 0.5789\n",
            "Epoch 131 / 1000 - loss: 0.7031 - accuracy: 0.6768 - val_loss: 0.8419 - val_accuracy: 0.5789\n",
            "Epoch 132 / 1000 - loss: 0.7016 - accuracy: 0.6768 - val_loss: 0.8401 - val_accuracy: 0.5789\n",
            "Epoch 133 / 1000 - loss: 0.7001 - accuracy: 0.6801 - val_loss: 0.8375 - val_accuracy: 0.5789\n",
            "Epoch 134 / 1000 - loss: 0.6986 - accuracy: 0.6801 - val_loss: 0.8354 - val_accuracy: 0.5789\n",
            "Epoch 135 / 1000 - loss: 0.6972 - accuracy: 0.6801 - val_loss: 0.8329 - val_accuracy: 0.5789\n",
            "Epoch 136 / 1000 - loss: 0.6957 - accuracy: 0.6835 - val_loss: 0.8316 - val_accuracy: 0.5789\n",
            "Epoch 137 / 1000 - loss: 0.6943 - accuracy: 0.6835 - val_loss: 0.8280 - val_accuracy: 0.5789\n",
            "Epoch 138 / 1000 - loss: 0.6928 - accuracy: 0.6835 - val_loss: 0.8261 - val_accuracy: 0.5789\n",
            "Epoch 139 / 1000 - loss: 0.6914 - accuracy: 0.6835 - val_loss: 0.8247 - val_accuracy: 0.5789\n",
            "Epoch 140 / 1000 - loss: 0.6899 - accuracy: 0.6835 - val_loss: 0.8226 - val_accuracy: 0.6053\n",
            "Epoch 141 / 1000 - loss: 0.6885 - accuracy: 0.6835 - val_loss: 0.8217 - val_accuracy: 0.6053\n",
            "Epoch 142 / 1000 - loss: 0.6870 - accuracy: 0.6835 - val_loss: 0.8205 - val_accuracy: 0.6053\n",
            "Epoch 143 / 1000 - loss: 0.6856 - accuracy: 0.6835 - val_loss: 0.8187 - val_accuracy: 0.6053\n",
            "Epoch 144 / 1000 - loss: 0.6842 - accuracy: 0.6869 - val_loss: 0.8163 - val_accuracy: 0.6053\n",
            "Epoch 145 / 1000 - loss: 0.6828 - accuracy: 0.6869 - val_loss: 0.8144 - val_accuracy: 0.6053\n",
            "Epoch 146 / 1000 - loss: 0.6813 - accuracy: 0.6902 - val_loss: 0.8127 - val_accuracy: 0.6053\n",
            "Epoch 147 / 1000 - loss: 0.6799 - accuracy: 0.6869 - val_loss: 0.8117 - val_accuracy: 0.6053\n",
            "Epoch 148 / 1000 - loss: 0.6786 - accuracy: 0.6902 - val_loss: 0.8103 - val_accuracy: 0.6053\n",
            "Epoch 149 / 1000 - loss: 0.6771 - accuracy: 0.6869 - val_loss: 0.8090 - val_accuracy: 0.6053\n",
            "Epoch 150 / 1000 - loss: 0.6757 - accuracy: 0.6869 - val_loss: 0.8071 - val_accuracy: 0.6053\n",
            "Epoch 151 / 1000 - loss: 0.6743 - accuracy: 0.6902 - val_loss: 0.8039 - val_accuracy: 0.6053\n",
            "Epoch 152 / 1000 - loss: 0.6729 - accuracy: 0.6869 - val_loss: 0.8030 - val_accuracy: 0.6053\n",
            "Epoch 153 / 1000 - loss: 0.6716 - accuracy: 0.6902 - val_loss: 0.8017 - val_accuracy: 0.6053\n",
            "Epoch 154 / 1000 - loss: 0.6702 - accuracy: 0.6902 - val_loss: 0.8006 - val_accuracy: 0.6053\n",
            "Epoch 155 / 1000 - loss: 0.6688 - accuracy: 0.6869 - val_loss: 0.8006 - val_accuracy: 0.6053\n",
            "Epoch 156 / 1000 - loss: 0.6675 - accuracy: 0.6869 - val_loss: 0.7989 - val_accuracy: 0.6053\n",
            "Epoch 157 / 1000 - loss: 0.6661 - accuracy: 0.6902 - val_loss: 0.7959 - val_accuracy: 0.6053\n",
            "Epoch 158 / 1000 - loss: 0.6647 - accuracy: 0.7071 - val_loss: 0.7931 - val_accuracy: 0.6053\n",
            "Epoch 159 / 1000 - loss: 0.6634 - accuracy: 0.7071 - val_loss: 0.7910 - val_accuracy: 0.6053\n",
            "Epoch 160 / 1000 - loss: 0.6620 - accuracy: 0.7071 - val_loss: 0.7891 - val_accuracy: 0.6053\n",
            "Epoch 161 / 1000 - loss: 0.6606 - accuracy: 0.7071 - val_loss: 0.7877 - val_accuracy: 0.6053\n",
            "Epoch 162 / 1000 - loss: 0.6593 - accuracy: 0.7071 - val_loss: 0.7880 - val_accuracy: 0.6053\n",
            "Epoch 163 / 1000 - loss: 0.6580 - accuracy: 0.7071 - val_loss: 0.7839 - val_accuracy: 0.6053\n",
            "Epoch 164 / 1000 - loss: 0.6566 - accuracy: 0.7205 - val_loss: 0.7830 - val_accuracy: 0.6316\n",
            "Epoch 165 / 1000 - loss: 0.6553 - accuracy: 0.7172 - val_loss: 0.7823 - val_accuracy: 0.6316\n",
            "Epoch 166 / 1000 - loss: 0.6539 - accuracy: 0.7172 - val_loss: 0.7804 - val_accuracy: 0.6316\n",
            "Epoch 167 / 1000 - loss: 0.6526 - accuracy: 0.7205 - val_loss: 0.7783 - val_accuracy: 0.6316\n",
            "Epoch 168 / 1000 - loss: 0.6513 - accuracy: 0.7239 - val_loss: 0.7764 - val_accuracy: 0.6316\n",
            "Epoch 169 / 1000 - loss: 0.6500 - accuracy: 0.7239 - val_loss: 0.7739 - val_accuracy: 0.6579\n",
            "Epoch 170 / 1000 - loss: 0.6487 - accuracy: 0.7239 - val_loss: 0.7720 - val_accuracy: 0.6579\n",
            "Epoch 171 / 1000 - loss: 0.6473 - accuracy: 0.7239 - val_loss: 0.7705 - val_accuracy: 0.6316\n",
            "Epoch 172 / 1000 - loss: 0.6460 - accuracy: 0.7239 - val_loss: 0.7695 - val_accuracy: 0.6316\n",
            "Epoch 173 / 1000 - loss: 0.6447 - accuracy: 0.7273 - val_loss: 0.7677 - val_accuracy: 0.6579\n",
            "Epoch 174 / 1000 - loss: 0.6434 - accuracy: 0.7273 - val_loss: 0.7648 - val_accuracy: 0.6579\n",
            "Epoch 175 / 1000 - loss: 0.6421 - accuracy: 0.7273 - val_loss: 0.7636 - val_accuracy: 0.6579\n",
            "Epoch 176 / 1000 - loss: 0.6408 - accuracy: 0.7340 - val_loss: 0.7617 - val_accuracy: 0.6579\n",
            "Epoch 177 / 1000 - loss: 0.6395 - accuracy: 0.7306 - val_loss: 0.7612 - val_accuracy: 0.6579\n",
            "Epoch 178 / 1000 - loss: 0.6382 - accuracy: 0.7340 - val_loss: 0.7585 - val_accuracy: 0.6579\n",
            "Epoch 179 / 1000 - loss: 0.6369 - accuracy: 0.7374 - val_loss: 0.7577 - val_accuracy: 0.6579\n",
            "Epoch 180 / 1000 - loss: 0.6357 - accuracy: 0.7374 - val_loss: 0.7555 - val_accuracy: 0.6579\n",
            "Epoch 181 / 1000 - loss: 0.6344 - accuracy: 0.7374 - val_loss: 0.7544 - val_accuracy: 0.6579\n",
            "Epoch 182 / 1000 - loss: 0.6331 - accuracy: 0.7508 - val_loss: 0.7530 - val_accuracy: 0.6842\n",
            "Epoch 183 / 1000 - loss: 0.6318 - accuracy: 0.7542 - val_loss: 0.7519 - val_accuracy: 0.6842\n",
            "Epoch 184 / 1000 - loss: 0.6306 - accuracy: 0.7407 - val_loss: 0.7510 - val_accuracy: 0.6842\n",
            "Epoch 185 / 1000 - loss: 0.6293 - accuracy: 0.7542 - val_loss: 0.7485 - val_accuracy: 0.6842\n",
            "Epoch 186 / 1000 - loss: 0.6280 - accuracy: 0.7542 - val_loss: 0.7464 - val_accuracy: 0.6842\n",
            "Epoch 187 / 1000 - loss: 0.6268 - accuracy: 0.7542 - val_loss: 0.7450 - val_accuracy: 0.6842\n",
            "Epoch 188 / 1000 - loss: 0.6255 - accuracy: 0.7407 - val_loss: 0.7436 - val_accuracy: 0.6842\n",
            "Epoch 189 / 1000 - loss: 0.6243 - accuracy: 0.7407 - val_loss: 0.7422 - val_accuracy: 0.6842\n",
            "Epoch 190 / 1000 - loss: 0.6230 - accuracy: 0.7542 - val_loss: 0.7396 - val_accuracy: 0.6842\n",
            "Epoch 191 / 1000 - loss: 0.6218 - accuracy: 0.7946 - val_loss: 0.7381 - val_accuracy: 0.6842\n",
            "Epoch 192 / 1000 - loss: 0.6205 - accuracy: 0.7677 - val_loss: 0.7366 - val_accuracy: 0.6842\n",
            "Epoch 193 / 1000 - loss: 0.6193 - accuracy: 0.7946 - val_loss: 0.7349 - val_accuracy: 0.6842\n",
            "Epoch 194 / 1000 - loss: 0.6180 - accuracy: 0.7811 - val_loss: 0.7346 - val_accuracy: 0.6842\n",
            "Epoch 195 / 1000 - loss: 0.6168 - accuracy: 0.7946 - val_loss: 0.7329 - val_accuracy: 0.6842\n",
            "Epoch 196 / 1000 - loss: 0.6156 - accuracy: 0.7946 - val_loss: 0.7316 - val_accuracy: 0.6842\n",
            "Epoch 197 / 1000 - loss: 0.6144 - accuracy: 0.7946 - val_loss: 0.7301 - val_accuracy: 0.6842\n",
            "Epoch 198 / 1000 - loss: 0.6131 - accuracy: 0.7946 - val_loss: 0.7290 - val_accuracy: 0.6842\n",
            "Epoch 199 / 1000 - loss: 0.6119 - accuracy: 0.7811 - val_loss: 0.7275 - val_accuracy: 0.6842\n",
            "Epoch 200 / 1000 - loss: 0.6107 - accuracy: 0.7677 - val_loss: 0.7271 - val_accuracy: 0.7105\n",
            "Epoch 201 / 1000 - loss: 0.6095 - accuracy: 0.7811 - val_loss: 0.7251 - val_accuracy: 0.7105\n",
            "Epoch 202 / 1000 - loss: 0.6083 - accuracy: 0.7542 - val_loss: 0.7236 - val_accuracy: 0.7105\n",
            "Epoch 203 / 1000 - loss: 0.6071 - accuracy: 0.7576 - val_loss: 0.7204 - val_accuracy: 0.7105\n",
            "Epoch 204 / 1000 - loss: 0.6059 - accuracy: 0.7980 - val_loss: 0.7191 - val_accuracy: 0.7105\n",
            "Epoch 205 / 1000 - loss: 0.6047 - accuracy: 0.7980 - val_loss: 0.7173 - val_accuracy: 0.7105\n",
            "Epoch 206 / 1000 - loss: 0.6035 - accuracy: 0.7980 - val_loss: 0.7164 - val_accuracy: 0.7105\n",
            "Epoch 207 / 1000 - loss: 0.6023 - accuracy: 0.8148 - val_loss: 0.7147 - val_accuracy: 0.7105\n",
            "Epoch 208 / 1000 - loss: 0.6011 - accuracy: 0.8182 - val_loss: 0.7132 - val_accuracy: 0.7105\n",
            "Epoch 209 / 1000 - loss: 0.5999 - accuracy: 0.8182 - val_loss: 0.7115 - val_accuracy: 0.7105\n",
            "Epoch 210 / 1000 - loss: 0.5987 - accuracy: 0.8182 - val_loss: 0.7108 - val_accuracy: 0.7105\n",
            "Epoch 211 / 1000 - loss: 0.5975 - accuracy: 0.8182 - val_loss: 0.7095 - val_accuracy: 0.7105\n",
            "Epoch 212 / 1000 - loss: 0.5963 - accuracy: 0.8182 - val_loss: 0.7079 - val_accuracy: 0.7105\n",
            "Epoch 213 / 1000 - loss: 0.5952 - accuracy: 0.8182 - val_loss: 0.7072 - val_accuracy: 0.7105\n",
            "Epoch 214 / 1000 - loss: 0.5940 - accuracy: 0.8182 - val_loss: 0.7064 - val_accuracy: 0.7105\n",
            "Epoch 215 / 1000 - loss: 0.5928 - accuracy: 0.8182 - val_loss: 0.7060 - val_accuracy: 0.7105\n",
            "Epoch 216 / 1000 - loss: 0.5917 - accuracy: 0.8182 - val_loss: 0.7035 - val_accuracy: 0.7105\n",
            "Epoch 217 / 1000 - loss: 0.5905 - accuracy: 0.8215 - val_loss: 0.7022 - val_accuracy: 0.7105\n",
            "Epoch 218 / 1000 - loss: 0.5893 - accuracy: 0.8215 - val_loss: 0.7013 - val_accuracy: 0.7105\n",
            "Epoch 219 / 1000 - loss: 0.5882 - accuracy: 0.8215 - val_loss: 0.6992 - val_accuracy: 0.7105\n",
            "Epoch 220 / 1000 - loss: 0.5870 - accuracy: 0.8215 - val_loss: 0.6983 - val_accuracy: 0.7105\n",
            "Epoch 221 / 1000 - loss: 0.5859 - accuracy: 0.8215 - val_loss: 0.6975 - val_accuracy: 0.7105\n",
            "Epoch 222 / 1000 - loss: 0.5847 - accuracy: 0.8215 - val_loss: 0.6962 - val_accuracy: 0.7105\n",
            "Epoch 223 / 1000 - loss: 0.5836 - accuracy: 0.8215 - val_loss: 0.6954 - val_accuracy: 0.7368\n",
            "Epoch 224 / 1000 - loss: 0.5825 - accuracy: 0.8215 - val_loss: 0.6932 - val_accuracy: 0.7368\n",
            "Epoch 225 / 1000 - loss: 0.5813 - accuracy: 0.8215 - val_loss: 0.6914 - val_accuracy: 0.7368\n",
            "Epoch 226 / 1000 - loss: 0.5802 - accuracy: 0.8249 - val_loss: 0.6891 - val_accuracy: 0.7368\n",
            "Epoch 227 / 1000 - loss: 0.5790 - accuracy: 0.8249 - val_loss: 0.6884 - val_accuracy: 0.7368\n",
            "Epoch 228 / 1000 - loss: 0.5779 - accuracy: 0.8249 - val_loss: 0.6865 - val_accuracy: 0.7368\n",
            "Epoch 229 / 1000 - loss: 0.5768 - accuracy: 0.8249 - val_loss: 0.6857 - val_accuracy: 0.7368\n",
            "Epoch 230 / 1000 - loss: 0.5757 - accuracy: 0.8249 - val_loss: 0.6842 - val_accuracy: 0.7368\n",
            "Epoch 231 / 1000 - loss: 0.5746 - accuracy: 0.8249 - val_loss: 0.6819 - val_accuracy: 0.7368\n",
            "Epoch 232 / 1000 - loss: 0.5735 - accuracy: 0.8249 - val_loss: 0.6803 - val_accuracy: 0.7368\n",
            "Epoch 233 / 1000 - loss: 0.5724 - accuracy: 0.8249 - val_loss: 0.6787 - val_accuracy: 0.7368\n",
            "Epoch 234 / 1000 - loss: 0.5712 - accuracy: 0.8384 - val_loss: 0.6781 - val_accuracy: 0.7368\n",
            "Epoch 235 / 1000 - loss: 0.5701 - accuracy: 0.8418 - val_loss: 0.6780 - val_accuracy: 0.7368\n",
            "Epoch 236 / 1000 - loss: 0.5690 - accuracy: 0.8283 - val_loss: 0.6764 - val_accuracy: 0.7368\n",
            "Epoch 237 / 1000 - loss: 0.5679 - accuracy: 0.8418 - val_loss: 0.6755 - val_accuracy: 0.7368\n",
            "Epoch 238 / 1000 - loss: 0.5668 - accuracy: 0.8418 - val_loss: 0.6742 - val_accuracy: 0.7632\n",
            "Epoch 239 / 1000 - loss: 0.5657 - accuracy: 0.8418 - val_loss: 0.6723 - val_accuracy: 0.7632\n",
            "Epoch 240 / 1000 - loss: 0.5646 - accuracy: 0.8418 - val_loss: 0.6710 - val_accuracy: 0.7632\n",
            "Epoch 241 / 1000 - loss: 0.5635 - accuracy: 0.8418 - val_loss: 0.6701 - val_accuracy: 0.7632\n",
            "Epoch 242 / 1000 - loss: 0.5625 - accuracy: 0.8418 - val_loss: 0.6693 - val_accuracy: 0.7632\n",
            "Epoch 243 / 1000 - loss: 0.5614 - accuracy: 0.8418 - val_loss: 0.6679 - val_accuracy: 0.7632\n",
            "Epoch 244 / 1000 - loss: 0.5603 - accuracy: 0.8418 - val_loss: 0.6659 - val_accuracy: 0.7632\n",
            "Epoch 245 / 1000 - loss: 0.5592 - accuracy: 0.8418 - val_loss: 0.6637 - val_accuracy: 0.7632\n",
            "Epoch 246 / 1000 - loss: 0.5581 - accuracy: 0.8418 - val_loss: 0.6630 - val_accuracy: 0.7632\n",
            "Epoch 247 / 1000 - loss: 0.5571 - accuracy: 0.8418 - val_loss: 0.6622 - val_accuracy: 0.7632\n",
            "Epoch 248 / 1000 - loss: 0.5560 - accuracy: 0.8418 - val_loss: 0.6605 - val_accuracy: 0.7632\n",
            "Epoch 249 / 1000 - loss: 0.5549 - accuracy: 0.8418 - val_loss: 0.6593 - val_accuracy: 0.7632\n",
            "Epoch 250 / 1000 - loss: 0.5539 - accuracy: 0.8418 - val_loss: 0.6580 - val_accuracy: 0.7632\n",
            "Epoch 251 / 1000 - loss: 0.5528 - accuracy: 0.8418 - val_loss: 0.6574 - val_accuracy: 0.7632\n",
            "Epoch 252 / 1000 - loss: 0.5518 - accuracy: 0.8552 - val_loss: 0.6558 - val_accuracy: 0.7632\n",
            "Epoch 253 / 1000 - loss: 0.5507 - accuracy: 0.8586 - val_loss: 0.6540 - val_accuracy: 0.7632\n",
            "Epoch 254 / 1000 - loss: 0.5497 - accuracy: 0.8586 - val_loss: 0.6526 - val_accuracy: 0.7895\n",
            "Epoch 255 / 1000 - loss: 0.5486 - accuracy: 0.8586 - val_loss: 0.6512 - val_accuracy: 0.7895\n",
            "Epoch 256 / 1000 - loss: 0.5476 - accuracy: 0.8586 - val_loss: 0.6503 - val_accuracy: 0.7895\n",
            "Epoch 257 / 1000 - loss: 0.5465 - accuracy: 0.8586 - val_loss: 0.6496 - val_accuracy: 0.7895\n",
            "Epoch 258 / 1000 - loss: 0.5455 - accuracy: 0.8586 - val_loss: 0.6483 - val_accuracy: 0.7895\n",
            "Epoch 259 / 1000 - loss: 0.5444 - accuracy: 0.8586 - val_loss: 0.6479 - val_accuracy: 0.7895\n",
            "Epoch 260 / 1000 - loss: 0.5434 - accuracy: 0.8586 - val_loss: 0.6468 - val_accuracy: 0.7895\n",
            "Epoch 261 / 1000 - loss: 0.5424 - accuracy: 0.8586 - val_loss: 0.6457 - val_accuracy: 0.7895\n",
            "Epoch 262 / 1000 - loss: 0.5414 - accuracy: 0.8586 - val_loss: 0.6448 - val_accuracy: 0.7895\n",
            "Epoch 263 / 1000 - loss: 0.5403 - accuracy: 0.8586 - val_loss: 0.6434 - val_accuracy: 0.7895\n",
            "Epoch 264 / 1000 - loss: 0.5393 - accuracy: 0.8586 - val_loss: 0.6422 - val_accuracy: 0.7895\n",
            "Epoch 265 / 1000 - loss: 0.5383 - accuracy: 0.8721 - val_loss: 0.6401 - val_accuracy: 0.7895\n",
            "Epoch 266 / 1000 - loss: 0.5373 - accuracy: 0.8721 - val_loss: 0.6392 - val_accuracy: 0.7895\n",
            "Epoch 267 / 1000 - loss: 0.5363 - accuracy: 0.8721 - val_loss: 0.6382 - val_accuracy: 0.7895\n",
            "Epoch 268 / 1000 - loss: 0.5353 - accuracy: 0.8721 - val_loss: 0.6378 - val_accuracy: 0.7895\n",
            "Epoch 269 / 1000 - loss: 0.5342 - accuracy: 0.8721 - val_loss: 0.6368 - val_accuracy: 0.7895\n",
            "Epoch 270 / 1000 - loss: 0.5332 - accuracy: 0.8721 - val_loss: 0.6355 - val_accuracy: 0.7895\n",
            "Epoch 271 / 1000 - loss: 0.5322 - accuracy: 0.8721 - val_loss: 0.6346 - val_accuracy: 0.7895\n",
            "Epoch 272 / 1000 - loss: 0.5312 - accuracy: 0.8754 - val_loss: 0.6337 - val_accuracy: 0.7895\n",
            "Epoch 273 / 1000 - loss: 0.5302 - accuracy: 0.8721 - val_loss: 0.6334 - val_accuracy: 0.7895\n",
            "Epoch 274 / 1000 - loss: 0.5293 - accuracy: 0.8721 - val_loss: 0.6328 - val_accuracy: 0.7895\n",
            "Epoch 275 / 1000 - loss: 0.5283 - accuracy: 0.8721 - val_loss: 0.6319 - val_accuracy: 0.7895\n",
            "Epoch 276 / 1000 - loss: 0.5273 - accuracy: 0.8754 - val_loss: 0.6311 - val_accuracy: 0.7895\n",
            "Epoch 277 / 1000 - loss: 0.5263 - accuracy: 0.8754 - val_loss: 0.6299 - val_accuracy: 0.7895\n",
            "Epoch 278 / 1000 - loss: 0.5253 - accuracy: 0.8754 - val_loss: 0.6286 - val_accuracy: 0.7895\n",
            "Epoch 279 / 1000 - loss: 0.5243 - accuracy: 0.8754 - val_loss: 0.6274 - val_accuracy: 0.7895\n",
            "Epoch 280 / 1000 - loss: 0.5234 - accuracy: 0.8754 - val_loss: 0.6263 - val_accuracy: 0.8158\n",
            "Epoch 281 / 1000 - loss: 0.5224 - accuracy: 0.8754 - val_loss: 0.6248 - val_accuracy: 0.8158\n",
            "Epoch 282 / 1000 - loss: 0.5214 - accuracy: 0.8788 - val_loss: 0.6229 - val_accuracy: 0.8158\n",
            "Epoch 283 / 1000 - loss: 0.5204 - accuracy: 0.8788 - val_loss: 0.6223 - val_accuracy: 0.8158\n",
            "Epoch 284 / 1000 - loss: 0.5195 - accuracy: 0.8788 - val_loss: 0.6213 - val_accuracy: 0.8158\n",
            "Epoch 285 / 1000 - loss: 0.5185 - accuracy: 0.8754 - val_loss: 0.6211 - val_accuracy: 0.8158\n",
            "Epoch 286 / 1000 - loss: 0.5175 - accuracy: 0.8754 - val_loss: 0.6210 - val_accuracy: 0.8158\n",
            "Epoch 287 / 1000 - loss: 0.5166 - accuracy: 0.8754 - val_loss: 0.6197 - val_accuracy: 0.8158\n",
            "Epoch 288 / 1000 - loss: 0.5156 - accuracy: 0.8788 - val_loss: 0.6180 - val_accuracy: 0.8158\n",
            "Epoch 289 / 1000 - loss: 0.5147 - accuracy: 0.8788 - val_loss: 0.6175 - val_accuracy: 0.8158\n",
            "Epoch 290 / 1000 - loss: 0.5137 - accuracy: 0.8788 - val_loss: 0.6168 - val_accuracy: 0.8158\n",
            "Epoch 291 / 1000 - loss: 0.5128 - accuracy: 0.8788 - val_loss: 0.6154 - val_accuracy: 0.8158\n",
            "Epoch 292 / 1000 - loss: 0.5118 - accuracy: 0.8788 - val_loss: 0.6143 - val_accuracy: 0.8158\n",
            "Epoch 293 / 1000 - loss: 0.5109 - accuracy: 0.8788 - val_loss: 0.6139 - val_accuracy: 0.8158\n",
            "Epoch 294 / 1000 - loss: 0.5099 - accuracy: 0.8788 - val_loss: 0.6129 - val_accuracy: 0.8158\n",
            "Epoch 295 / 1000 - loss: 0.5090 - accuracy: 0.8788 - val_loss: 0.6115 - val_accuracy: 0.8158\n",
            "Epoch 296 / 1000 - loss: 0.5081 - accuracy: 0.8788 - val_loss: 0.6106 - val_accuracy: 0.8158\n",
            "Epoch 297 / 1000 - loss: 0.5071 - accuracy: 0.8788 - val_loss: 0.6096 - val_accuracy: 0.8158\n",
            "Epoch 298 / 1000 - loss: 0.5062 - accuracy: 0.8788 - val_loss: 0.6084 - val_accuracy: 0.8158\n",
            "Epoch 299 / 1000 - loss: 0.5053 - accuracy: 0.8788 - val_loss: 0.6069 - val_accuracy: 0.8158\n",
            "Epoch 300 / 1000 - loss: 0.5044 - accuracy: 0.8923 - val_loss: 0.6057 - val_accuracy: 0.8158\n",
            "Epoch 301 / 1000 - loss: 0.5034 - accuracy: 0.8788 - val_loss: 0.6050 - val_accuracy: 0.8158\n",
            "Epoch 302 / 1000 - loss: 0.5025 - accuracy: 0.8788 - val_loss: 0.6039 - val_accuracy: 0.8158\n",
            "Epoch 303 / 1000 - loss: 0.5016 - accuracy: 0.8788 - val_loss: 0.6030 - val_accuracy: 0.8158\n",
            "Epoch 304 / 1000 - loss: 0.5007 - accuracy: 0.8923 - val_loss: 0.6017 - val_accuracy: 0.8158\n",
            "Epoch 305 / 1000 - loss: 0.4998 - accuracy: 0.9057 - val_loss: 0.6002 - val_accuracy: 0.8158\n",
            "Epoch 306 / 1000 - loss: 0.4989 - accuracy: 0.9057 - val_loss: 0.5996 - val_accuracy: 0.8158\n",
            "Epoch 307 / 1000 - loss: 0.4980 - accuracy: 0.9057 - val_loss: 0.5984 - val_accuracy: 0.8158\n",
            "Epoch 308 / 1000 - loss: 0.4970 - accuracy: 0.9057 - val_loss: 0.5977 - val_accuracy: 0.8158\n",
            "Epoch 309 / 1000 - loss: 0.4961 - accuracy: 0.9057 - val_loss: 0.5968 - val_accuracy: 0.8158\n",
            "Epoch 310 / 1000 - loss: 0.4952 - accuracy: 0.9057 - val_loss: 0.5958 - val_accuracy: 0.8158\n",
            "Epoch 311 / 1000 - loss: 0.4943 - accuracy: 0.9057 - val_loss: 0.5949 - val_accuracy: 0.8158\n",
            "Epoch 312 / 1000 - loss: 0.4935 - accuracy: 0.9057 - val_loss: 0.5936 - val_accuracy: 0.8158\n",
            "Epoch 313 / 1000 - loss: 0.4926 - accuracy: 0.9057 - val_loss: 0.5932 - val_accuracy: 0.8158\n",
            "Epoch 314 / 1000 - loss: 0.4917 - accuracy: 0.9057 - val_loss: 0.5928 - val_accuracy: 0.8158\n",
            "Epoch 315 / 1000 - loss: 0.4908 - accuracy: 0.9057 - val_loss: 0.5916 - val_accuracy: 0.8158\n",
            "Epoch 316 / 1000 - loss: 0.4899 - accuracy: 0.9057 - val_loss: 0.5903 - val_accuracy: 0.8158\n",
            "Epoch 317 / 1000 - loss: 0.4890 - accuracy: 0.9057 - val_loss: 0.5891 - val_accuracy: 0.8158\n",
            "Epoch 318 / 1000 - loss: 0.4881 - accuracy: 0.9057 - val_loss: 0.5890 - val_accuracy: 0.8158\n",
            "Epoch 319 / 1000 - loss: 0.4873 - accuracy: 0.9057 - val_loss: 0.5886 - val_accuracy: 0.8158\n",
            "Epoch 320 / 1000 - loss: 0.4864 - accuracy: 0.9057 - val_loss: 0.5870 - val_accuracy: 0.8158\n",
            "Epoch 321 / 1000 - loss: 0.4855 - accuracy: 0.9057 - val_loss: 0.5864 - val_accuracy: 0.8158\n",
            "Epoch 322 / 1000 - loss: 0.4846 - accuracy: 0.9057 - val_loss: 0.5853 - val_accuracy: 0.8158\n",
            "Epoch 323 / 1000 - loss: 0.4838 - accuracy: 0.9057 - val_loss: 0.5842 - val_accuracy: 0.8158\n",
            "Epoch 324 / 1000 - loss: 0.4829 - accuracy: 0.9091 - val_loss: 0.5832 - val_accuracy: 0.8158\n",
            "Epoch 325 / 1000 - loss: 0.4820 - accuracy: 0.9057 - val_loss: 0.5828 - val_accuracy: 0.8158\n",
            "Epoch 326 / 1000 - loss: 0.4812 - accuracy: 0.9057 - val_loss: 0.5823 - val_accuracy: 0.8158\n",
            "Epoch 327 / 1000 - loss: 0.4803 - accuracy: 0.9057 - val_loss: 0.5814 - val_accuracy: 0.8158\n",
            "Epoch 328 / 1000 - loss: 0.4795 - accuracy: 0.9057 - val_loss: 0.5803 - val_accuracy: 0.8158\n",
            "Epoch 329 / 1000 - loss: 0.4786 - accuracy: 0.9057 - val_loss: 0.5796 - val_accuracy: 0.8158\n",
            "Epoch 330 / 1000 - loss: 0.4778 - accuracy: 0.9057 - val_loss: 0.5782 - val_accuracy: 0.8158\n",
            "Epoch 331 / 1000 - loss: 0.4769 - accuracy: 0.9057 - val_loss: 0.5777 - val_accuracy: 0.8158\n",
            "Epoch 332 / 1000 - loss: 0.4761 - accuracy: 0.9057 - val_loss: 0.5770 - val_accuracy: 0.8158\n",
            "Epoch 333 / 1000 - loss: 0.4752 - accuracy: 0.9057 - val_loss: 0.5763 - val_accuracy: 0.8158\n",
            "Epoch 334 / 1000 - loss: 0.4744 - accuracy: 0.9057 - val_loss: 0.5755 - val_accuracy: 0.8158\n",
            "Epoch 335 / 1000 - loss: 0.4735 - accuracy: 0.9091 - val_loss: 0.5742 - val_accuracy: 0.8158\n",
            "Epoch 336 / 1000 - loss: 0.4727 - accuracy: 0.9057 - val_loss: 0.5743 - val_accuracy: 0.8158\n",
            "Epoch 337 / 1000 - loss: 0.4719 - accuracy: 0.9091 - val_loss: 0.5732 - val_accuracy: 0.8158\n",
            "Epoch 338 / 1000 - loss: 0.4710 - accuracy: 0.9091 - val_loss: 0.5719 - val_accuracy: 0.8158\n",
            "Epoch 339 / 1000 - loss: 0.4702 - accuracy: 0.9057 - val_loss: 0.5717 - val_accuracy: 0.8158\n",
            "Epoch 340 / 1000 - loss: 0.4694 - accuracy: 0.9057 - val_loss: 0.5713 - val_accuracy: 0.8158\n",
            "Epoch 341 / 1000 - loss: 0.4686 - accuracy: 0.9057 - val_loss: 0.5706 - val_accuracy: 0.8158\n",
            "Epoch 342 / 1000 - loss: 0.4677 - accuracy: 0.9057 - val_loss: 0.5695 - val_accuracy: 0.8158\n",
            "Epoch 343 / 1000 - loss: 0.4669 - accuracy: 0.9057 - val_loss: 0.5690 - val_accuracy: 0.8158\n",
            "Epoch 344 / 1000 - loss: 0.4661 - accuracy: 0.9125 - val_loss: 0.5677 - val_accuracy: 0.8158\n",
            "Epoch 345 / 1000 - loss: 0.4653 - accuracy: 0.9125 - val_loss: 0.5666 - val_accuracy: 0.8158\n",
            "Epoch 346 / 1000 - loss: 0.4645 - accuracy: 0.9125 - val_loss: 0.5657 - val_accuracy: 0.8158\n",
            "Epoch 347 / 1000 - loss: 0.4637 - accuracy: 0.9125 - val_loss: 0.5650 - val_accuracy: 0.8158\n",
            "Epoch 348 / 1000 - loss: 0.4628 - accuracy: 0.9125 - val_loss: 0.5650 - val_accuracy: 0.8158\n",
            "Epoch 349 / 1000 - loss: 0.4620 - accuracy: 0.9091 - val_loss: 0.5651 - val_accuracy: 0.8158\n",
            "Epoch 350 / 1000 - loss: 0.4612 - accuracy: 0.9125 - val_loss: 0.5639 - val_accuracy: 0.8158\n",
            "Epoch 351 / 1000 - loss: 0.4604 - accuracy: 0.9125 - val_loss: 0.5623 - val_accuracy: 0.8158\n",
            "Epoch 352 / 1000 - loss: 0.4596 - accuracy: 0.9125 - val_loss: 0.5619 - val_accuracy: 0.8158\n",
            "Epoch 353 / 1000 - loss: 0.4588 - accuracy: 0.9125 - val_loss: 0.5607 - val_accuracy: 0.8158\n",
            "Epoch 354 / 1000 - loss: 0.4580 - accuracy: 0.9125 - val_loss: 0.5596 - val_accuracy: 0.8158\n",
            "Epoch 355 / 1000 - loss: 0.4572 - accuracy: 0.9125 - val_loss: 0.5591 - val_accuracy: 0.8158\n",
            "Epoch 356 / 1000 - loss: 0.4565 - accuracy: 0.9125 - val_loss: 0.5583 - val_accuracy: 0.8158\n",
            "Epoch 357 / 1000 - loss: 0.4557 - accuracy: 0.9125 - val_loss: 0.5572 - val_accuracy: 0.8158\n",
            "Epoch 358 / 1000 - loss: 0.4549 - accuracy: 0.9125 - val_loss: 0.5568 - val_accuracy: 0.8158\n",
            "Epoch 359 / 1000 - loss: 0.4541 - accuracy: 0.9125 - val_loss: 0.5564 - val_accuracy: 0.8158\n",
            "Epoch 360 / 1000 - loss: 0.4533 - accuracy: 0.9125 - val_loss: 0.5557 - val_accuracy: 0.8158\n",
            "Epoch 361 / 1000 - loss: 0.4525 - accuracy: 0.9125 - val_loss: 0.5551 - val_accuracy: 0.8158\n",
            "Epoch 362 / 1000 - loss: 0.4518 - accuracy: 0.9125 - val_loss: 0.5539 - val_accuracy: 0.8158\n",
            "Epoch 363 / 1000 - loss: 0.4510 - accuracy: 0.9125 - val_loss: 0.5529 - val_accuracy: 0.8158\n",
            "Epoch 364 / 1000 - loss: 0.4502 - accuracy: 0.9125 - val_loss: 0.5527 - val_accuracy: 0.8158\n",
            "Epoch 365 / 1000 - loss: 0.4494 - accuracy: 0.9125 - val_loss: 0.5527 - val_accuracy: 0.8158\n",
            "Epoch 366 / 1000 - loss: 0.4487 - accuracy: 0.9125 - val_loss: 0.5515 - val_accuracy: 0.8158\n",
            "Epoch 367 / 1000 - loss: 0.4479 - accuracy: 0.9125 - val_loss: 0.5501 - val_accuracy: 0.8158\n",
            "Epoch 368 / 1000 - loss: 0.4471 - accuracy: 0.9125 - val_loss: 0.5493 - val_accuracy: 0.8158\n",
            "Epoch 369 / 1000 - loss: 0.4464 - accuracy: 0.9125 - val_loss: 0.5486 - val_accuracy: 0.8158\n",
            "Epoch 370 / 1000 - loss: 0.4456 - accuracy: 0.9125 - val_loss: 0.5488 - val_accuracy: 0.8158\n",
            "Epoch 371 / 1000 - loss: 0.4449 - accuracy: 0.9125 - val_loss: 0.5475 - val_accuracy: 0.8158\n",
            "Epoch 372 / 1000 - loss: 0.4441 - accuracy: 0.9259 - val_loss: 0.5468 - val_accuracy: 0.8158\n",
            "Epoch 373 / 1000 - loss: 0.4433 - accuracy: 0.9259 - val_loss: 0.5460 - val_accuracy: 0.8158\n",
            "Epoch 374 / 1000 - loss: 0.4426 - accuracy: 0.9259 - val_loss: 0.5459 - val_accuracy: 0.8158\n",
            "Epoch 375 / 1000 - loss: 0.4418 - accuracy: 0.9259 - val_loss: 0.5455 - val_accuracy: 0.8158\n",
            "Epoch 376 / 1000 - loss: 0.4411 - accuracy: 0.9259 - val_loss: 0.5437 - val_accuracy: 0.8158\n",
            "Epoch 377 / 1000 - loss: 0.4404 - accuracy: 0.9259 - val_loss: 0.5436 - val_accuracy: 0.8158\n",
            "Epoch 378 / 1000 - loss: 0.4396 - accuracy: 0.9259 - val_loss: 0.5434 - val_accuracy: 0.8158\n",
            "Epoch 379 / 1000 - loss: 0.4389 - accuracy: 0.9259 - val_loss: 0.5426 - val_accuracy: 0.8158\n",
            "Epoch 380 / 1000 - loss: 0.4381 - accuracy: 0.9259 - val_loss: 0.5415 - val_accuracy: 0.8158\n",
            "Epoch 381 / 1000 - loss: 0.4374 - accuracy: 0.9259 - val_loss: 0.5412 - val_accuracy: 0.8158\n",
            "Epoch 382 / 1000 - loss: 0.4367 - accuracy: 0.9259 - val_loss: 0.5406 - val_accuracy: 0.8158\n",
            "Epoch 383 / 1000 - loss: 0.4359 - accuracy: 0.9259 - val_loss: 0.5396 - val_accuracy: 0.8158\n",
            "Epoch 384 / 1000 - loss: 0.4352 - accuracy: 0.9259 - val_loss: 0.5385 - val_accuracy: 0.8158\n",
            "Epoch 385 / 1000 - loss: 0.4345 - accuracy: 0.9259 - val_loss: 0.5379 - val_accuracy: 0.8158\n",
            "Epoch 386 / 1000 - loss: 0.4337 - accuracy: 0.9293 - val_loss: 0.5375 - val_accuracy: 0.8158\n",
            "Epoch 387 / 1000 - loss: 0.4330 - accuracy: 0.9293 - val_loss: 0.5375 - val_accuracy: 0.8158\n",
            "Epoch 388 / 1000 - loss: 0.4323 - accuracy: 0.9293 - val_loss: 0.5365 - val_accuracy: 0.8158\n",
            "Epoch 389 / 1000 - loss: 0.4316 - accuracy: 0.9259 - val_loss: 0.5365 - val_accuracy: 0.8158\n",
            "Epoch 390 / 1000 - loss: 0.4309 - accuracy: 0.9259 - val_loss: 0.5353 - val_accuracy: 0.8158\n",
            "Epoch 391 / 1000 - loss: 0.4301 - accuracy: 0.9327 - val_loss: 0.5346 - val_accuracy: 0.8158\n",
            "Epoch 392 / 1000 - loss: 0.4294 - accuracy: 0.9327 - val_loss: 0.5341 - val_accuracy: 0.8158\n",
            "Epoch 393 / 1000 - loss: 0.4287 - accuracy: 0.9327 - val_loss: 0.5333 - val_accuracy: 0.8158\n",
            "Epoch 394 / 1000 - loss: 0.4280 - accuracy: 0.9360 - val_loss: 0.5321 - val_accuracy: 0.8158\n",
            "Epoch 395 / 1000 - loss: 0.4273 - accuracy: 0.9360 - val_loss: 0.5315 - val_accuracy: 0.8158\n",
            "Epoch 396 / 1000 - loss: 0.4266 - accuracy: 0.9360 - val_loss: 0.5301 - val_accuracy: 0.8158\n",
            "Epoch 397 / 1000 - loss: 0.4259 - accuracy: 0.9360 - val_loss: 0.5297 - val_accuracy: 0.8158\n",
            "Epoch 398 / 1000 - loss: 0.4252 - accuracy: 0.9360 - val_loss: 0.5290 - val_accuracy: 0.8158\n",
            "Epoch 399 / 1000 - loss: 0.4245 - accuracy: 0.9360 - val_loss: 0.5286 - val_accuracy: 0.8158\n",
            "Epoch 400 / 1000 - loss: 0.4238 - accuracy: 0.9360 - val_loss: 0.5285 - val_accuracy: 0.8158\n",
            "Epoch 401 / 1000 - loss: 0.4231 - accuracy: 0.9360 - val_loss: 0.5282 - val_accuracy: 0.8158\n",
            "Epoch 402 / 1000 - loss: 0.4224 - accuracy: 0.9360 - val_loss: 0.5275 - val_accuracy: 0.8158\n",
            "Epoch 403 / 1000 - loss: 0.4217 - accuracy: 0.9360 - val_loss: 0.5271 - val_accuracy: 0.8158\n",
            "Epoch 404 / 1000 - loss: 0.4210 - accuracy: 0.9360 - val_loss: 0.5264 - val_accuracy: 0.8158\n",
            "Epoch 405 / 1000 - loss: 0.4203 - accuracy: 0.9360 - val_loss: 0.5250 - val_accuracy: 0.8158\n",
            "Epoch 406 / 1000 - loss: 0.4196 - accuracy: 0.9327 - val_loss: 0.5255 - val_accuracy: 0.8158\n",
            "Epoch 407 / 1000 - loss: 0.4189 - accuracy: 0.9360 - val_loss: 0.5247 - val_accuracy: 0.8158\n",
            "Epoch 408 / 1000 - loss: 0.4182 - accuracy: 0.9360 - val_loss: 0.5237 - val_accuracy: 0.8158\n",
            "Epoch 409 / 1000 - loss: 0.4176 - accuracy: 0.9360 - val_loss: 0.5233 - val_accuracy: 0.8158\n",
            "Epoch 410 / 1000 - loss: 0.4169 - accuracy: 0.9360 - val_loss: 0.5225 - val_accuracy: 0.8158\n",
            "Epoch 411 / 1000 - loss: 0.4162 - accuracy: 0.9360 - val_loss: 0.5216 - val_accuracy: 0.8158\n",
            "Epoch 412 / 1000 - loss: 0.4155 - accuracy: 0.9360 - val_loss: 0.5210 - val_accuracy: 0.8158\n",
            "Epoch 413 / 1000 - loss: 0.4148 - accuracy: 0.9360 - val_loss: 0.5206 - val_accuracy: 0.8158\n",
            "Epoch 414 / 1000 - loss: 0.4142 - accuracy: 0.9327 - val_loss: 0.5209 - val_accuracy: 0.8158\n",
            "Epoch 415 / 1000 - loss: 0.4135 - accuracy: 0.9327 - val_loss: 0.5201 - val_accuracy: 0.8158\n",
            "Epoch 416 / 1000 - loss: 0.4128 - accuracy: 0.9327 - val_loss: 0.5196 - val_accuracy: 0.8158\n",
            "Epoch 417 / 1000 - loss: 0.4122 - accuracy: 0.9327 - val_loss: 0.5191 - val_accuracy: 0.8158\n",
            "Epoch 418 / 1000 - loss: 0.4115 - accuracy: 0.9360 - val_loss: 0.5180 - val_accuracy: 0.8158\n",
            "Epoch 419 / 1000 - loss: 0.4108 - accuracy: 0.9360 - val_loss: 0.5173 - val_accuracy: 0.8158\n",
            "Epoch 420 / 1000 - loss: 0.4102 - accuracy: 0.9360 - val_loss: 0.5166 - val_accuracy: 0.8158\n",
            "Epoch 421 / 1000 - loss: 0.4095 - accuracy: 0.9360 - val_loss: 0.5154 - val_accuracy: 0.8158\n",
            "Epoch 422 / 1000 - loss: 0.4089 - accuracy: 0.9360 - val_loss: 0.5148 - val_accuracy: 0.8158\n",
            "Epoch 423 / 1000 - loss: 0.4082 - accuracy: 0.9360 - val_loss: 0.5148 - val_accuracy: 0.8158\n",
            "Epoch 424 / 1000 - loss: 0.4075 - accuracy: 0.9360 - val_loss: 0.5144 - val_accuracy: 0.8158\n",
            "Epoch 425 / 1000 - loss: 0.4069 - accuracy: 0.9360 - val_loss: 0.5138 - val_accuracy: 0.8158\n",
            "Epoch 426 / 1000 - loss: 0.4062 - accuracy: 0.9360 - val_loss: 0.5130 - val_accuracy: 0.8158\n",
            "Epoch 427 / 1000 - loss: 0.4056 - accuracy: 0.9360 - val_loss: 0.5133 - val_accuracy: 0.8158\n",
            "Epoch 428 / 1000 - loss: 0.4049 - accuracy: 0.9360 - val_loss: 0.5120 - val_accuracy: 0.8158\n",
            "Epoch 429 / 1000 - loss: 0.4043 - accuracy: 0.9360 - val_loss: 0.5114 - val_accuracy: 0.8158\n",
            "Epoch 430 / 1000 - loss: 0.4037 - accuracy: 0.9360 - val_loss: 0.5109 - val_accuracy: 0.8158\n",
            "Epoch 431 / 1000 - loss: 0.4030 - accuracy: 0.9360 - val_loss: 0.5105 - val_accuracy: 0.8158\n",
            "Epoch 432 / 1000 - loss: 0.4024 - accuracy: 0.9360 - val_loss: 0.5093 - val_accuracy: 0.8158\n",
            "Epoch 433 / 1000 - loss: 0.4017 - accuracy: 0.9360 - val_loss: 0.5088 - val_accuracy: 0.8158\n",
            "Epoch 434 / 1000 - loss: 0.4011 - accuracy: 0.9360 - val_loss: 0.5083 - val_accuracy: 0.8158\n",
            "Epoch 435 / 1000 - loss: 0.4005 - accuracy: 0.9360 - val_loss: 0.5081 - val_accuracy: 0.8158\n",
            "Epoch 436 / 1000 - loss: 0.3998 - accuracy: 0.9360 - val_loss: 0.5076 - val_accuracy: 0.8158\n",
            "Epoch 437 / 1000 - loss: 0.3992 - accuracy: 0.9360 - val_loss: 0.5075 - val_accuracy: 0.8158\n",
            "Epoch 438 / 1000 - loss: 0.3986 - accuracy: 0.9360 - val_loss: 0.5074 - val_accuracy: 0.8158\n",
            "Epoch 439 / 1000 - loss: 0.3979 - accuracy: 0.9360 - val_loss: 0.5069 - val_accuracy: 0.8158\n",
            "Epoch 440 / 1000 - loss: 0.3973 - accuracy: 0.9360 - val_loss: 0.5064 - val_accuracy: 0.8158\n",
            "Epoch 441 / 1000 - loss: 0.3967 - accuracy: 0.9360 - val_loss: 0.5056 - val_accuracy: 0.8158\n",
            "Epoch 442 / 1000 - loss: 0.3961 - accuracy: 0.9394 - val_loss: 0.5042 - val_accuracy: 0.8158\n",
            "Epoch 443 / 1000 - loss: 0.3955 - accuracy: 0.9394 - val_loss: 0.5039 - val_accuracy: 0.8158\n",
            "Epoch 444 / 1000 - loss: 0.3948 - accuracy: 0.9394 - val_loss: 0.5034 - val_accuracy: 0.8158\n",
            "Epoch 445 / 1000 - loss: 0.3942 - accuracy: 0.9394 - val_loss: 0.5032 - val_accuracy: 0.8158\n",
            "Epoch 446 / 1000 - loss: 0.3936 - accuracy: 0.9394 - val_loss: 0.5026 - val_accuracy: 0.8158\n",
            "Epoch 447 / 1000 - loss: 0.3930 - accuracy: 0.9394 - val_loss: 0.5022 - val_accuracy: 0.8158\n",
            "Epoch 448 / 1000 - loss: 0.3924 - accuracy: 0.9394 - val_loss: 0.5016 - val_accuracy: 0.8158\n",
            "Epoch 449 / 1000 - loss: 0.3918 - accuracy: 0.9394 - val_loss: 0.5012 - val_accuracy: 0.8158\n",
            "Epoch 450 / 1000 - loss: 0.3912 - accuracy: 0.9394 - val_loss: 0.5006 - val_accuracy: 0.8158\n",
            "Epoch 451 / 1000 - loss: 0.3905 - accuracy: 0.9394 - val_loss: 0.4998 - val_accuracy: 0.8158\n",
            "Epoch 452 / 1000 - loss: 0.3899 - accuracy: 0.9394 - val_loss: 0.4996 - val_accuracy: 0.8158\n",
            "Epoch 453 / 1000 - loss: 0.3893 - accuracy: 0.9394 - val_loss: 0.4993 - val_accuracy: 0.8158\n",
            "Epoch 454 / 1000 - loss: 0.3887 - accuracy: 0.9394 - val_loss: 0.4987 - val_accuracy: 0.8158\n",
            "Epoch 455 / 1000 - loss: 0.3881 - accuracy: 0.9394 - val_loss: 0.4984 - val_accuracy: 0.8158\n",
            "Epoch 456 / 1000 - loss: 0.3875 - accuracy: 0.9394 - val_loss: 0.4978 - val_accuracy: 0.8158\n",
            "Epoch 457 / 1000 - loss: 0.3869 - accuracy: 0.9394 - val_loss: 0.4974 - val_accuracy: 0.8158\n",
            "Epoch 458 / 1000 - loss: 0.3863 - accuracy: 0.9394 - val_loss: 0.4970 - val_accuracy: 0.8158\n",
            "Epoch 459 / 1000 - loss: 0.3857 - accuracy: 0.9394 - val_loss: 0.4960 - val_accuracy: 0.8158\n",
            "Epoch 460 / 1000 - loss: 0.3852 - accuracy: 0.9394 - val_loss: 0.4952 - val_accuracy: 0.8158\n",
            "Epoch 461 / 1000 - loss: 0.3846 - accuracy: 0.9394 - val_loss: 0.4945 - val_accuracy: 0.8158\n",
            "Epoch 462 / 1000 - loss: 0.3840 - accuracy: 0.9394 - val_loss: 0.4946 - val_accuracy: 0.8158\n",
            "Epoch 463 / 1000 - loss: 0.3834 - accuracy: 0.9394 - val_loss: 0.4943 - val_accuracy: 0.8158\n",
            "Epoch 464 / 1000 - loss: 0.3828 - accuracy: 0.9394 - val_loss: 0.4938 - val_accuracy: 0.8158\n",
            "Epoch 465 / 1000 - loss: 0.3822 - accuracy: 0.9394 - val_loss: 0.4936 - val_accuracy: 0.8158\n",
            "Epoch 466 / 1000 - loss: 0.3816 - accuracy: 0.9394 - val_loss: 0.4932 - val_accuracy: 0.8158\n",
            "Epoch 467 / 1000 - loss: 0.3811 - accuracy: 0.9394 - val_loss: 0.4926 - val_accuracy: 0.8158\n",
            "Epoch 468 / 1000 - loss: 0.3805 - accuracy: 0.9394 - val_loss: 0.4927 - val_accuracy: 0.8158\n",
            "Epoch 469 / 1000 - loss: 0.3799 - accuracy: 0.9394 - val_loss: 0.4921 - val_accuracy: 0.8158\n",
            "Epoch 470 / 1000 - loss: 0.3793 - accuracy: 0.9394 - val_loss: 0.4916 - val_accuracy: 0.8158\n",
            "Epoch 471 / 1000 - loss: 0.3788 - accuracy: 0.9428 - val_loss: 0.4909 - val_accuracy: 0.8158\n",
            "Epoch 472 / 1000 - loss: 0.3782 - accuracy: 0.9428 - val_loss: 0.4908 - val_accuracy: 0.8158\n",
            "Epoch 473 / 1000 - loss: 0.3776 - accuracy: 0.9428 - val_loss: 0.4903 - val_accuracy: 0.8158\n",
            "Epoch 474 / 1000 - loss: 0.3770 - accuracy: 0.9428 - val_loss: 0.4898 - val_accuracy: 0.8158\n",
            "Epoch 475 / 1000 - loss: 0.3765 - accuracy: 0.9428 - val_loss: 0.4892 - val_accuracy: 0.8158\n",
            "Epoch 476 / 1000 - loss: 0.3759 - accuracy: 0.9428 - val_loss: 0.4884 - val_accuracy: 0.8158\n",
            "Epoch 477 / 1000 - loss: 0.3753 - accuracy: 0.9461 - val_loss: 0.4878 - val_accuracy: 0.8158\n",
            "Epoch 478 / 1000 - loss: 0.3748 - accuracy: 0.9461 - val_loss: 0.4874 - val_accuracy: 0.8158\n",
            "Epoch 479 / 1000 - loss: 0.3742 - accuracy: 0.9461 - val_loss: 0.4869 - val_accuracy: 0.8158\n",
            "Epoch 480 / 1000 - loss: 0.3737 - accuracy: 0.9461 - val_loss: 0.4864 - val_accuracy: 0.8158\n",
            "Epoch 481 / 1000 - loss: 0.3731 - accuracy: 0.9461 - val_loss: 0.4859 - val_accuracy: 0.8158\n",
            "Epoch 482 / 1000 - loss: 0.3725 - accuracy: 0.9461 - val_loss: 0.4855 - val_accuracy: 0.8158\n",
            "Epoch 483 / 1000 - loss: 0.3720 - accuracy: 0.9461 - val_loss: 0.4847 - val_accuracy: 0.8158\n",
            "Epoch 484 / 1000 - loss: 0.3714 - accuracy: 0.9461 - val_loss: 0.4846 - val_accuracy: 0.8158\n",
            "Epoch 485 / 1000 - loss: 0.3709 - accuracy: 0.9461 - val_loss: 0.4841 - val_accuracy: 0.8158\n",
            "Epoch 486 / 1000 - loss: 0.3703 - accuracy: 0.9461 - val_loss: 0.4839 - val_accuracy: 0.8158\n",
            "Epoch 487 / 1000 - loss: 0.3698 - accuracy: 0.9461 - val_loss: 0.4832 - val_accuracy: 0.8158\n",
            "Epoch 488 / 1000 - loss: 0.3692 - accuracy: 0.9461 - val_loss: 0.4825 - val_accuracy: 0.8158\n",
            "Epoch 489 / 1000 - loss: 0.3687 - accuracy: 0.9461 - val_loss: 0.4820 - val_accuracy: 0.8158\n",
            "Epoch 490 / 1000 - loss: 0.3681 - accuracy: 0.9461 - val_loss: 0.4823 - val_accuracy: 0.8158\n",
            "Epoch 491 / 1000 - loss: 0.3676 - accuracy: 0.9461 - val_loss: 0.4816 - val_accuracy: 0.8158\n",
            "Epoch 492 / 1000 - loss: 0.3671 - accuracy: 0.9461 - val_loss: 0.4811 - val_accuracy: 0.8158\n",
            "Epoch 493 / 1000 - loss: 0.3665 - accuracy: 0.9461 - val_loss: 0.4809 - val_accuracy: 0.8158\n",
            "Epoch 494 / 1000 - loss: 0.3660 - accuracy: 0.9495 - val_loss: 0.4803 - val_accuracy: 0.8158\n",
            "Epoch 495 / 1000 - loss: 0.3654 - accuracy: 0.9495 - val_loss: 0.4794 - val_accuracy: 0.8158\n",
            "Epoch 496 / 1000 - loss: 0.3649 - accuracy: 0.9495 - val_loss: 0.4793 - val_accuracy: 0.8158\n",
            "Epoch 497 / 1000 - loss: 0.3644 - accuracy: 0.9495 - val_loss: 0.4794 - val_accuracy: 0.8158\n",
            "Epoch 498 / 1000 - loss: 0.3638 - accuracy: 0.9495 - val_loss: 0.4789 - val_accuracy: 0.8158\n",
            "Epoch 499 / 1000 - loss: 0.3633 - accuracy: 0.9495 - val_loss: 0.4782 - val_accuracy: 0.8158\n",
            "Epoch 500 / 1000 - loss: 0.3628 - accuracy: 0.9495 - val_loss: 0.4777 - val_accuracy: 0.8158\n",
            "Epoch 501 / 1000 - loss: 0.3622 - accuracy: 0.9495 - val_loss: 0.4779 - val_accuracy: 0.8158\n",
            "Epoch 502 / 1000 - loss: 0.3617 - accuracy: 0.9495 - val_loss: 0.4771 - val_accuracy: 0.8158\n",
            "Epoch 503 / 1000 - loss: 0.3612 - accuracy: 0.9495 - val_loss: 0.4769 - val_accuracy: 0.8158\n",
            "Epoch 504 / 1000 - loss: 0.3607 - accuracy: 0.9495 - val_loss: 0.4762 - val_accuracy: 0.8158\n",
            "Epoch 505 / 1000 - loss: 0.3601 - accuracy: 0.9495 - val_loss: 0.4765 - val_accuracy: 0.8158\n",
            "Epoch 506 / 1000 - loss: 0.3596 - accuracy: 0.9495 - val_loss: 0.4759 - val_accuracy: 0.8158\n",
            "Epoch 507 / 1000 - loss: 0.3591 - accuracy: 0.9495 - val_loss: 0.4753 - val_accuracy: 0.8158\n",
            "Epoch 508 / 1000 - loss: 0.3586 - accuracy: 0.9495 - val_loss: 0.4751 - val_accuracy: 0.8158\n",
            "Epoch 509 / 1000 - loss: 0.3581 - accuracy: 0.9495 - val_loss: 0.4743 - val_accuracy: 0.8158\n",
            "Epoch 510 / 1000 - loss: 0.3575 - accuracy: 0.9495 - val_loss: 0.4738 - val_accuracy: 0.8158\n",
            "Epoch 511 / 1000 - loss: 0.3570 - accuracy: 0.9495 - val_loss: 0.4736 - val_accuracy: 0.8158\n",
            "Epoch 512 / 1000 - loss: 0.3565 - accuracy: 0.9495 - val_loss: 0.4732 - val_accuracy: 0.8158\n",
            "Epoch 513 / 1000 - loss: 0.3560 - accuracy: 0.9495 - val_loss: 0.4721 - val_accuracy: 0.8158\n",
            "Epoch 514 / 1000 - loss: 0.3555 - accuracy: 0.9495 - val_loss: 0.4719 - val_accuracy: 0.8158\n",
            "Epoch 515 / 1000 - loss: 0.3550 - accuracy: 0.9495 - val_loss: 0.4715 - val_accuracy: 0.8158\n",
            "Epoch 516 / 1000 - loss: 0.3545 - accuracy: 0.9495 - val_loss: 0.4720 - val_accuracy: 0.8158\n",
            "Epoch 517 / 1000 - loss: 0.3540 - accuracy: 0.9495 - val_loss: 0.4717 - val_accuracy: 0.8158\n",
            "Epoch 518 / 1000 - loss: 0.3535 - accuracy: 0.9495 - val_loss: 0.4710 - val_accuracy: 0.8158\n",
            "Epoch 519 / 1000 - loss: 0.3530 - accuracy: 0.9495 - val_loss: 0.4702 - val_accuracy: 0.8158\n",
            "Epoch 520 / 1000 - loss: 0.3525 - accuracy: 0.9495 - val_loss: 0.4699 - val_accuracy: 0.8158\n",
            "Epoch 521 / 1000 - loss: 0.3520 - accuracy: 0.9495 - val_loss: 0.4696 - val_accuracy: 0.8158\n",
            "Epoch 522 / 1000 - loss: 0.3515 - accuracy: 0.9495 - val_loss: 0.4691 - val_accuracy: 0.8158\n",
            "Epoch 523 / 1000 - loss: 0.3510 - accuracy: 0.9495 - val_loss: 0.4687 - val_accuracy: 0.8158\n",
            "Epoch 524 / 1000 - loss: 0.3505 - accuracy: 0.9495 - val_loss: 0.4690 - val_accuracy: 0.8158\n",
            "Epoch 525 / 1000 - loss: 0.3500 - accuracy: 0.9495 - val_loss: 0.4681 - val_accuracy: 0.8158\n",
            "Epoch 526 / 1000 - loss: 0.3495 - accuracy: 0.9495 - val_loss: 0.4684 - val_accuracy: 0.8158\n",
            "Epoch 527 / 1000 - loss: 0.3490 - accuracy: 0.9495 - val_loss: 0.4674 - val_accuracy: 0.8158\n",
            "Epoch 528 / 1000 - loss: 0.3485 - accuracy: 0.9495 - val_loss: 0.4669 - val_accuracy: 0.8158\n",
            "Epoch 529 / 1000 - loss: 0.3480 - accuracy: 0.9495 - val_loss: 0.4661 - val_accuracy: 0.8158\n",
            "Epoch 530 / 1000 - loss: 0.3475 - accuracy: 0.9495 - val_loss: 0.4659 - val_accuracy: 0.8158\n",
            "Epoch 531 / 1000 - loss: 0.3470 - accuracy: 0.9495 - val_loss: 0.4655 - val_accuracy: 0.8158\n",
            "Epoch 532 / 1000 - loss: 0.3465 - accuracy: 0.9495 - val_loss: 0.4653 - val_accuracy: 0.8158\n",
            "Epoch 533 / 1000 - loss: 0.3461 - accuracy: 0.9495 - val_loss: 0.4647 - val_accuracy: 0.8158\n",
            "Epoch 534 / 1000 - loss: 0.3456 - accuracy: 0.9495 - val_loss: 0.4644 - val_accuracy: 0.8158\n",
            "Epoch 535 / 1000 - loss: 0.3451 - accuracy: 0.9495 - val_loss: 0.4642 - val_accuracy: 0.8158\n",
            "Epoch 536 / 1000 - loss: 0.3446 - accuracy: 0.9495 - val_loss: 0.4635 - val_accuracy: 0.8158\n",
            "Epoch 537 / 1000 - loss: 0.3441 - accuracy: 0.9495 - val_loss: 0.4633 - val_accuracy: 0.8158\n",
            "Epoch 538 / 1000 - loss: 0.3437 - accuracy: 0.9495 - val_loss: 0.4631 - val_accuracy: 0.8158\n",
            "Epoch 539 / 1000 - loss: 0.3432 - accuracy: 0.9495 - val_loss: 0.4627 - val_accuracy: 0.8158\n",
            "Epoch 540 / 1000 - loss: 0.3427 - accuracy: 0.9495 - val_loss: 0.4620 - val_accuracy: 0.8158\n",
            "Epoch 541 / 1000 - loss: 0.3422 - accuracy: 0.9495 - val_loss: 0.4617 - val_accuracy: 0.8158\n",
            "Epoch 542 / 1000 - loss: 0.3418 - accuracy: 0.9495 - val_loss: 0.4613 - val_accuracy: 0.8158\n",
            "Epoch 543 / 1000 - loss: 0.3413 - accuracy: 0.9495 - val_loss: 0.4607 - val_accuracy: 0.8158\n",
            "Epoch 544 / 1000 - loss: 0.3408 - accuracy: 0.9495 - val_loss: 0.4604 - val_accuracy: 0.8158\n",
            "Epoch 545 / 1000 - loss: 0.3403 - accuracy: 0.9495 - val_loss: 0.4600 - val_accuracy: 0.8158\n",
            "Epoch 546 / 1000 - loss: 0.3399 - accuracy: 0.9495 - val_loss: 0.4597 - val_accuracy: 0.8158\n",
            "Epoch 547 / 1000 - loss: 0.3394 - accuracy: 0.9495 - val_loss: 0.4597 - val_accuracy: 0.8158\n",
            "Epoch 548 / 1000 - loss: 0.3389 - accuracy: 0.9495 - val_loss: 0.4597 - val_accuracy: 0.8158\n",
            "Epoch 549 / 1000 - loss: 0.3385 - accuracy: 0.9495 - val_loss: 0.4598 - val_accuracy: 0.8158\n",
            "Epoch 550 / 1000 - loss: 0.3380 - accuracy: 0.9495 - val_loss: 0.4595 - val_accuracy: 0.8158\n",
            "Epoch 551 / 1000 - loss: 0.3376 - accuracy: 0.9495 - val_loss: 0.4592 - val_accuracy: 0.8158\n",
            "Epoch 552 / 1000 - loss: 0.3371 - accuracy: 0.9495 - val_loss: 0.4586 - val_accuracy: 0.8158\n",
            "Epoch 553 / 1000 - loss: 0.3366 - accuracy: 0.9495 - val_loss: 0.4578 - val_accuracy: 0.8158\n",
            "Epoch 554 / 1000 - loss: 0.3362 - accuracy: 0.9495 - val_loss: 0.4575 - val_accuracy: 0.8158\n",
            "Epoch 555 / 1000 - loss: 0.3357 - accuracy: 0.9495 - val_loss: 0.4574 - val_accuracy: 0.8158\n",
            "Epoch 556 / 1000 - loss: 0.3353 - accuracy: 0.9495 - val_loss: 0.4567 - val_accuracy: 0.8158\n",
            "Epoch 557 / 1000 - loss: 0.3348 - accuracy: 0.9495 - val_loss: 0.4562 - val_accuracy: 0.8158\n",
            "Epoch 558 / 1000 - loss: 0.3344 - accuracy: 0.9495 - val_loss: 0.4555 - val_accuracy: 0.8158\n",
            "Epoch 559 / 1000 - loss: 0.3339 - accuracy: 0.9495 - val_loss: 0.4549 - val_accuracy: 0.8158\n",
            "Epoch 560 / 1000 - loss: 0.3335 - accuracy: 0.9495 - val_loss: 0.4540 - val_accuracy: 0.8158\n",
            "Epoch 561 / 1000 - loss: 0.3330 - accuracy: 0.9495 - val_loss: 0.4541 - val_accuracy: 0.8158\n",
            "Epoch 562 / 1000 - loss: 0.3326 - accuracy: 0.9495 - val_loss: 0.4537 - val_accuracy: 0.8158\n",
            "Epoch 563 / 1000 - loss: 0.3321 - accuracy: 0.9495 - val_loss: 0.4535 - val_accuracy: 0.8158\n",
            "Epoch 564 / 1000 - loss: 0.3317 - accuracy: 0.9495 - val_loss: 0.4535 - val_accuracy: 0.8158\n",
            "Epoch 565 / 1000 - loss: 0.3312 - accuracy: 0.9495 - val_loss: 0.4531 - val_accuracy: 0.8158\n",
            "Epoch 566 / 1000 - loss: 0.3308 - accuracy: 0.9495 - val_loss: 0.4527 - val_accuracy: 0.8158\n",
            "Epoch 567 / 1000 - loss: 0.3303 - accuracy: 0.9495 - val_loss: 0.4525 - val_accuracy: 0.8158\n",
            "Epoch 568 / 1000 - loss: 0.3299 - accuracy: 0.9495 - val_loss: 0.4523 - val_accuracy: 0.8158\n",
            "Epoch 569 / 1000 - loss: 0.3294 - accuracy: 0.9495 - val_loss: 0.4523 - val_accuracy: 0.8158\n",
            "Epoch 570 / 1000 - loss: 0.3290 - accuracy: 0.9495 - val_loss: 0.4519 - val_accuracy: 0.8158\n",
            "Epoch 571 / 1000 - loss: 0.3286 - accuracy: 0.9495 - val_loss: 0.4522 - val_accuracy: 0.8158\n",
            "Epoch 572 / 1000 - loss: 0.3281 - accuracy: 0.9495 - val_loss: 0.4517 - val_accuracy: 0.8158\n",
            "Epoch 573 / 1000 - loss: 0.3277 - accuracy: 0.9529 - val_loss: 0.4509 - val_accuracy: 0.8158\n",
            "Epoch 574 / 1000 - loss: 0.3273 - accuracy: 0.9529 - val_loss: 0.4505 - val_accuracy: 0.8158\n",
            "Epoch 575 / 1000 - loss: 0.3268 - accuracy: 0.9529 - val_loss: 0.4505 - val_accuracy: 0.8158\n",
            "Epoch 576 / 1000 - loss: 0.3264 - accuracy: 0.9529 - val_loss: 0.4501 - val_accuracy: 0.8158\n",
            "Epoch 577 / 1000 - loss: 0.3260 - accuracy: 0.9529 - val_loss: 0.4505 - val_accuracy: 0.8158\n",
            "Epoch 578 / 1000 - loss: 0.3255 - accuracy: 0.9529 - val_loss: 0.4500 - val_accuracy: 0.8158\n",
            "Epoch 579 / 1000 - loss: 0.3251 - accuracy: 0.9529 - val_loss: 0.4497 - val_accuracy: 0.8158\n",
            "Epoch 580 / 1000 - loss: 0.3247 - accuracy: 0.9529 - val_loss: 0.4493 - val_accuracy: 0.8158\n",
            "Epoch 581 / 1000 - loss: 0.3243 - accuracy: 0.9529 - val_loss: 0.4487 - val_accuracy: 0.8158\n",
            "Epoch 582 / 1000 - loss: 0.3238 - accuracy: 0.9529 - val_loss: 0.4482 - val_accuracy: 0.8158\n",
            "Epoch 583 / 1000 - loss: 0.3234 - accuracy: 0.9529 - val_loss: 0.4478 - val_accuracy: 0.8158\n",
            "Epoch 584 / 1000 - loss: 0.3230 - accuracy: 0.9529 - val_loss: 0.4477 - val_accuracy: 0.8158\n",
            "Epoch 585 / 1000 - loss: 0.3226 - accuracy: 0.9529 - val_loss: 0.4470 - val_accuracy: 0.8158\n",
            "Epoch 586 / 1000 - loss: 0.3221 - accuracy: 0.9529 - val_loss: 0.4468 - val_accuracy: 0.8158\n",
            "Epoch 587 / 1000 - loss: 0.3217 - accuracy: 0.9529 - val_loss: 0.4464 - val_accuracy: 0.8158\n",
            "Epoch 588 / 1000 - loss: 0.3213 - accuracy: 0.9529 - val_loss: 0.4461 - val_accuracy: 0.8158\n",
            "Epoch 589 / 1000 - loss: 0.3209 - accuracy: 0.9529 - val_loss: 0.4459 - val_accuracy: 0.8158\n",
            "Epoch 590 / 1000 - loss: 0.3205 - accuracy: 0.9529 - val_loss: 0.4454 - val_accuracy: 0.8158\n",
            "Epoch 591 / 1000 - loss: 0.3201 - accuracy: 0.9529 - val_loss: 0.4450 - val_accuracy: 0.8158\n",
            "Epoch 592 / 1000 - loss: 0.3196 - accuracy: 0.9529 - val_loss: 0.4449 - val_accuracy: 0.8158\n",
            "Epoch 593 / 1000 - loss: 0.3192 - accuracy: 0.9529 - val_loss: 0.4446 - val_accuracy: 0.8158\n",
            "Epoch 594 / 1000 - loss: 0.3188 - accuracy: 0.9529 - val_loss: 0.4444 - val_accuracy: 0.8158\n",
            "Epoch 595 / 1000 - loss: 0.3184 - accuracy: 0.9529 - val_loss: 0.4440 - val_accuracy: 0.8158\n",
            "Epoch 596 / 1000 - loss: 0.3180 - accuracy: 0.9529 - val_loss: 0.4439 - val_accuracy: 0.8158\n",
            "Epoch 597 / 1000 - loss: 0.3176 - accuracy: 0.9529 - val_loss: 0.4438 - val_accuracy: 0.8158\n",
            "Epoch 598 / 1000 - loss: 0.3172 - accuracy: 0.9529 - val_loss: 0.4432 - val_accuracy: 0.8158\n",
            "Epoch 599 / 1000 - loss: 0.3168 - accuracy: 0.9529 - val_loss: 0.4428 - val_accuracy: 0.8158\n",
            "Epoch 600 / 1000 - loss: 0.3164 - accuracy: 0.9529 - val_loss: 0.4424 - val_accuracy: 0.8158\n",
            "Epoch 601 / 1000 - loss: 0.3160 - accuracy: 0.9529 - val_loss: 0.4420 - val_accuracy: 0.8158\n",
            "Epoch 602 / 1000 - loss: 0.3156 - accuracy: 0.9529 - val_loss: 0.4415 - val_accuracy: 0.8158\n",
            "Epoch 603 / 1000 - loss: 0.3151 - accuracy: 0.9529 - val_loss: 0.4414 - val_accuracy: 0.8158\n",
            "Epoch 604 / 1000 - loss: 0.3147 - accuracy: 0.9529 - val_loss: 0.4412 - val_accuracy: 0.8158\n",
            "Epoch 605 / 1000 - loss: 0.3143 - accuracy: 0.9529 - val_loss: 0.4413 - val_accuracy: 0.8158\n",
            "Epoch 606 / 1000 - loss: 0.3139 - accuracy: 0.9529 - val_loss: 0.4409 - val_accuracy: 0.8158\n",
            "Epoch 607 / 1000 - loss: 0.3135 - accuracy: 0.9529 - val_loss: 0.4408 - val_accuracy: 0.8158\n",
            "Epoch 608 / 1000 - loss: 0.3132 - accuracy: 0.9529 - val_loss: 0.4402 - val_accuracy: 0.8158\n",
            "Epoch 609 / 1000 - loss: 0.3128 - accuracy: 0.9529 - val_loss: 0.4400 - val_accuracy: 0.8158\n",
            "Epoch 610 / 1000 - loss: 0.3124 - accuracy: 0.9529 - val_loss: 0.4396 - val_accuracy: 0.8158\n",
            "Epoch 611 / 1000 - loss: 0.3120 - accuracy: 0.9529 - val_loss: 0.4392 - val_accuracy: 0.8158\n",
            "Epoch 612 / 1000 - loss: 0.3116 - accuracy: 0.9529 - val_loss: 0.4393 - val_accuracy: 0.8158\n",
            "Epoch 613 / 1000 - loss: 0.3112 - accuracy: 0.9529 - val_loss: 0.4388 - val_accuracy: 0.8158\n",
            "Epoch 614 / 1000 - loss: 0.3108 - accuracy: 0.9529 - val_loss: 0.4386 - val_accuracy: 0.8158\n",
            "Epoch 615 / 1000 - loss: 0.3104 - accuracy: 0.9529 - val_loss: 0.4385 - val_accuracy: 0.8158\n",
            "Epoch 616 / 1000 - loss: 0.3100 - accuracy: 0.9529 - val_loss: 0.4380 - val_accuracy: 0.8158\n",
            "Epoch 617 / 1000 - loss: 0.3096 - accuracy: 0.9529 - val_loss: 0.4378 - val_accuracy: 0.8158\n",
            "Epoch 618 / 1000 - loss: 0.3092 - accuracy: 0.9529 - val_loss: 0.4378 - val_accuracy: 0.8158\n",
            "Epoch 619 / 1000 - loss: 0.3088 - accuracy: 0.9529 - val_loss: 0.4375 - val_accuracy: 0.8158\n",
            "Epoch 620 / 1000 - loss: 0.3085 - accuracy: 0.9529 - val_loss: 0.4372 - val_accuracy: 0.8158\n",
            "Epoch 621 / 1000 - loss: 0.3081 - accuracy: 0.9529 - val_loss: 0.4369 - val_accuracy: 0.8158\n",
            "Epoch 622 / 1000 - loss: 0.3077 - accuracy: 0.9562 - val_loss: 0.4362 - val_accuracy: 0.8158\n",
            "Epoch 623 / 1000 - loss: 0.3073 - accuracy: 0.9562 - val_loss: 0.4361 - val_accuracy: 0.8158\n",
            "Epoch 624 / 1000 - loss: 0.3069 - accuracy: 0.9562 - val_loss: 0.4357 - val_accuracy: 0.8158\n",
            "Epoch 625 / 1000 - loss: 0.3065 - accuracy: 0.9562 - val_loss: 0.4355 - val_accuracy: 0.8158\n",
            "Epoch 626 / 1000 - loss: 0.3062 - accuracy: 0.9562 - val_loss: 0.4351 - val_accuracy: 0.8158\n",
            "Epoch 627 / 1000 - loss: 0.3058 - accuracy: 0.9562 - val_loss: 0.4348 - val_accuracy: 0.8158\n",
            "Epoch 628 / 1000 - loss: 0.3054 - accuracy: 0.9562 - val_loss: 0.4346 - val_accuracy: 0.8158\n",
            "Epoch 629 / 1000 - loss: 0.3050 - accuracy: 0.9562 - val_loss: 0.4345 - val_accuracy: 0.8158\n",
            "Epoch 630 / 1000 - loss: 0.3047 - accuracy: 0.9562 - val_loss: 0.4339 - val_accuracy: 0.8158\n",
            "Epoch 631 / 1000 - loss: 0.3043 - accuracy: 0.9562 - val_loss: 0.4341 - val_accuracy: 0.8158\n",
            "Epoch 632 / 1000 - loss: 0.3039 - accuracy: 0.9562 - val_loss: 0.4338 - val_accuracy: 0.8158\n",
            "Epoch 633 / 1000 - loss: 0.3035 - accuracy: 0.9562 - val_loss: 0.4337 - val_accuracy: 0.8158\n",
            "Epoch 634 / 1000 - loss: 0.3032 - accuracy: 0.9562 - val_loss: 0.4336 - val_accuracy: 0.8158\n",
            "Epoch 635 / 1000 - loss: 0.3028 - accuracy: 0.9562 - val_loss: 0.4330 - val_accuracy: 0.8158\n",
            "Epoch 636 / 1000 - loss: 0.3024 - accuracy: 0.9562 - val_loss: 0.4327 - val_accuracy: 0.8158\n",
            "Epoch 637 / 1000 - loss: 0.3021 - accuracy: 0.9562 - val_loss: 0.4326 - val_accuracy: 0.8158\n",
            "Epoch 638 / 1000 - loss: 0.3017 - accuracy: 0.9562 - val_loss: 0.4324 - val_accuracy: 0.8158\n",
            "Epoch 639 / 1000 - loss: 0.3013 - accuracy: 0.9562 - val_loss: 0.4320 - val_accuracy: 0.8158\n",
            "Epoch 640 / 1000 - loss: 0.3010 - accuracy: 0.9562 - val_loss: 0.4316 - val_accuracy: 0.8158\n",
            "Epoch 641 / 1000 - loss: 0.3006 - accuracy: 0.9562 - val_loss: 0.4314 - val_accuracy: 0.8158\n",
            "Epoch 642 / 1000 - loss: 0.3002 - accuracy: 0.9562 - val_loss: 0.4314 - val_accuracy: 0.8158\n",
            "Epoch 643 / 1000 - loss: 0.2999 - accuracy: 0.9562 - val_loss: 0.4312 - val_accuracy: 0.8158\n",
            "Epoch 644 / 1000 - loss: 0.2995 - accuracy: 0.9562 - val_loss: 0.4312 - val_accuracy: 0.8158\n",
            "Epoch 645 / 1000 - loss: 0.2991 - accuracy: 0.9562 - val_loss: 0.4310 - val_accuracy: 0.8158\n",
            "Epoch 646 / 1000 - loss: 0.2988 - accuracy: 0.9562 - val_loss: 0.4309 - val_accuracy: 0.8158\n",
            "Epoch 647 / 1000 - loss: 0.2984 - accuracy: 0.9562 - val_loss: 0.4304 - val_accuracy: 0.8158\n",
            "Epoch 648 / 1000 - loss: 0.2981 - accuracy: 0.9562 - val_loss: 0.4305 - val_accuracy: 0.8158\n",
            "Epoch 649 / 1000 - loss: 0.2977 - accuracy: 0.9562 - val_loss: 0.4302 - val_accuracy: 0.8158\n",
            "Epoch 650 / 1000 - loss: 0.2974 - accuracy: 0.9562 - val_loss: 0.4298 - val_accuracy: 0.8158\n",
            "Epoch 651 / 1000 - loss: 0.2970 - accuracy: 0.9562 - val_loss: 0.4292 - val_accuracy: 0.8158\n",
            "Epoch 652 / 1000 - loss: 0.2966 - accuracy: 0.9562 - val_loss: 0.4290 - val_accuracy: 0.8158\n",
            "Epoch 653 / 1000 - loss: 0.2963 - accuracy: 0.9562 - val_loss: 0.4293 - val_accuracy: 0.8158\n",
            "Epoch 654 / 1000 - loss: 0.2959 - accuracy: 0.9562 - val_loss: 0.4289 - val_accuracy: 0.8158\n",
            "Epoch 655 / 1000 - loss: 0.2956 - accuracy: 0.9562 - val_loss: 0.4287 - val_accuracy: 0.8158\n",
            "Epoch 656 / 1000 - loss: 0.2952 - accuracy: 0.9562 - val_loss: 0.4281 - val_accuracy: 0.8158\n",
            "Epoch 657 / 1000 - loss: 0.2949 - accuracy: 0.9562 - val_loss: 0.4277 - val_accuracy: 0.8158\n",
            "Epoch 658 / 1000 - loss: 0.2945 - accuracy: 0.9562 - val_loss: 0.4269 - val_accuracy: 0.8158\n",
            "Epoch 659 / 1000 - loss: 0.2942 - accuracy: 0.9562 - val_loss: 0.4268 - val_accuracy: 0.8158\n",
            "Epoch 660 / 1000 - loss: 0.2938 - accuracy: 0.9596 - val_loss: 0.4263 - val_accuracy: 0.8158\n",
            "Epoch 661 / 1000 - loss: 0.2935 - accuracy: 0.9596 - val_loss: 0.4263 - val_accuracy: 0.8158\n",
            "Epoch 662 / 1000 - loss: 0.2931 - accuracy: 0.9596 - val_loss: 0.4265 - val_accuracy: 0.8158\n",
            "Epoch 663 / 1000 - loss: 0.2928 - accuracy: 0.9596 - val_loss: 0.4260 - val_accuracy: 0.8158\n",
            "Epoch 664 / 1000 - loss: 0.2924 - accuracy: 0.9596 - val_loss: 0.4258 - val_accuracy: 0.8158\n",
            "Epoch 665 / 1000 - loss: 0.2921 - accuracy: 0.9596 - val_loss: 0.4254 - val_accuracy: 0.8158\n",
            "Epoch 666 / 1000 - loss: 0.2918 - accuracy: 0.9596 - val_loss: 0.4252 - val_accuracy: 0.8158\n",
            "Epoch 667 / 1000 - loss: 0.2914 - accuracy: 0.9596 - val_loss: 0.4247 - val_accuracy: 0.8158\n",
            "Epoch 668 / 1000 - loss: 0.2911 - accuracy: 0.9596 - val_loss: 0.4245 - val_accuracy: 0.8158\n",
            "Epoch 669 / 1000 - loss: 0.2907 - accuracy: 0.9596 - val_loss: 0.4244 - val_accuracy: 0.8158\n",
            "Epoch 670 / 1000 - loss: 0.2904 - accuracy: 0.9596 - val_loss: 0.4242 - val_accuracy: 0.8158\n",
            "Epoch 671 / 1000 - loss: 0.2901 - accuracy: 0.9596 - val_loss: 0.4237 - val_accuracy: 0.8158\n",
            "Epoch 672 / 1000 - loss: 0.2897 - accuracy: 0.9596 - val_loss: 0.4234 - val_accuracy: 0.8158\n",
            "Epoch 673 / 1000 - loss: 0.2894 - accuracy: 0.9596 - val_loss: 0.4235 - val_accuracy: 0.8158\n",
            "Epoch 674 / 1000 - loss: 0.2890 - accuracy: 0.9596 - val_loss: 0.4235 - val_accuracy: 0.8158\n",
            "Epoch 675 / 1000 - loss: 0.2887 - accuracy: 0.9596 - val_loss: 0.4232 - val_accuracy: 0.8158\n",
            "Epoch 676 / 1000 - loss: 0.2884 - accuracy: 0.9596 - val_loss: 0.4230 - val_accuracy: 0.8158\n",
            "Epoch 677 / 1000 - loss: 0.2880 - accuracy: 0.9596 - val_loss: 0.4230 - val_accuracy: 0.8158\n",
            "Epoch 678 / 1000 - loss: 0.2877 - accuracy: 0.9596 - val_loss: 0.4227 - val_accuracy: 0.8158\n",
            "Epoch 679 / 1000 - loss: 0.2874 - accuracy: 0.9596 - val_loss: 0.4219 - val_accuracy: 0.8158\n",
            "Epoch 680 / 1000 - loss: 0.2870 - accuracy: 0.9596 - val_loss: 0.4220 - val_accuracy: 0.8158\n",
            "Epoch 681 / 1000 - loss: 0.2867 - accuracy: 0.9596 - val_loss: 0.4216 - val_accuracy: 0.8158\n",
            "Epoch 682 / 1000 - loss: 0.2864 - accuracy: 0.9596 - val_loss: 0.4215 - val_accuracy: 0.8158\n",
            "Epoch 683 / 1000 - loss: 0.2860 - accuracy: 0.9596 - val_loss: 0.4215 - val_accuracy: 0.8158\n",
            "Epoch 684 / 1000 - loss: 0.2857 - accuracy: 0.9596 - val_loss: 0.4220 - val_accuracy: 0.8158\n",
            "Epoch 685 / 1000 - loss: 0.2854 - accuracy: 0.9596 - val_loss: 0.4214 - val_accuracy: 0.8158\n",
            "Epoch 686 / 1000 - loss: 0.2851 - accuracy: 0.9596 - val_loss: 0.4210 - val_accuracy: 0.8158\n",
            "Epoch 687 / 1000 - loss: 0.2847 - accuracy: 0.9596 - val_loss: 0.4208 - val_accuracy: 0.8158\n",
            "Epoch 688 / 1000 - loss: 0.2844 - accuracy: 0.9596 - val_loss: 0.4207 - val_accuracy: 0.8158\n",
            "Epoch 689 / 1000 - loss: 0.2841 - accuracy: 0.9596 - val_loss: 0.4203 - val_accuracy: 0.8158\n",
            "Epoch 690 / 1000 - loss: 0.2838 - accuracy: 0.9596 - val_loss: 0.4203 - val_accuracy: 0.8158\n",
            "Epoch 691 / 1000 - loss: 0.2834 - accuracy: 0.9596 - val_loss: 0.4209 - val_accuracy: 0.8158\n",
            "Epoch 692 / 1000 - loss: 0.2831 - accuracy: 0.9596 - val_loss: 0.4207 - val_accuracy: 0.8158\n",
            "Epoch 693 / 1000 - loss: 0.2828 - accuracy: 0.9596 - val_loss: 0.4204 - val_accuracy: 0.8158\n",
            "Epoch 694 / 1000 - loss: 0.2825 - accuracy: 0.9596 - val_loss: 0.4197 - val_accuracy: 0.8158\n",
            "Epoch 695 / 1000 - loss: 0.2822 - accuracy: 0.9596 - val_loss: 0.4195 - val_accuracy: 0.8158\n",
            "Epoch 696 / 1000 - loss: 0.2818 - accuracy: 0.9596 - val_loss: 0.4192 - val_accuracy: 0.8158\n",
            "Epoch 697 / 1000 - loss: 0.2815 - accuracy: 0.9596 - val_loss: 0.4189 - val_accuracy: 0.8158\n",
            "Epoch 698 / 1000 - loss: 0.2812 - accuracy: 0.9596 - val_loss: 0.4186 - val_accuracy: 0.8158\n",
            "Epoch 699 / 1000 - loss: 0.2809 - accuracy: 0.9596 - val_loss: 0.4184 - val_accuracy: 0.8158\n",
            "Epoch 700 / 1000 - loss: 0.2806 - accuracy: 0.9596 - val_loss: 0.4178 - val_accuracy: 0.8158\n",
            "Epoch 701 / 1000 - loss: 0.2802 - accuracy: 0.9596 - val_loss: 0.4174 - val_accuracy: 0.8158\n",
            "Epoch 702 / 1000 - loss: 0.2799 - accuracy: 0.9596 - val_loss: 0.4169 - val_accuracy: 0.8158\n",
            "Epoch 703 / 1000 - loss: 0.2796 - accuracy: 0.9596 - val_loss: 0.4169 - val_accuracy: 0.8158\n",
            "Epoch 704 / 1000 - loss: 0.2793 - accuracy: 0.9596 - val_loss: 0.4167 - val_accuracy: 0.8158\n",
            "Epoch 705 / 1000 - loss: 0.2790 - accuracy: 0.9596 - val_loss: 0.4162 - val_accuracy: 0.8158\n",
            "Epoch 706 / 1000 - loss: 0.2787 - accuracy: 0.9596 - val_loss: 0.4162 - val_accuracy: 0.8158\n",
            "Epoch 707 / 1000 - loss: 0.2784 - accuracy: 0.9596 - val_loss: 0.4160 - val_accuracy: 0.8158\n",
            "Epoch 708 / 1000 - loss: 0.2781 - accuracy: 0.9596 - val_loss: 0.4159 - val_accuracy: 0.8158\n",
            "Epoch 709 / 1000 - loss: 0.2777 - accuracy: 0.9596 - val_loss: 0.4156 - val_accuracy: 0.8158\n",
            "Epoch 710 / 1000 - loss: 0.2774 - accuracy: 0.9596 - val_loss: 0.4149 - val_accuracy: 0.8158\n",
            "Epoch 711 / 1000 - loss: 0.2771 - accuracy: 0.9596 - val_loss: 0.4146 - val_accuracy: 0.8158\n",
            "Epoch 712 / 1000 - loss: 0.2768 - accuracy: 0.9596 - val_loss: 0.4143 - val_accuracy: 0.8158\n",
            "Epoch 713 / 1000 - loss: 0.2765 - accuracy: 0.9596 - val_loss: 0.4143 - val_accuracy: 0.8158\n",
            "Epoch 714 / 1000 - loss: 0.2762 - accuracy: 0.9596 - val_loss: 0.4144 - val_accuracy: 0.8158\n",
            "Epoch 715 / 1000 - loss: 0.2759 - accuracy: 0.9596 - val_loss: 0.4141 - val_accuracy: 0.8158\n",
            "Epoch 716 / 1000 - loss: 0.2756 - accuracy: 0.9596 - val_loss: 0.4139 - val_accuracy: 0.8158\n",
            "Epoch 717 / 1000 - loss: 0.2753 - accuracy: 0.9596 - val_loss: 0.4144 - val_accuracy: 0.8158\n",
            "Epoch 718 / 1000 - loss: 0.2750 - accuracy: 0.9596 - val_loss: 0.4139 - val_accuracy: 0.8158\n",
            "Epoch 719 / 1000 - loss: 0.2747 - accuracy: 0.9596 - val_loss: 0.4137 - val_accuracy: 0.8158\n",
            "Epoch 720 / 1000 - loss: 0.2744 - accuracy: 0.9596 - val_loss: 0.4134 - val_accuracy: 0.8158\n",
            "Epoch 721 / 1000 - loss: 0.2741 - accuracy: 0.9596 - val_loss: 0.4130 - val_accuracy: 0.8158\n",
            "Epoch 722 / 1000 - loss: 0.2738 - accuracy: 0.9596 - val_loss: 0.4127 - val_accuracy: 0.8158\n",
            "Epoch 723 / 1000 - loss: 0.2735 - accuracy: 0.9596 - val_loss: 0.4125 - val_accuracy: 0.8158\n",
            "Epoch 724 / 1000 - loss: 0.2732 - accuracy: 0.9596 - val_loss: 0.4123 - val_accuracy: 0.8158\n",
            "Epoch 725 / 1000 - loss: 0.2729 - accuracy: 0.9596 - val_loss: 0.4123 - val_accuracy: 0.8158\n",
            "Epoch 726 / 1000 - loss: 0.2726 - accuracy: 0.9596 - val_loss: 0.4121 - val_accuracy: 0.8158\n",
            "Epoch 727 / 1000 - loss: 0.2723 - accuracy: 0.9596 - val_loss: 0.4119 - val_accuracy: 0.8158\n",
            "Epoch 728 / 1000 - loss: 0.2720 - accuracy: 0.9596 - val_loss: 0.4119 - val_accuracy: 0.8158\n",
            "Epoch 729 / 1000 - loss: 0.2717 - accuracy: 0.9596 - val_loss: 0.4114 - val_accuracy: 0.8158\n",
            "Epoch 730 / 1000 - loss: 0.2714 - accuracy: 0.9596 - val_loss: 0.4110 - val_accuracy: 0.8158\n",
            "Epoch 731 / 1000 - loss: 0.2711 - accuracy: 0.9596 - val_loss: 0.4114 - val_accuracy: 0.8158\n",
            "Epoch 732 / 1000 - loss: 0.2708 - accuracy: 0.9596 - val_loss: 0.4113 - val_accuracy: 0.8158\n",
            "Epoch 733 / 1000 - loss: 0.2705 - accuracy: 0.9596 - val_loss: 0.4110 - val_accuracy: 0.8158\n",
            "Epoch 734 / 1000 - loss: 0.2702 - accuracy: 0.9596 - val_loss: 0.4108 - val_accuracy: 0.8158\n",
            "Epoch 735 / 1000 - loss: 0.2699 - accuracy: 0.9596 - val_loss: 0.4108 - val_accuracy: 0.8158\n",
            "Epoch 736 / 1000 - loss: 0.2697 - accuracy: 0.9596 - val_loss: 0.4107 - val_accuracy: 0.8158\n",
            "Epoch 737 / 1000 - loss: 0.2694 - accuracy: 0.9596 - val_loss: 0.4107 - val_accuracy: 0.8158\n",
            "Epoch 738 / 1000 - loss: 0.2691 - accuracy: 0.9596 - val_loss: 0.4104 - val_accuracy: 0.8158\n",
            "Epoch 739 / 1000 - loss: 0.2688 - accuracy: 0.9596 - val_loss: 0.4101 - val_accuracy: 0.8158\n",
            "Epoch 740 / 1000 - loss: 0.2685 - accuracy: 0.9596 - val_loss: 0.4098 - val_accuracy: 0.8158\n",
            "Epoch 741 / 1000 - loss: 0.2682 - accuracy: 0.9596 - val_loss: 0.4092 - val_accuracy: 0.8158\n",
            "Epoch 742 / 1000 - loss: 0.2679 - accuracy: 0.9596 - val_loss: 0.4097 - val_accuracy: 0.8158\n",
            "Epoch 743 / 1000 - loss: 0.2676 - accuracy: 0.9596 - val_loss: 0.4095 - val_accuracy: 0.8158\n",
            "Epoch 744 / 1000 - loss: 0.2674 - accuracy: 0.9596 - val_loss: 0.4091 - val_accuracy: 0.8158\n",
            "Epoch 745 / 1000 - loss: 0.2671 - accuracy: 0.9596 - val_loss: 0.4094 - val_accuracy: 0.8158\n",
            "Epoch 746 / 1000 - loss: 0.2668 - accuracy: 0.9596 - val_loss: 0.4096 - val_accuracy: 0.8158\n",
            "Epoch 747 / 1000 - loss: 0.2665 - accuracy: 0.9596 - val_loss: 0.4090 - val_accuracy: 0.8158\n",
            "Epoch 748 / 1000 - loss: 0.2662 - accuracy: 0.9596 - val_loss: 0.4086 - val_accuracy: 0.8158\n",
            "Epoch 749 / 1000 - loss: 0.2659 - accuracy: 0.9596 - val_loss: 0.4085 - val_accuracy: 0.8158\n",
            "Epoch 750 / 1000 - loss: 0.2657 - accuracy: 0.9596 - val_loss: 0.4081 - val_accuracy: 0.8158\n",
            "Epoch 751 / 1000 - loss: 0.2654 - accuracy: 0.9596 - val_loss: 0.4082 - val_accuracy: 0.8158\n",
            "Epoch 752 / 1000 - loss: 0.2651 - accuracy: 0.9596 - val_loss: 0.4080 - val_accuracy: 0.8158\n",
            "Epoch 753 / 1000 - loss: 0.2648 - accuracy: 0.9596 - val_loss: 0.4080 - val_accuracy: 0.8158\n",
            "Epoch 754 / 1000 - loss: 0.2645 - accuracy: 0.9596 - val_loss: 0.4076 - val_accuracy: 0.8158\n",
            "Epoch 755 / 1000 - loss: 0.2643 - accuracy: 0.9596 - val_loss: 0.4073 - val_accuracy: 0.8158\n",
            "Epoch 756 / 1000 - loss: 0.2640 - accuracy: 0.9596 - val_loss: 0.4075 - val_accuracy: 0.8158\n",
            "Epoch 757 / 1000 - loss: 0.2637 - accuracy: 0.9596 - val_loss: 0.4069 - val_accuracy: 0.8158\n",
            "Epoch 758 / 1000 - loss: 0.2634 - accuracy: 0.9596 - val_loss: 0.4066 - val_accuracy: 0.8158\n",
            "Epoch 759 / 1000 - loss: 0.2632 - accuracy: 0.9596 - val_loss: 0.4062 - val_accuracy: 0.8158\n",
            "Epoch 760 / 1000 - loss: 0.2629 - accuracy: 0.9596 - val_loss: 0.4060 - val_accuracy: 0.8158\n",
            "Epoch 761 / 1000 - loss: 0.2626 - accuracy: 0.9596 - val_loss: 0.4061 - val_accuracy: 0.8158\n",
            "Epoch 762 / 1000 - loss: 0.2623 - accuracy: 0.9596 - val_loss: 0.4060 - val_accuracy: 0.8158\n",
            "Epoch 763 / 1000 - loss: 0.2621 - accuracy: 0.9596 - val_loss: 0.4059 - val_accuracy: 0.8158\n",
            "Epoch 764 / 1000 - loss: 0.2618 - accuracy: 0.9596 - val_loss: 0.4057 - val_accuracy: 0.8158\n",
            "Epoch 765 / 1000 - loss: 0.2615 - accuracy: 0.9596 - val_loss: 0.4054 - val_accuracy: 0.8158\n",
            "Epoch 766 / 1000 - loss: 0.2612 - accuracy: 0.9596 - val_loss: 0.4057 - val_accuracy: 0.8158\n",
            "Epoch 767 / 1000 - loss: 0.2610 - accuracy: 0.9596 - val_loss: 0.4054 - val_accuracy: 0.8158\n",
            "Epoch 768 / 1000 - loss: 0.2607 - accuracy: 0.9596 - val_loss: 0.4050 - val_accuracy: 0.8158\n",
            "Epoch 769 / 1000 - loss: 0.2604 - accuracy: 0.9596 - val_loss: 0.4047 - val_accuracy: 0.8158\n",
            "Epoch 770 / 1000 - loss: 0.2602 - accuracy: 0.9596 - val_loss: 0.4045 - val_accuracy: 0.8158\n",
            "Epoch 771 / 1000 - loss: 0.2599 - accuracy: 0.9596 - val_loss: 0.4043 - val_accuracy: 0.8158\n",
            "Epoch 772 / 1000 - loss: 0.2596 - accuracy: 0.9596 - val_loss: 0.4037 - val_accuracy: 0.8158\n",
            "Epoch 773 / 1000 - loss: 0.2594 - accuracy: 0.9596 - val_loss: 0.4033 - val_accuracy: 0.8158\n",
            "Epoch 774 / 1000 - loss: 0.2591 - accuracy: 0.9596 - val_loss: 0.4033 - val_accuracy: 0.8158\n",
            "Epoch 775 / 1000 - loss: 0.2588 - accuracy: 0.9596 - val_loss: 0.4031 - val_accuracy: 0.8158\n",
            "Epoch 776 / 1000 - loss: 0.2586 - accuracy: 0.9596 - val_loss: 0.4030 - val_accuracy: 0.8158\n",
            "Epoch 777 / 1000 - loss: 0.2583 - accuracy: 0.9596 - val_loss: 0.4033 - val_accuracy: 0.8158\n",
            "Epoch 778 / 1000 - loss: 0.2580 - accuracy: 0.9596 - val_loss: 0.4029 - val_accuracy: 0.8158\n",
            "Epoch 779 / 1000 - loss: 0.2578 - accuracy: 0.9596 - val_loss: 0.4028 - val_accuracy: 0.8158\n",
            "Epoch 780 / 1000 - loss: 0.2575 - accuracy: 0.9596 - val_loss: 0.4026 - val_accuracy: 0.8158\n",
            "Epoch 781 / 1000 - loss: 0.2572 - accuracy: 0.9596 - val_loss: 0.4025 - val_accuracy: 0.8158\n",
            "Epoch 782 / 1000 - loss: 0.2570 - accuracy: 0.9596 - val_loss: 0.4024 - val_accuracy: 0.8158\n",
            "Epoch 783 / 1000 - loss: 0.2567 - accuracy: 0.9596 - val_loss: 0.4025 - val_accuracy: 0.8158\n",
            "Epoch 784 / 1000 - loss: 0.2565 - accuracy: 0.9596 - val_loss: 0.4022 - val_accuracy: 0.8158\n",
            "Epoch 785 / 1000 - loss: 0.2562 - accuracy: 0.9596 - val_loss: 0.4018 - val_accuracy: 0.8158\n",
            "Epoch 786 / 1000 - loss: 0.2559 - accuracy: 0.9596 - val_loss: 0.4019 - val_accuracy: 0.8158\n",
            "Epoch 787 / 1000 - loss: 0.2557 - accuracy: 0.9596 - val_loss: 0.4020 - val_accuracy: 0.8158\n",
            "Epoch 788 / 1000 - loss: 0.2554 - accuracy: 0.9596 - val_loss: 0.4018 - val_accuracy: 0.8158\n",
            "Epoch 789 / 1000 - loss: 0.2552 - accuracy: 0.9596 - val_loss: 0.4015 - val_accuracy: 0.8158\n",
            "Epoch 790 / 1000 - loss: 0.2549 - accuracy: 0.9596 - val_loss: 0.4012 - val_accuracy: 0.8158\n",
            "Epoch 791 / 1000 - loss: 0.2546 - accuracy: 0.9596 - val_loss: 0.4007 - val_accuracy: 0.8158\n",
            "Epoch 792 / 1000 - loss: 0.2544 - accuracy: 0.9596 - val_loss: 0.4012 - val_accuracy: 0.8158\n",
            "Epoch 793 / 1000 - loss: 0.2541 - accuracy: 0.9596 - val_loss: 0.4011 - val_accuracy: 0.8158\n",
            "Epoch 794 / 1000 - loss: 0.2539 - accuracy: 0.9596 - val_loss: 0.4008 - val_accuracy: 0.8158\n",
            "Epoch 795 / 1000 - loss: 0.2536 - accuracy: 0.9596 - val_loss: 0.4005 - val_accuracy: 0.8158\n",
            "Epoch 796 / 1000 - loss: 0.2534 - accuracy: 0.9596 - val_loss: 0.4005 - val_accuracy: 0.8158\n",
            "Epoch 797 / 1000 - loss: 0.2531 - accuracy: 0.9596 - val_loss: 0.4002 - val_accuracy: 0.8158\n",
            "Epoch 798 / 1000 - loss: 0.2529 - accuracy: 0.9596 - val_loss: 0.4000 - val_accuracy: 0.8158\n",
            "Epoch 799 / 1000 - loss: 0.2526 - accuracy: 0.9596 - val_loss: 0.3999 - val_accuracy: 0.8158\n",
            "Epoch 800 / 1000 - loss: 0.2524 - accuracy: 0.9596 - val_loss: 0.3993 - val_accuracy: 0.8158\n",
            "Epoch 801 / 1000 - loss: 0.2521 - accuracy: 0.9596 - val_loss: 0.3992 - val_accuracy: 0.8158\n",
            "Epoch 802 / 1000 - loss: 0.2519 - accuracy: 0.9596 - val_loss: 0.3991 - val_accuracy: 0.8158\n",
            "Epoch 803 / 1000 - loss: 0.2516 - accuracy: 0.9596 - val_loss: 0.3988 - val_accuracy: 0.8158\n",
            "Epoch 804 / 1000 - loss: 0.2514 - accuracy: 0.9596 - val_loss: 0.3987 - val_accuracy: 0.8158\n",
            "Epoch 805 / 1000 - loss: 0.2511 - accuracy: 0.9596 - val_loss: 0.3986 - val_accuracy: 0.8158\n",
            "Epoch 806 / 1000 - loss: 0.2509 - accuracy: 0.9596 - val_loss: 0.3984 - val_accuracy: 0.8421\n",
            "Epoch 807 / 1000 - loss: 0.2506 - accuracy: 0.9596 - val_loss: 0.3981 - val_accuracy: 0.8421\n",
            "Epoch 808 / 1000 - loss: 0.2504 - accuracy: 0.9596 - val_loss: 0.3978 - val_accuracy: 0.8421\n",
            "Epoch 809 / 1000 - loss: 0.2501 - accuracy: 0.9596 - val_loss: 0.3978 - val_accuracy: 0.8421\n",
            "Epoch 810 / 1000 - loss: 0.2499 - accuracy: 0.9596 - val_loss: 0.3976 - val_accuracy: 0.8421\n",
            "Epoch 811 / 1000 - loss: 0.2496 - accuracy: 0.9596 - val_loss: 0.3976 - val_accuracy: 0.8421\n",
            "Epoch 812 / 1000 - loss: 0.2494 - accuracy: 0.9596 - val_loss: 0.3973 - val_accuracy: 0.8421\n",
            "Epoch 813 / 1000 - loss: 0.2491 - accuracy: 0.9596 - val_loss: 0.3973 - val_accuracy: 0.8421\n",
            "Epoch 814 / 1000 - loss: 0.2489 - accuracy: 0.9596 - val_loss: 0.3969 - val_accuracy: 0.8421\n",
            "Epoch 815 / 1000 - loss: 0.2486 - accuracy: 0.9596 - val_loss: 0.3968 - val_accuracy: 0.8421\n",
            "Epoch 816 / 1000 - loss: 0.2484 - accuracy: 0.9630 - val_loss: 0.3969 - val_accuracy: 0.8421\n",
            "Epoch 817 / 1000 - loss: 0.2482 - accuracy: 0.9630 - val_loss: 0.3968 - val_accuracy: 0.8421\n",
            "Epoch 818 / 1000 - loss: 0.2479 - accuracy: 0.9596 - val_loss: 0.3970 - val_accuracy: 0.8421\n",
            "Epoch 819 / 1000 - loss: 0.2477 - accuracy: 0.9596 - val_loss: 0.3969 - val_accuracy: 0.8421\n",
            "Epoch 820 / 1000 - loss: 0.2474 - accuracy: 0.9630 - val_loss: 0.3965 - val_accuracy: 0.8421\n",
            "Epoch 821 / 1000 - loss: 0.2472 - accuracy: 0.9630 - val_loss: 0.3962 - val_accuracy: 0.8421\n",
            "Epoch 822 / 1000 - loss: 0.2469 - accuracy: 0.9630 - val_loss: 0.3961 - val_accuracy: 0.8421\n",
            "Epoch 823 / 1000 - loss: 0.2467 - accuracy: 0.9630 - val_loss: 0.3961 - val_accuracy: 0.8421\n",
            "Epoch 824 / 1000 - loss: 0.2465 - accuracy: 0.9630 - val_loss: 0.3959 - val_accuracy: 0.8421\n",
            "Epoch 825 / 1000 - loss: 0.2462 - accuracy: 0.9630 - val_loss: 0.3957 - val_accuracy: 0.8421\n",
            "Epoch 826 / 1000 - loss: 0.2460 - accuracy: 0.9630 - val_loss: 0.3957 - val_accuracy: 0.8421\n",
            "Epoch 827 / 1000 - loss: 0.2458 - accuracy: 0.9630 - val_loss: 0.3959 - val_accuracy: 0.8421\n",
            "Epoch 828 / 1000 - loss: 0.2455 - accuracy: 0.9630 - val_loss: 0.3956 - val_accuracy: 0.8421\n",
            "Epoch 829 / 1000 - loss: 0.2453 - accuracy: 0.9630 - val_loss: 0.3952 - val_accuracy: 0.8421\n",
            "Epoch 830 / 1000 - loss: 0.2450 - accuracy: 0.9630 - val_loss: 0.3951 - val_accuracy: 0.8421\n",
            "Epoch 831 / 1000 - loss: 0.2448 - accuracy: 0.9630 - val_loss: 0.3948 - val_accuracy: 0.8421\n",
            "Epoch 832 / 1000 - loss: 0.2446 - accuracy: 0.9630 - val_loss: 0.3948 - val_accuracy: 0.8421\n",
            "Epoch 833 / 1000 - loss: 0.2443 - accuracy: 0.9630 - val_loss: 0.3947 - val_accuracy: 0.8421\n",
            "Epoch 834 / 1000 - loss: 0.2441 - accuracy: 0.9630 - val_loss: 0.3946 - val_accuracy: 0.8421\n",
            "Epoch 835 / 1000 - loss: 0.2439 - accuracy: 0.9630 - val_loss: 0.3942 - val_accuracy: 0.8421\n",
            "Epoch 836 / 1000 - loss: 0.2436 - accuracy: 0.9630 - val_loss: 0.3939 - val_accuracy: 0.8421\n",
            "Epoch 837 / 1000 - loss: 0.2434 - accuracy: 0.9630 - val_loss: 0.3939 - val_accuracy: 0.8421\n",
            "Epoch 838 / 1000 - loss: 0.2432 - accuracy: 0.9630 - val_loss: 0.3935 - val_accuracy: 0.8421\n",
            "Epoch 839 / 1000 - loss: 0.2429 - accuracy: 0.9630 - val_loss: 0.3934 - val_accuracy: 0.8421\n",
            "Epoch 840 / 1000 - loss: 0.2427 - accuracy: 0.9630 - val_loss: 0.3935 - val_accuracy: 0.8421\n",
            "Epoch 841 / 1000 - loss: 0.2425 - accuracy: 0.9630 - val_loss: 0.3931 - val_accuracy: 0.8421\n",
            "Epoch 842 / 1000 - loss: 0.2422 - accuracy: 0.9630 - val_loss: 0.3930 - val_accuracy: 0.8421\n",
            "Epoch 843 / 1000 - loss: 0.2420 - accuracy: 0.9630 - val_loss: 0.3930 - val_accuracy: 0.8421\n",
            "Epoch 844 / 1000 - loss: 0.2418 - accuracy: 0.9630 - val_loss: 0.3931 - val_accuracy: 0.8421\n",
            "Epoch 845 / 1000 - loss: 0.2416 - accuracy: 0.9630 - val_loss: 0.3929 - val_accuracy: 0.8421\n",
            "Epoch 846 / 1000 - loss: 0.2413 - accuracy: 0.9630 - val_loss: 0.3932 - val_accuracy: 0.8421\n",
            "Epoch 847 / 1000 - loss: 0.2411 - accuracy: 0.9630 - val_loss: 0.3928 - val_accuracy: 0.8421\n",
            "Epoch 848 / 1000 - loss: 0.2409 - accuracy: 0.9630 - val_loss: 0.3926 - val_accuracy: 0.8421\n",
            "Epoch 849 / 1000 - loss: 0.2406 - accuracy: 0.9630 - val_loss: 0.3922 - val_accuracy: 0.8421\n",
            "Epoch 850 / 1000 - loss: 0.2404 - accuracy: 0.9630 - val_loss: 0.3921 - val_accuracy: 0.8421\n",
            "Epoch 851 / 1000 - loss: 0.2402 - accuracy: 0.9630 - val_loss: 0.3919 - val_accuracy: 0.8421\n",
            "Epoch 852 / 1000 - loss: 0.2400 - accuracy: 0.9630 - val_loss: 0.3919 - val_accuracy: 0.8421\n",
            "Epoch 853 / 1000 - loss: 0.2397 - accuracy: 0.9630 - val_loss: 0.3916 - val_accuracy: 0.8421\n",
            "Epoch 854 / 1000 - loss: 0.2395 - accuracy: 0.9630 - val_loss: 0.3915 - val_accuracy: 0.8421\n",
            "Epoch 855 / 1000 - loss: 0.2393 - accuracy: 0.9630 - val_loss: 0.3913 - val_accuracy: 0.8421\n",
            "Epoch 856 / 1000 - loss: 0.2391 - accuracy: 0.9630 - val_loss: 0.3920 - val_accuracy: 0.8421\n",
            "Epoch 857 / 1000 - loss: 0.2388 - accuracy: 0.9630 - val_loss: 0.3917 - val_accuracy: 0.8421\n",
            "Epoch 858 / 1000 - loss: 0.2386 - accuracy: 0.9630 - val_loss: 0.3913 - val_accuracy: 0.8421\n",
            "Epoch 859 / 1000 - loss: 0.2384 - accuracy: 0.9630 - val_loss: 0.3910 - val_accuracy: 0.8421\n",
            "Epoch 860 / 1000 - loss: 0.2382 - accuracy: 0.9630 - val_loss: 0.3907 - val_accuracy: 0.8421\n",
            "Epoch 861 / 1000 - loss: 0.2380 - accuracy: 0.9630 - val_loss: 0.3907 - val_accuracy: 0.8421\n",
            "Epoch 862 / 1000 - loss: 0.2377 - accuracy: 0.9630 - val_loss: 0.3905 - val_accuracy: 0.8421\n",
            "Epoch 863 / 1000 - loss: 0.2375 - accuracy: 0.9630 - val_loss: 0.3903 - val_accuracy: 0.8421\n",
            "Epoch 864 / 1000 - loss: 0.2373 - accuracy: 0.9630 - val_loss: 0.3902 - val_accuracy: 0.8421\n",
            "Epoch 865 / 1000 - loss: 0.2371 - accuracy: 0.9630 - val_loss: 0.3902 - val_accuracy: 0.8421\n",
            "Epoch 866 / 1000 - loss: 0.2368 - accuracy: 0.9630 - val_loss: 0.3901 - val_accuracy: 0.8421\n",
            "Epoch 867 / 1000 - loss: 0.2366 - accuracy: 0.9630 - val_loss: 0.3898 - val_accuracy: 0.8421\n",
            "Epoch 868 / 1000 - loss: 0.2364 - accuracy: 0.9630 - val_loss: 0.3896 - val_accuracy: 0.8421\n",
            "Epoch 869 / 1000 - loss: 0.2362 - accuracy: 0.9630 - val_loss: 0.3892 - val_accuracy: 0.8421\n",
            "Epoch 870 / 1000 - loss: 0.2360 - accuracy: 0.9630 - val_loss: 0.3889 - val_accuracy: 0.8421\n",
            "Epoch 871 / 1000 - loss: 0.2358 - accuracy: 0.9630 - val_loss: 0.3890 - val_accuracy: 0.8421\n",
            "Epoch 872 / 1000 - loss: 0.2355 - accuracy: 0.9630 - val_loss: 0.3888 - val_accuracy: 0.8421\n",
            "Epoch 873 / 1000 - loss: 0.2353 - accuracy: 0.9630 - val_loss: 0.3886 - val_accuracy: 0.8421\n",
            "Epoch 874 / 1000 - loss: 0.2351 - accuracy: 0.9630 - val_loss: 0.3885 - val_accuracy: 0.8421\n",
            "Epoch 875 / 1000 - loss: 0.2349 - accuracy: 0.9630 - val_loss: 0.3885 - val_accuracy: 0.8421\n",
            "Epoch 876 / 1000 - loss: 0.2347 - accuracy: 0.9630 - val_loss: 0.3883 - val_accuracy: 0.8421\n",
            "Epoch 877 / 1000 - loss: 0.2345 - accuracy: 0.9630 - val_loss: 0.3882 - val_accuracy: 0.8421\n",
            "Epoch 878 / 1000 - loss: 0.2342 - accuracy: 0.9663 - val_loss: 0.3880 - val_accuracy: 0.8421\n",
            "Epoch 879 / 1000 - loss: 0.2340 - accuracy: 0.9630 - val_loss: 0.3881 - val_accuracy: 0.8421\n",
            "Epoch 880 / 1000 - loss: 0.2338 - accuracy: 0.9663 - val_loss: 0.3879 - val_accuracy: 0.8421\n",
            "Epoch 881 / 1000 - loss: 0.2336 - accuracy: 0.9663 - val_loss: 0.3879 - val_accuracy: 0.8421\n",
            "Epoch 882 / 1000 - loss: 0.2334 - accuracy: 0.9663 - val_loss: 0.3876 - val_accuracy: 0.8421\n",
            "Epoch 883 / 1000 - loss: 0.2332 - accuracy: 0.9663 - val_loss: 0.3877 - val_accuracy: 0.8421\n",
            "Epoch 884 / 1000 - loss: 0.2330 - accuracy: 0.9663 - val_loss: 0.3874 - val_accuracy: 0.8421\n",
            "Epoch 885 / 1000 - loss: 0.2328 - accuracy: 0.9663 - val_loss: 0.3873 - val_accuracy: 0.8421\n",
            "Epoch 886 / 1000 - loss: 0.2325 - accuracy: 0.9663 - val_loss: 0.3872 - val_accuracy: 0.8421\n",
            "Epoch 887 / 1000 - loss: 0.2323 - accuracy: 0.9663 - val_loss: 0.3872 - val_accuracy: 0.8421\n",
            "Epoch 888 / 1000 - loss: 0.2321 - accuracy: 0.9663 - val_loss: 0.3874 - val_accuracy: 0.8421\n",
            "Epoch 889 / 1000 - loss: 0.2319 - accuracy: 0.9663 - val_loss: 0.3871 - val_accuracy: 0.8421\n",
            "Epoch 890 / 1000 - loss: 0.2317 - accuracy: 0.9663 - val_loss: 0.3872 - val_accuracy: 0.8421\n",
            "Epoch 891 / 1000 - loss: 0.2315 - accuracy: 0.9663 - val_loss: 0.3870 - val_accuracy: 0.8421\n",
            "Epoch 892 / 1000 - loss: 0.2313 - accuracy: 0.9663 - val_loss: 0.3868 - val_accuracy: 0.8421\n",
            "Epoch 893 / 1000 - loss: 0.2311 - accuracy: 0.9663 - val_loss: 0.3865 - val_accuracy: 0.8421\n",
            "Epoch 894 / 1000 - loss: 0.2309 - accuracy: 0.9663 - val_loss: 0.3861 - val_accuracy: 0.8421\n",
            "Epoch 895 / 1000 - loss: 0.2307 - accuracy: 0.9663 - val_loss: 0.3859 - val_accuracy: 0.8421\n",
            "Epoch 896 / 1000 - loss: 0.2305 - accuracy: 0.9663 - val_loss: 0.3859 - val_accuracy: 0.8421\n",
            "Epoch 897 / 1000 - loss: 0.2303 - accuracy: 0.9663 - val_loss: 0.3858 - val_accuracy: 0.8421\n",
            "Epoch 898 / 1000 - loss: 0.2300 - accuracy: 0.9663 - val_loss: 0.3857 - val_accuracy: 0.8421\n",
            "Epoch 899 / 1000 - loss: 0.2298 - accuracy: 0.9663 - val_loss: 0.3856 - val_accuracy: 0.8421\n",
            "Epoch 900 / 1000 - loss: 0.2296 - accuracy: 0.9663 - val_loss: 0.3857 - val_accuracy: 0.8421\n",
            "Epoch 901 / 1000 - loss: 0.2294 - accuracy: 0.9663 - val_loss: 0.3855 - val_accuracy: 0.8421\n",
            "Epoch 902 / 1000 - loss: 0.2292 - accuracy: 0.9663 - val_loss: 0.3852 - val_accuracy: 0.8421\n",
            "Epoch 903 / 1000 - loss: 0.2290 - accuracy: 0.9663 - val_loss: 0.3850 - val_accuracy: 0.8421\n",
            "Epoch 904 / 1000 - loss: 0.2288 - accuracy: 0.9663 - val_loss: 0.3848 - val_accuracy: 0.8421\n",
            "Epoch 905 / 1000 - loss: 0.2286 - accuracy: 0.9663 - val_loss: 0.3845 - val_accuracy: 0.8421\n",
            "Epoch 906 / 1000 - loss: 0.2284 - accuracy: 0.9663 - val_loss: 0.3843 - val_accuracy: 0.8421\n",
            "Epoch 907 / 1000 - loss: 0.2282 - accuracy: 0.9663 - val_loss: 0.3841 - val_accuracy: 0.8421\n",
            "Epoch 908 / 1000 - loss: 0.2280 - accuracy: 0.9663 - val_loss: 0.3841 - val_accuracy: 0.8421\n",
            "Epoch 909 / 1000 - loss: 0.2278 - accuracy: 0.9663 - val_loss: 0.3839 - val_accuracy: 0.8421\n",
            "Epoch 910 / 1000 - loss: 0.2276 - accuracy: 0.9663 - val_loss: 0.3837 - val_accuracy: 0.8421\n",
            "Epoch 911 / 1000 - loss: 0.2274 - accuracy: 0.9663 - val_loss: 0.3837 - val_accuracy: 0.8421\n",
            "Epoch 912 / 1000 - loss: 0.2272 - accuracy: 0.9663 - val_loss: 0.3834 - val_accuracy: 0.8421\n",
            "Epoch 913 / 1000 - loss: 0.2270 - accuracy: 0.9663 - val_loss: 0.3835 - val_accuracy: 0.8421\n",
            "Epoch 914 / 1000 - loss: 0.2268 - accuracy: 0.9663 - val_loss: 0.3834 - val_accuracy: 0.8421\n",
            "Epoch 915 / 1000 - loss: 0.2266 - accuracy: 0.9663 - val_loss: 0.3834 - val_accuracy: 0.8421\n",
            "Epoch 916 / 1000 - loss: 0.2264 - accuracy: 0.9663 - val_loss: 0.3832 - val_accuracy: 0.8421\n",
            "Epoch 917 / 1000 - loss: 0.2262 - accuracy: 0.9663 - val_loss: 0.3831 - val_accuracy: 0.8421\n",
            "Epoch 918 / 1000 - loss: 0.2260 - accuracy: 0.9663 - val_loss: 0.3828 - val_accuracy: 0.8421\n",
            "Epoch 919 / 1000 - loss: 0.2258 - accuracy: 0.9663 - val_loss: 0.3828 - val_accuracy: 0.8421\n",
            "Epoch 920 / 1000 - loss: 0.2256 - accuracy: 0.9663 - val_loss: 0.3829 - val_accuracy: 0.8421\n",
            "Epoch 921 / 1000 - loss: 0.2254 - accuracy: 0.9663 - val_loss: 0.3827 - val_accuracy: 0.8421\n",
            "Epoch 922 / 1000 - loss: 0.2252 - accuracy: 0.9663 - val_loss: 0.3828 - val_accuracy: 0.8421\n",
            "Epoch 923 / 1000 - loss: 0.2250 - accuracy: 0.9663 - val_loss: 0.3828 - val_accuracy: 0.8421\n",
            "Epoch 924 / 1000 - loss: 0.2248 - accuracy: 0.9663 - val_loss: 0.3827 - val_accuracy: 0.8421\n",
            "Epoch 925 / 1000 - loss: 0.2246 - accuracy: 0.9663 - val_loss: 0.3826 - val_accuracy: 0.8421\n",
            "Epoch 926 / 1000 - loss: 0.2244 - accuracy: 0.9663 - val_loss: 0.3825 - val_accuracy: 0.8421\n",
            "Epoch 927 / 1000 - loss: 0.2242 - accuracy: 0.9663 - val_loss: 0.3823 - val_accuracy: 0.8421\n",
            "Epoch 928 / 1000 - loss: 0.2240 - accuracy: 0.9663 - val_loss: 0.3820 - val_accuracy: 0.8421\n",
            "Epoch 929 / 1000 - loss: 0.2239 - accuracy: 0.9663 - val_loss: 0.3819 - val_accuracy: 0.8421\n",
            "Epoch 930 / 1000 - loss: 0.2237 - accuracy: 0.9663 - val_loss: 0.3817 - val_accuracy: 0.8421\n",
            "Epoch 931 / 1000 - loss: 0.2235 - accuracy: 0.9663 - val_loss: 0.3818 - val_accuracy: 0.8421\n",
            "Epoch 932 / 1000 - loss: 0.2233 - accuracy: 0.9663 - val_loss: 0.3819 - val_accuracy: 0.8421\n",
            "Epoch 933 / 1000 - loss: 0.2231 - accuracy: 0.9663 - val_loss: 0.3818 - val_accuracy: 0.8421\n",
            "Epoch 934 / 1000 - loss: 0.2229 - accuracy: 0.9663 - val_loss: 0.3819 - val_accuracy: 0.8421\n",
            "Epoch 935 / 1000 - loss: 0.2227 - accuracy: 0.9663 - val_loss: 0.3816 - val_accuracy: 0.8421\n",
            "Epoch 936 / 1000 - loss: 0.2225 - accuracy: 0.9663 - val_loss: 0.3814 - val_accuracy: 0.8421\n",
            "Epoch 937 / 1000 - loss: 0.2223 - accuracy: 0.9663 - val_loss: 0.3812 - val_accuracy: 0.8421\n",
            "Epoch 938 / 1000 - loss: 0.2221 - accuracy: 0.9663 - val_loss: 0.3812 - val_accuracy: 0.8421\n",
            "Epoch 939 / 1000 - loss: 0.2219 - accuracy: 0.9663 - val_loss: 0.3809 - val_accuracy: 0.8421\n",
            "Epoch 940 / 1000 - loss: 0.2217 - accuracy: 0.9663 - val_loss: 0.3808 - val_accuracy: 0.8421\n",
            "Epoch 941 / 1000 - loss: 0.2215 - accuracy: 0.9663 - val_loss: 0.3809 - val_accuracy: 0.8421\n",
            "Epoch 942 / 1000 - loss: 0.2214 - accuracy: 0.9663 - val_loss: 0.3806 - val_accuracy: 0.8421\n",
            "Epoch 943 / 1000 - loss: 0.2212 - accuracy: 0.9663 - val_loss: 0.3804 - val_accuracy: 0.8421\n",
            "Epoch 944 / 1000 - loss: 0.2210 - accuracy: 0.9663 - val_loss: 0.3803 - val_accuracy: 0.8421\n",
            "Epoch 945 / 1000 - loss: 0.2208 - accuracy: 0.9663 - val_loss: 0.3802 - val_accuracy: 0.8421\n",
            "Epoch 946 / 1000 - loss: 0.2206 - accuracy: 0.9663 - val_loss: 0.3802 - val_accuracy: 0.8421\n",
            "Epoch 947 / 1000 - loss: 0.2204 - accuracy: 0.9663 - val_loss: 0.3804 - val_accuracy: 0.8421\n",
            "Epoch 948 / 1000 - loss: 0.2202 - accuracy: 0.9663 - val_loss: 0.3802 - val_accuracy: 0.8421\n",
            "Epoch 949 / 1000 - loss: 0.2200 - accuracy: 0.9663 - val_loss: 0.3800 - val_accuracy: 0.8421\n",
            "Epoch 950 / 1000 - loss: 0.2199 - accuracy: 0.9663 - val_loss: 0.3799 - val_accuracy: 0.8421\n",
            "Epoch 951 / 1000 - loss: 0.2197 - accuracy: 0.9663 - val_loss: 0.3800 - val_accuracy: 0.8421\n",
            "Epoch 952 / 1000 - loss: 0.2195 - accuracy: 0.9663 - val_loss: 0.3800 - val_accuracy: 0.8421\n",
            "Epoch 953 / 1000 - loss: 0.2193 - accuracy: 0.9663 - val_loss: 0.3798 - val_accuracy: 0.8421\n",
            "Epoch 954 / 1000 - loss: 0.2191 - accuracy: 0.9663 - val_loss: 0.3798 - val_accuracy: 0.8421\n",
            "Epoch 955 / 1000 - loss: 0.2189 - accuracy: 0.9663 - val_loss: 0.3798 - val_accuracy: 0.8421\n",
            "Epoch 956 / 1000 - loss: 0.2187 - accuracy: 0.9663 - val_loss: 0.3797 - val_accuracy: 0.8421\n",
            "Epoch 957 / 1000 - loss: 0.2186 - accuracy: 0.9663 - val_loss: 0.3796 - val_accuracy: 0.8421\n",
            "Epoch 958 / 1000 - loss: 0.2184 - accuracy: 0.9663 - val_loss: 0.3793 - val_accuracy: 0.8421\n",
            "Epoch 959 / 1000 - loss: 0.2182 - accuracy: 0.9663 - val_loss: 0.3789 - val_accuracy: 0.8421\n",
            "Epoch 960 / 1000 - loss: 0.2180 - accuracy: 0.9663 - val_loss: 0.3786 - val_accuracy: 0.8421\n",
            "Epoch 961 / 1000 - loss: 0.2178 - accuracy: 0.9663 - val_loss: 0.3787 - val_accuracy: 0.8421\n",
            "Epoch 962 / 1000 - loss: 0.2176 - accuracy: 0.9663 - val_loss: 0.3786 - val_accuracy: 0.8421\n",
            "Epoch 963 / 1000 - loss: 0.2175 - accuracy: 0.9663 - val_loss: 0.3785 - val_accuracy: 0.8421\n",
            "Epoch 964 / 1000 - loss: 0.2173 - accuracy: 0.9663 - val_loss: 0.3783 - val_accuracy: 0.8421\n",
            "Epoch 965 / 1000 - loss: 0.2171 - accuracy: 0.9663 - val_loss: 0.3783 - val_accuracy: 0.8421\n",
            "Epoch 966 / 1000 - loss: 0.2169 - accuracy: 0.9663 - val_loss: 0.3783 - val_accuracy: 0.8421\n",
            "Epoch 967 / 1000 - loss: 0.2167 - accuracy: 0.9663 - val_loss: 0.3784 - val_accuracy: 0.8421\n",
            "Epoch 968 / 1000 - loss: 0.2166 - accuracy: 0.9663 - val_loss: 0.3783 - val_accuracy: 0.8421\n",
            "Epoch 969 / 1000 - loss: 0.2164 - accuracy: 0.9663 - val_loss: 0.3782 - val_accuracy: 0.8421\n",
            "Epoch 970 / 1000 - loss: 0.2162 - accuracy: 0.9663 - val_loss: 0.3780 - val_accuracy: 0.8421\n",
            "Epoch 971 / 1000 - loss: 0.2160 - accuracy: 0.9663 - val_loss: 0.3779 - val_accuracy: 0.8421\n",
            "Epoch 972 / 1000 - loss: 0.2158 - accuracy: 0.9663 - val_loss: 0.3778 - val_accuracy: 0.8421\n",
            "Epoch 973 / 1000 - loss: 0.2157 - accuracy: 0.9663 - val_loss: 0.3778 - val_accuracy: 0.8421\n",
            "Epoch 974 / 1000 - loss: 0.2155 - accuracy: 0.9663 - val_loss: 0.3777 - val_accuracy: 0.8421\n",
            "Epoch 975 / 1000 - loss: 0.2153 - accuracy: 0.9663 - val_loss: 0.3776 - val_accuracy: 0.8421\n",
            "Epoch 976 / 1000 - loss: 0.2151 - accuracy: 0.9663 - val_loss: 0.3775 - val_accuracy: 0.8421\n",
            "Epoch 977 / 1000 - loss: 0.2150 - accuracy: 0.9663 - val_loss: 0.3774 - val_accuracy: 0.8421\n",
            "Epoch 978 / 1000 - loss: 0.2148 - accuracy: 0.9663 - val_loss: 0.3772 - val_accuracy: 0.8421\n",
            "Epoch 979 / 1000 - loss: 0.2146 - accuracy: 0.9663 - val_loss: 0.3776 - val_accuracy: 0.8421\n",
            "Epoch 980 / 1000 - loss: 0.2144 - accuracy: 0.9663 - val_loss: 0.3775 - val_accuracy: 0.8421\n",
            "Epoch 981 / 1000 - loss: 0.2142 - accuracy: 0.9663 - val_loss: 0.3772 - val_accuracy: 0.8421\n",
            "Epoch 982 / 1000 - loss: 0.2141 - accuracy: 0.9663 - val_loss: 0.3771 - val_accuracy: 0.8421\n",
            "Epoch 983 / 1000 - loss: 0.2139 - accuracy: 0.9663 - val_loss: 0.3768 - val_accuracy: 0.8421\n",
            "Epoch 984 / 1000 - loss: 0.2137 - accuracy: 0.9663 - val_loss: 0.3765 - val_accuracy: 0.8421\n",
            "Epoch 985 / 1000 - loss: 0.2135 - accuracy: 0.9663 - val_loss: 0.3763 - val_accuracy: 0.8421\n",
            "Epoch 986 / 1000 - loss: 0.2134 - accuracy: 0.9663 - val_loss: 0.3760 - val_accuracy: 0.8421\n",
            "Epoch 987 / 1000 - loss: 0.2132 - accuracy: 0.9663 - val_loss: 0.3758 - val_accuracy: 0.8421\n",
            "Epoch 988 / 1000 - loss: 0.2130 - accuracy: 0.9663 - val_loss: 0.3756 - val_accuracy: 0.8421\n",
            "Epoch 989 / 1000 - loss: 0.2129 - accuracy: 0.9663 - val_loss: 0.3757 - val_accuracy: 0.8421\n",
            "Epoch 990 / 1000 - loss: 0.2127 - accuracy: 0.9663 - val_loss: 0.3757 - val_accuracy: 0.8421\n",
            "Epoch 991 / 1000 - loss: 0.2125 - accuracy: 0.9663 - val_loss: 0.3756 - val_accuracy: 0.8421\n",
            "Epoch 992 / 1000 - loss: 0.2123 - accuracy: 0.9663 - val_loss: 0.3755 - val_accuracy: 0.8421\n",
            "Epoch 993 / 1000 - loss: 0.2122 - accuracy: 0.9663 - val_loss: 0.3757 - val_accuracy: 0.8421\n",
            "Epoch 994 / 1000 - loss: 0.2120 - accuracy: 0.9663 - val_loss: 0.3756 - val_accuracy: 0.8421\n",
            "Epoch 995 / 1000 - loss: 0.2118 - accuracy: 0.9663 - val_loss: 0.3753 - val_accuracy: 0.8421\n",
            "Epoch 996 / 1000 - loss: 0.2116 - accuracy: 0.9663 - val_loss: 0.3753 - val_accuracy: 0.8421\n",
            "Epoch 997 / 1000 - loss: 0.2115 - accuracy: 0.9663 - val_loss: 0.3751 - val_accuracy: 0.8421\n",
            "Epoch 998 / 1000 - loss: 0.2113 - accuracy: 0.9663 - val_loss: 0.3751 - val_accuracy: 0.8421\n",
            "Epoch 999 / 1000 - loss: 0.2111 - accuracy: 0.9663 - val_loss: 0.3749 - val_accuracy: 0.8421\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1152x720 with 4 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "id": "mbGv2V4JNd8b",
        "outputId": "2200fc66-3c85-4c71-8da7-e02381492cfd"
      },
      "source": [
        "plt.title('Weights change')\n",
        "plt.ylabel('Weights')\n",
        "plt.xlabel('Iterations')\n",
        "plt.plot(np.array(all_w).reshape((len(all_w),all_w[0].shape[0]*all_w[0].shape[1])))\n",
        "plt.show()"
      ],
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "uth2pIDkPYLp",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "bf744768-23d8-4e80-d971-36423ffc5b16"
      },
      "source": [
        "out_history=[]\n",
        "all_w=[]\n",
        "eta=0.004\n",
        "model3 = NN_model([\n",
        "            RBF_layer(in_dim=train_data.shape[0], out_dim=32 , cent_lr = eta, wid_lr=eta,data =train_data.T,act= \"gaussian\"),\n",
        "            One_layer(in_dim=32,out_dim=5,act=\"softmax\",lr=eta,apt_type='AdaGrad',val=0.2)],\n",
        "            batch_size = 8,loss='Multi_class'\n",
        "    ) \n",
        "history,all_w = model3.training(train_data,valid_data,train_label,valid_label,epochs=1000)\n",
        "out_history.append(history)\n",
        "\n",
        "def draw(hh,title):\n",
        "    fig = plt.figure(figsize=(16,10))\n",
        "    fig.subplots_adjust(hspace=0.25, wspace=0.2)\n",
        "    plt.subplot(2,2,1)\n",
        "    for i,h in enumerate(hh):\n",
        "        plt.plot(h.T[2])\n",
        "    plt.title('Training accuracy')\n",
        "    plt.ylabel('Accuracy')\n",
        "    plt.xlabel('Epoch')\n",
        "\n",
        "    plt.subplot(2,2,2)\n",
        "    for i,h in enumerate(hh):\n",
        "        plt.plot(h.T[3])\n",
        "    plt.title('Validation accuracy')\n",
        "    plt.ylabel('Accuracy')\n",
        "    plt.xlabel('Epoch')\n",
        "    plt.subplot(2,2,3)\n",
        "    for i,h in enumerate(hh):\n",
        "        plt.plot(h.T[0])\n",
        "    plt.title('Training loss')\n",
        "    plt.ylabel('Loss')\n",
        "    plt.xlabel('Epoch')\n",
        "\n",
        "    plt.subplot(2,2,4)\n",
        "    for i,h in enumerate(hh):\n",
        "        plt.plot(h.T[1])\n",
        "    plt.title('Validation loss')\n",
        "    plt.ylabel('Loss')\n",
        "    plt.xlabel('Epoch')\n",
        "\n",
        "    fig.suptitle(title, fontsize=16)\n",
        "    plt.show()\n",
        "draw(out_history,\"AdaGrad\")"
      ],
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 0 / 1000 - loss: 1.5621 - accuracy: 0.2761 - val_loss: 1.5750 - val_accuracy: 0.2895\n",
            "Epoch 1 / 1000 - loss: 1.5409 - accuracy: 0.4007 - val_loss: 1.5706 - val_accuracy: 0.2368\n",
            "Epoch 2 / 1000 - loss: 1.5222 - accuracy: 0.4007 - val_loss: 1.5612 - val_accuracy: 0.2368\n",
            "Epoch 3 / 1000 - loss: 1.5065 - accuracy: 0.4175 - val_loss: 1.5511 - val_accuracy: 0.2632\n",
            "Epoch 4 / 1000 - loss: 1.4919 - accuracy: 0.4175 - val_loss: 1.5409 - val_accuracy: 0.2632\n",
            "Epoch 5 / 1000 - loss: 1.4777 - accuracy: 0.5118 - val_loss: 1.5326 - val_accuracy: 0.3158\n",
            "Epoch 6 / 1000 - loss: 1.4636 - accuracy: 0.5118 - val_loss: 1.5251 - val_accuracy: 0.3158\n",
            "Epoch 7 / 1000 - loss: 1.4501 - accuracy: 0.5118 - val_loss: 1.5166 - val_accuracy: 0.3158\n",
            "Epoch 8 / 1000 - loss: 1.4372 - accuracy: 0.5118 - val_loss: 1.5106 - val_accuracy: 0.3158\n",
            "Epoch 9 / 1000 - loss: 1.4250 - accuracy: 0.5387 - val_loss: 1.5019 - val_accuracy: 0.3421\n",
            "Epoch 10 / 1000 - loss: 1.4130 - accuracy: 0.5387 - val_loss: 1.4936 - val_accuracy: 0.3421\n",
            "Epoch 11 / 1000 - loss: 1.4013 - accuracy: 0.5387 - val_loss: 1.4878 - val_accuracy: 0.3421\n",
            "Epoch 12 / 1000 - loss: 1.3896 - accuracy: 0.5387 - val_loss: 1.4809 - val_accuracy: 0.3421\n",
            "Epoch 13 / 1000 - loss: 1.3782 - accuracy: 0.5387 - val_loss: 1.4736 - val_accuracy: 0.3421\n",
            "Epoch 14 / 1000 - loss: 1.3671 - accuracy: 0.5387 - val_loss: 1.4672 - val_accuracy: 0.3421\n",
            "Epoch 15 / 1000 - loss: 1.3564 - accuracy: 0.5387 - val_loss: 1.4629 - val_accuracy: 0.3421\n",
            "Epoch 16 / 1000 - loss: 1.3460 - accuracy: 0.5387 - val_loss: 1.4568 - val_accuracy: 0.3421\n",
            "Epoch 17 / 1000 - loss: 1.3357 - accuracy: 0.5387 - val_loss: 1.4498 - val_accuracy: 0.3421\n",
            "Epoch 18 / 1000 - loss: 1.3257 - accuracy: 0.5387 - val_loss: 1.4440 - val_accuracy: 0.3421\n",
            "Epoch 19 / 1000 - loss: 1.3159 - accuracy: 0.5387 - val_loss: 1.4376 - val_accuracy: 0.3421\n",
            "Epoch 20 / 1000 - loss: 1.3067 - accuracy: 0.5387 - val_loss: 1.4312 - val_accuracy: 0.3421\n",
            "Epoch 21 / 1000 - loss: 1.2974 - accuracy: 0.5387 - val_loss: 1.4264 - val_accuracy: 0.3421\n",
            "Epoch 22 / 1000 - loss: 1.2884 - accuracy: 0.5387 - val_loss: 1.4205 - val_accuracy: 0.3421\n",
            "Epoch 23 / 1000 - loss: 1.2798 - accuracy: 0.5387 - val_loss: 1.4146 - val_accuracy: 0.3421\n",
            "Epoch 24 / 1000 - loss: 1.2714 - accuracy: 0.5387 - val_loss: 1.4083 - val_accuracy: 0.3421\n",
            "Epoch 25 / 1000 - loss: 1.2632 - accuracy: 0.5387 - val_loss: 1.4029 - val_accuracy: 0.3421\n",
            "Epoch 26 / 1000 - loss: 1.2554 - accuracy: 0.5387 - val_loss: 1.3980 - val_accuracy: 0.3421\n",
            "Epoch 27 / 1000 - loss: 1.2478 - accuracy: 0.5387 - val_loss: 1.3921 - val_accuracy: 0.3421\n",
            "Epoch 28 / 1000 - loss: 1.2405 - accuracy: 0.5387 - val_loss: 1.3872 - val_accuracy: 0.3421\n",
            "Epoch 29 / 1000 - loss: 1.2336 - accuracy: 0.5387 - val_loss: 1.3833 - val_accuracy: 0.3421\n",
            "Epoch 30 / 1000 - loss: 1.2269 - accuracy: 0.5387 - val_loss: 1.3789 - val_accuracy: 0.3421\n",
            "Epoch 31 / 1000 - loss: 1.2205 - accuracy: 0.5387 - val_loss: 1.3731 - val_accuracy: 0.3421\n",
            "Epoch 32 / 1000 - loss: 1.2141 - accuracy: 0.5387 - val_loss: 1.3698 - val_accuracy: 0.3421\n",
            "Epoch 33 / 1000 - loss: 1.2082 - accuracy: 0.5387 - val_loss: 1.3651 - val_accuracy: 0.3421\n",
            "Epoch 34 / 1000 - loss: 1.2024 - accuracy: 0.5387 - val_loss: 1.3600 - val_accuracy: 0.3421\n",
            "Epoch 35 / 1000 - loss: 1.1967 - accuracy: 0.5387 - val_loss: 1.3554 - val_accuracy: 0.3421\n",
            "Epoch 36 / 1000 - loss: 1.1911 - accuracy: 0.5387 - val_loss: 1.3516 - val_accuracy: 0.3421\n",
            "Epoch 37 / 1000 - loss: 1.1859 - accuracy: 0.5387 - val_loss: 1.3473 - val_accuracy: 0.3421\n",
            "Epoch 38 / 1000 - loss: 1.1806 - accuracy: 0.5387 - val_loss: 1.3445 - val_accuracy: 0.3421\n",
            "Epoch 39 / 1000 - loss: 1.1756 - accuracy: 0.5387 - val_loss: 1.3409 - val_accuracy: 0.3421\n",
            "Epoch 40 / 1000 - loss: 1.1706 - accuracy: 0.5387 - val_loss: 1.3373 - val_accuracy: 0.3421\n",
            "Epoch 41 / 1000 - loss: 1.1658 - accuracy: 0.5387 - val_loss: 1.3333 - val_accuracy: 0.3421\n",
            "Epoch 42 / 1000 - loss: 1.1612 - accuracy: 0.5387 - val_loss: 1.3298 - val_accuracy: 0.3421\n",
            "Epoch 43 / 1000 - loss: 1.1565 - accuracy: 0.5387 - val_loss: 1.3257 - val_accuracy: 0.3421\n",
            "Epoch 44 / 1000 - loss: 1.1521 - accuracy: 0.5387 - val_loss: 1.3228 - val_accuracy: 0.3421\n",
            "Epoch 45 / 1000 - loss: 1.1479 - accuracy: 0.5387 - val_loss: 1.3198 - val_accuracy: 0.3421\n",
            "Epoch 46 / 1000 - loss: 1.1438 - accuracy: 0.5387 - val_loss: 1.3165 - val_accuracy: 0.3421\n",
            "Epoch 47 / 1000 - loss: 1.1398 - accuracy: 0.5387 - val_loss: 1.3130 - val_accuracy: 0.3421\n",
            "Epoch 48 / 1000 - loss: 1.1359 - accuracy: 0.5387 - val_loss: 1.3100 - val_accuracy: 0.3421\n",
            "Epoch 49 / 1000 - loss: 1.1321 - accuracy: 0.5387 - val_loss: 1.3065 - val_accuracy: 0.3421\n",
            "Epoch 50 / 1000 - loss: 1.1283 - accuracy: 0.5387 - val_loss: 1.3034 - val_accuracy: 0.3421\n",
            "Epoch 51 / 1000 - loss: 1.1246 - accuracy: 0.5387 - val_loss: 1.3012 - val_accuracy: 0.3421\n",
            "Epoch 52 / 1000 - loss: 1.1210 - accuracy: 0.5387 - val_loss: 1.2993 - val_accuracy: 0.3421\n",
            "Epoch 53 / 1000 - loss: 1.1174 - accuracy: 0.5387 - val_loss: 1.2961 - val_accuracy: 0.3421\n",
            "Epoch 54 / 1000 - loss: 1.1140 - accuracy: 0.5387 - val_loss: 1.2931 - val_accuracy: 0.3421\n",
            "Epoch 55 / 1000 - loss: 1.1106 - accuracy: 0.5387 - val_loss: 1.2903 - val_accuracy: 0.3421\n",
            "Epoch 56 / 1000 - loss: 1.1073 - accuracy: 0.5387 - val_loss: 1.2885 - val_accuracy: 0.3421\n",
            "Epoch 57 / 1000 - loss: 1.1041 - accuracy: 0.5387 - val_loss: 1.2855 - val_accuracy: 0.3421\n",
            "Epoch 58 / 1000 - loss: 1.1008 - accuracy: 0.5387 - val_loss: 1.2825 - val_accuracy: 0.3421\n",
            "Epoch 59 / 1000 - loss: 1.0977 - accuracy: 0.5387 - val_loss: 1.2794 - val_accuracy: 0.3421\n",
            "Epoch 60 / 1000 - loss: 1.0946 - accuracy: 0.5387 - val_loss: 1.2768 - val_accuracy: 0.3421\n",
            "Epoch 61 / 1000 - loss: 1.0915 - accuracy: 0.5387 - val_loss: 1.2738 - val_accuracy: 0.3421\n",
            "Epoch 62 / 1000 - loss: 1.0886 - accuracy: 0.5387 - val_loss: 1.2708 - val_accuracy: 0.3421\n",
            "Epoch 63 / 1000 - loss: 1.0856 - accuracy: 0.5387 - val_loss: 1.2683 - val_accuracy: 0.3421\n",
            "Epoch 64 / 1000 - loss: 1.0830 - accuracy: 0.5387 - val_loss: 1.2648 - val_accuracy: 0.3421\n",
            "Epoch 65 / 1000 - loss: 1.0803 - accuracy: 0.5387 - val_loss: 1.2625 - val_accuracy: 0.3421\n",
            "Epoch 66 / 1000 - loss: 1.0775 - accuracy: 0.5455 - val_loss: 1.2604 - val_accuracy: 0.3421\n",
            "Epoch 67 / 1000 - loss: 1.0747 - accuracy: 0.5556 - val_loss: 1.2578 - val_accuracy: 0.3684\n",
            "Epoch 68 / 1000 - loss: 1.0721 - accuracy: 0.5589 - val_loss: 1.2553 - val_accuracy: 0.3684\n",
            "Epoch 69 / 1000 - loss: 1.0693 - accuracy: 0.5623 - val_loss: 1.2534 - val_accuracy: 0.3684\n",
            "Epoch 70 / 1000 - loss: 1.0668 - accuracy: 0.5623 - val_loss: 1.2517 - val_accuracy: 0.3684\n",
            "Epoch 71 / 1000 - loss: 1.0643 - accuracy: 0.5623 - val_loss: 1.2494 - val_accuracy: 0.3684\n",
            "Epoch 72 / 1000 - loss: 1.0618 - accuracy: 0.5623 - val_loss: 1.2475 - val_accuracy: 0.3684\n",
            "Epoch 73 / 1000 - loss: 1.0593 - accuracy: 0.5623 - val_loss: 1.2454 - val_accuracy: 0.3684\n",
            "Epoch 74 / 1000 - loss: 1.0570 - accuracy: 0.5623 - val_loss: 1.2436 - val_accuracy: 0.3684\n",
            "Epoch 75 / 1000 - loss: 1.0545 - accuracy: 0.5623 - val_loss: 1.2423 - val_accuracy: 0.3684\n",
            "Epoch 76 / 1000 - loss: 1.0524 - accuracy: 0.5623 - val_loss: 1.2394 - val_accuracy: 0.3684\n",
            "Epoch 77 / 1000 - loss: 1.0498 - accuracy: 0.5623 - val_loss: 1.2380 - val_accuracy: 0.3684\n",
            "Epoch 78 / 1000 - loss: 1.0475 - accuracy: 0.5623 - val_loss: 1.2363 - val_accuracy: 0.3684\n",
            "Epoch 79 / 1000 - loss: 1.0454 - accuracy: 0.5623 - val_loss: 1.2341 - val_accuracy: 0.3684\n",
            "Epoch 80 / 1000 - loss: 1.0431 - accuracy: 0.5623 - val_loss: 1.2330 - val_accuracy: 0.3684\n",
            "Epoch 81 / 1000 - loss: 1.0409 - accuracy: 0.5623 - val_loss: 1.2309 - val_accuracy: 0.3684\n",
            "Epoch 82 / 1000 - loss: 1.0388 - accuracy: 0.5623 - val_loss: 1.2288 - val_accuracy: 0.3684\n",
            "Epoch 83 / 1000 - loss: 1.0367 - accuracy: 0.5623 - val_loss: 1.2275 - val_accuracy: 0.3684\n",
            "Epoch 84 / 1000 - loss: 1.0345 - accuracy: 0.5623 - val_loss: 1.2255 - val_accuracy: 0.3684\n",
            "Epoch 85 / 1000 - loss: 1.0324 - accuracy: 0.5623 - val_loss: 1.2238 - val_accuracy: 0.3684\n",
            "Epoch 86 / 1000 - loss: 1.0304 - accuracy: 0.5623 - val_loss: 1.2221 - val_accuracy: 0.3684\n",
            "Epoch 87 / 1000 - loss: 1.0288 - accuracy: 0.5623 - val_loss: 1.2198 - val_accuracy: 0.3684\n",
            "Epoch 88 / 1000 - loss: 1.0266 - accuracy: 0.5623 - val_loss: 1.2179 - val_accuracy: 0.3684\n",
            "Epoch 89 / 1000 - loss: 1.0245 - accuracy: 0.5623 - val_loss: 1.2169 - val_accuracy: 0.3684\n",
            "Epoch 90 / 1000 - loss: 1.0225 - accuracy: 0.5623 - val_loss: 1.2150 - val_accuracy: 0.3684\n",
            "Epoch 91 / 1000 - loss: 1.0205 - accuracy: 0.5623 - val_loss: 1.2133 - val_accuracy: 0.3684\n",
            "Epoch 92 / 1000 - loss: 1.0186 - accuracy: 0.5690 - val_loss: 1.2125 - val_accuracy: 0.3684\n",
            "Epoch 93 / 1000 - loss: 1.0170 - accuracy: 0.5758 - val_loss: 1.2095 - val_accuracy: 0.3947\n",
            "Epoch 94 / 1000 - loss: 1.0148 - accuracy: 0.5825 - val_loss: 1.2087 - val_accuracy: 0.3947\n",
            "Epoch 95 / 1000 - loss: 1.0131 - accuracy: 0.5825 - val_loss: 1.2066 - val_accuracy: 0.3947\n",
            "Epoch 96 / 1000 - loss: 1.0112 - accuracy: 0.5791 - val_loss: 1.2049 - val_accuracy: 0.3947\n",
            "Epoch 97 / 1000 - loss: 1.0094 - accuracy: 0.5758 - val_loss: 1.2030 - val_accuracy: 0.3947\n",
            "Epoch 98 / 1000 - loss: 1.0076 - accuracy: 0.5758 - val_loss: 1.2021 - val_accuracy: 0.3947\n",
            "Epoch 99 / 1000 - loss: 1.0059 - accuracy: 0.5758 - val_loss: 1.1996 - val_accuracy: 0.3947\n",
            "Epoch 100 / 1000 - loss: 1.0040 - accuracy: 0.5657 - val_loss: 1.1986 - val_accuracy: 0.3947\n",
            "Epoch 101 / 1000 - loss: 1.0024 - accuracy: 0.5657 - val_loss: 1.1970 - val_accuracy: 0.3947\n",
            "Epoch 102 / 1000 - loss: 1.0006 - accuracy: 0.5657 - val_loss: 1.1951 - val_accuracy: 0.3947\n",
            "Epoch 103 / 1000 - loss: 0.9989 - accuracy: 0.5657 - val_loss: 1.1931 - val_accuracy: 0.3947\n",
            "Epoch 104 / 1000 - loss: 0.9972 - accuracy: 0.5657 - val_loss: 1.1922 - val_accuracy: 0.3947\n",
            "Epoch 105 / 1000 - loss: 0.9955 - accuracy: 0.5657 - val_loss: 1.1903 - val_accuracy: 0.3947\n",
            "Epoch 106 / 1000 - loss: 0.9939 - accuracy: 0.5657 - val_loss: 1.1891 - val_accuracy: 0.3947\n",
            "Epoch 107 / 1000 - loss: 0.9923 - accuracy: 0.5657 - val_loss: 1.1874 - val_accuracy: 0.3947\n",
            "Epoch 108 / 1000 - loss: 0.9906 - accuracy: 0.5657 - val_loss: 1.1861 - val_accuracy: 0.3947\n",
            "Epoch 109 / 1000 - loss: 0.9890 - accuracy: 0.5657 - val_loss: 1.1845 - val_accuracy: 0.3947\n",
            "Epoch 110 / 1000 - loss: 0.9873 - accuracy: 0.5657 - val_loss: 1.1833 - val_accuracy: 0.3947\n",
            "Epoch 111 / 1000 - loss: 0.9859 - accuracy: 0.5657 - val_loss: 1.1810 - val_accuracy: 0.3947\n",
            "Epoch 112 / 1000 - loss: 0.9842 - accuracy: 0.5657 - val_loss: 1.1799 - val_accuracy: 0.3947\n",
            "Epoch 113 / 1000 - loss: 0.9828 - accuracy: 0.5657 - val_loss: 1.1783 - val_accuracy: 0.3947\n",
            "Epoch 114 / 1000 - loss: 0.9811 - accuracy: 0.5657 - val_loss: 1.1772 - val_accuracy: 0.3947\n",
            "Epoch 115 / 1000 - loss: 0.9796 - accuracy: 0.5657 - val_loss: 1.1766 - val_accuracy: 0.3947\n",
            "Epoch 116 / 1000 - loss: 0.9780 - accuracy: 0.5657 - val_loss: 1.1750 - val_accuracy: 0.3947\n",
            "Epoch 117 / 1000 - loss: 0.9766 - accuracy: 0.5657 - val_loss: 1.1734 - val_accuracy: 0.3947\n",
            "Epoch 118 / 1000 - loss: 0.9750 - accuracy: 0.5657 - val_loss: 1.1722 - val_accuracy: 0.3947\n",
            "Epoch 119 / 1000 - loss: 0.9736 - accuracy: 0.5657 - val_loss: 1.1704 - val_accuracy: 0.3947\n",
            "Epoch 120 / 1000 - loss: 0.9722 - accuracy: 0.5657 - val_loss: 1.1691 - val_accuracy: 0.3947\n",
            "Epoch 121 / 1000 - loss: 0.9707 - accuracy: 0.5657 - val_loss: 1.1676 - val_accuracy: 0.3947\n",
            "Epoch 122 / 1000 - loss: 0.9693 - accuracy: 0.5657 - val_loss: 1.1655 - val_accuracy: 0.3947\n",
            "Epoch 123 / 1000 - loss: 0.9678 - accuracy: 0.5657 - val_loss: 1.1648 - val_accuracy: 0.3947\n",
            "Epoch 124 / 1000 - loss: 0.9663 - accuracy: 0.5657 - val_loss: 1.1635 - val_accuracy: 0.3947\n",
            "Epoch 125 / 1000 - loss: 0.9650 - accuracy: 0.5657 - val_loss: 1.1623 - val_accuracy: 0.3947\n",
            "Epoch 126 / 1000 - loss: 0.9635 - accuracy: 0.5657 - val_loss: 1.1609 - val_accuracy: 0.3947\n",
            "Epoch 127 / 1000 - loss: 0.9621 - accuracy: 0.5657 - val_loss: 1.1600 - val_accuracy: 0.3947\n",
            "Epoch 128 / 1000 - loss: 0.9607 - accuracy: 0.5690 - val_loss: 1.1587 - val_accuracy: 0.3947\n",
            "Epoch 129 / 1000 - loss: 0.9594 - accuracy: 0.5724 - val_loss: 1.1572 - val_accuracy: 0.3947\n",
            "Epoch 130 / 1000 - loss: 0.9580 - accuracy: 0.5758 - val_loss: 1.1559 - val_accuracy: 0.3947\n",
            "Epoch 131 / 1000 - loss: 0.9567 - accuracy: 0.5825 - val_loss: 1.1547 - val_accuracy: 0.3947\n",
            "Epoch 132 / 1000 - loss: 0.9553 - accuracy: 0.5825 - val_loss: 1.1530 - val_accuracy: 0.3947\n",
            "Epoch 133 / 1000 - loss: 0.9540 - accuracy: 0.5859 - val_loss: 1.1525 - val_accuracy: 0.3947\n",
            "Epoch 134 / 1000 - loss: 0.9526 - accuracy: 0.5892 - val_loss: 1.1509 - val_accuracy: 0.3947\n",
            "Epoch 135 / 1000 - loss: 0.9514 - accuracy: 0.5892 - val_loss: 1.1497 - val_accuracy: 0.3947\n",
            "Epoch 136 / 1000 - loss: 0.9500 - accuracy: 0.5892 - val_loss: 1.1483 - val_accuracy: 0.3947\n",
            "Epoch 137 / 1000 - loss: 0.9487 - accuracy: 0.5892 - val_loss: 1.1467 - val_accuracy: 0.3947\n",
            "Epoch 138 / 1000 - loss: 0.9473 - accuracy: 0.5892 - val_loss: 1.1458 - val_accuracy: 0.3947\n",
            "Epoch 139 / 1000 - loss: 0.9461 - accuracy: 0.5926 - val_loss: 1.1446 - val_accuracy: 0.3947\n",
            "Epoch 140 / 1000 - loss: 0.9447 - accuracy: 0.5926 - val_loss: 1.1434 - val_accuracy: 0.3947\n",
            "Epoch 141 / 1000 - loss: 0.9435 - accuracy: 0.5926 - val_loss: 1.1417 - val_accuracy: 0.3684\n",
            "Epoch 142 / 1000 - loss: 0.9422 - accuracy: 0.5892 - val_loss: 1.1405 - val_accuracy: 0.3684\n",
            "Epoch 143 / 1000 - loss: 0.9409 - accuracy: 0.5859 - val_loss: 1.1392 - val_accuracy: 0.3684\n",
            "Epoch 144 / 1000 - loss: 0.9398 - accuracy: 0.5859 - val_loss: 1.1381 - val_accuracy: 0.3684\n",
            "Epoch 145 / 1000 - loss: 0.9387 - accuracy: 0.5859 - val_loss: 1.1369 - val_accuracy: 0.3684\n",
            "Epoch 146 / 1000 - loss: 0.9376 - accuracy: 0.5859 - val_loss: 1.1359 - val_accuracy: 0.3684\n",
            "Epoch 147 / 1000 - loss: 0.9361 - accuracy: 0.5859 - val_loss: 1.1350 - val_accuracy: 0.3684\n",
            "Epoch 148 / 1000 - loss: 0.9352 - accuracy: 0.5926 - val_loss: 1.1344 - val_accuracy: 0.3684\n",
            "Epoch 149 / 1000 - loss: 0.9338 - accuracy: 0.5892 - val_loss: 1.1327 - val_accuracy: 0.3684\n",
            "Epoch 150 / 1000 - loss: 0.9327 - accuracy: 0.5926 - val_loss: 1.1323 - val_accuracy: 0.3947\n",
            "Epoch 151 / 1000 - loss: 0.9317 - accuracy: 0.5926 - val_loss: 1.1309 - val_accuracy: 0.4474\n",
            "Epoch 152 / 1000 - loss: 0.9304 - accuracy: 0.5926 - val_loss: 1.1298 - val_accuracy: 0.4474\n",
            "Epoch 153 / 1000 - loss: 0.9292 - accuracy: 0.5926 - val_loss: 1.1280 - val_accuracy: 0.4737\n",
            "Epoch 154 / 1000 - loss: 0.9282 - accuracy: 0.5926 - val_loss: 1.1271 - val_accuracy: 0.4737\n",
            "Epoch 155 / 1000 - loss: 0.9272 - accuracy: 0.5926 - val_loss: 1.1256 - val_accuracy: 0.4737\n",
            "Epoch 156 / 1000 - loss: 0.9258 - accuracy: 0.5993 - val_loss: 1.1255 - val_accuracy: 0.4737\n",
            "Epoch 157 / 1000 - loss: 0.9251 - accuracy: 0.6027 - val_loss: 1.1238 - val_accuracy: 0.4737\n",
            "Epoch 158 / 1000 - loss: 0.9237 - accuracy: 0.6027 - val_loss: 1.1232 - val_accuracy: 0.4737\n",
            "Epoch 159 / 1000 - loss: 0.9226 - accuracy: 0.6061 - val_loss: 1.1222 - val_accuracy: 0.4737\n",
            "Epoch 160 / 1000 - loss: 0.9215 - accuracy: 0.6027 - val_loss: 1.1207 - val_accuracy: 0.4474\n",
            "Epoch 161 / 1000 - loss: 0.9206 - accuracy: 0.6027 - val_loss: 1.1193 - val_accuracy: 0.4474\n",
            "Epoch 162 / 1000 - loss: 0.9196 - accuracy: 0.6061 - val_loss: 1.1182 - val_accuracy: 0.4474\n",
            "Epoch 163 / 1000 - loss: 0.9184 - accuracy: 0.6094 - val_loss: 1.1175 - val_accuracy: 0.4474\n",
            "Epoch 164 / 1000 - loss: 0.9172 - accuracy: 0.6128 - val_loss: 1.1172 - val_accuracy: 0.4474\n",
            "Epoch 165 / 1000 - loss: 0.9162 - accuracy: 0.6094 - val_loss: 1.1164 - val_accuracy: 0.4474\n",
            "Epoch 166 / 1000 - loss: 0.9151 - accuracy: 0.6094 - val_loss: 1.1159 - val_accuracy: 0.4474\n",
            "Epoch 167 / 1000 - loss: 0.9142 - accuracy: 0.6061 - val_loss: 1.1147 - val_accuracy: 0.4474\n",
            "Epoch 168 / 1000 - loss: 0.9131 - accuracy: 0.6094 - val_loss: 1.1140 - val_accuracy: 0.4474\n",
            "Epoch 169 / 1000 - loss: 0.9125 - accuracy: 0.6061 - val_loss: 1.1123 - val_accuracy: 0.4474\n",
            "Epoch 170 / 1000 - loss: 0.9112 - accuracy: 0.6061 - val_loss: 1.1117 - val_accuracy: 0.4474\n",
            "Epoch 171 / 1000 - loss: 0.9103 - accuracy: 0.6094 - val_loss: 1.1105 - val_accuracy: 0.4474\n",
            "Epoch 172 / 1000 - loss: 0.9091 - accuracy: 0.6094 - val_loss: 1.1096 - val_accuracy: 0.4474\n",
            "Epoch 173 / 1000 - loss: 0.9081 - accuracy: 0.6094 - val_loss: 1.1087 - val_accuracy: 0.4474\n",
            "Epoch 174 / 1000 - loss: 0.9075 - accuracy: 0.6094 - val_loss: 1.1079 - val_accuracy: 0.4474\n",
            "Epoch 175 / 1000 - loss: 0.9062 - accuracy: 0.6094 - val_loss: 1.1068 - val_accuracy: 0.4474\n",
            "Epoch 176 / 1000 - loss: 0.9052 - accuracy: 0.6094 - val_loss: 1.1060 - val_accuracy: 0.4474\n",
            "Epoch 177 / 1000 - loss: 0.9042 - accuracy: 0.6094 - val_loss: 1.1052 - val_accuracy: 0.4474\n",
            "Epoch 178 / 1000 - loss: 0.9033 - accuracy: 0.6094 - val_loss: 1.1046 - val_accuracy: 0.4474\n",
            "Epoch 179 / 1000 - loss: 0.9025 - accuracy: 0.6229 - val_loss: 1.1031 - val_accuracy: 0.4474\n",
            "Epoch 180 / 1000 - loss: 0.9015 - accuracy: 0.6195 - val_loss: 1.1022 - val_accuracy: 0.4474\n",
            "Epoch 181 / 1000 - loss: 0.9003 - accuracy: 0.6229 - val_loss: 1.1017 - val_accuracy: 0.4474\n",
            "Epoch 182 / 1000 - loss: 0.8994 - accuracy: 0.6229 - val_loss: 1.1010 - val_accuracy: 0.4474\n",
            "Epoch 183 / 1000 - loss: 0.8986 - accuracy: 0.6195 - val_loss: 1.1001 - val_accuracy: 0.4474\n",
            "Epoch 184 / 1000 - loss: 0.8976 - accuracy: 0.6229 - val_loss: 1.0985 - val_accuracy: 0.4474\n",
            "Epoch 185 / 1000 - loss: 0.8966 - accuracy: 0.6229 - val_loss: 1.0981 - val_accuracy: 0.4737\n",
            "Epoch 186 / 1000 - loss: 0.8958 - accuracy: 0.6229 - val_loss: 1.0970 - val_accuracy: 0.4474\n",
            "Epoch 187 / 1000 - loss: 0.8948 - accuracy: 0.6229 - val_loss: 1.0957 - val_accuracy: 0.4737\n",
            "Epoch 188 / 1000 - loss: 0.8937 - accuracy: 0.6229 - val_loss: 1.0954 - val_accuracy: 0.4737\n",
            "Epoch 189 / 1000 - loss: 0.8928 - accuracy: 0.6229 - val_loss: 1.0943 - val_accuracy: 0.4737\n",
            "Epoch 190 / 1000 - loss: 0.8921 - accuracy: 0.6229 - val_loss: 1.0928 - val_accuracy: 0.4737\n",
            "Epoch 191 / 1000 - loss: 0.8910 - accuracy: 0.6229 - val_loss: 1.0927 - val_accuracy: 0.4737\n",
            "Epoch 192 / 1000 - loss: 0.8900 - accuracy: 0.6229 - val_loss: 1.0922 - val_accuracy: 0.4737\n",
            "Epoch 193 / 1000 - loss: 0.8892 - accuracy: 0.6229 - val_loss: 1.0908 - val_accuracy: 0.4737\n",
            "Epoch 194 / 1000 - loss: 0.8883 - accuracy: 0.6229 - val_loss: 1.0898 - val_accuracy: 0.4737\n",
            "Epoch 195 / 1000 - loss: 0.8875 - accuracy: 0.6229 - val_loss: 1.0887 - val_accuracy: 0.4737\n",
            "Epoch 196 / 1000 - loss: 0.8867 - accuracy: 0.6229 - val_loss: 1.0883 - val_accuracy: 0.4737\n",
            "Epoch 197 / 1000 - loss: 0.8858 - accuracy: 0.6229 - val_loss: 1.0872 - val_accuracy: 0.4737\n",
            "Epoch 198 / 1000 - loss: 0.8850 - accuracy: 0.6229 - val_loss: 1.0873 - val_accuracy: 0.5000\n",
            "Epoch 199 / 1000 - loss: 0.8842 - accuracy: 0.6364 - val_loss: 1.0864 - val_accuracy: 0.5000\n",
            "Epoch 200 / 1000 - loss: 0.8835 - accuracy: 0.6364 - val_loss: 1.0845 - val_accuracy: 0.5000\n",
            "Epoch 201 / 1000 - loss: 0.8825 - accuracy: 0.6229 - val_loss: 1.0847 - val_accuracy: 0.5000\n",
            "Epoch 202 / 1000 - loss: 0.8816 - accuracy: 0.6364 - val_loss: 1.0841 - val_accuracy: 0.5000\n",
            "Epoch 203 / 1000 - loss: 0.8808 - accuracy: 0.6364 - val_loss: 1.0825 - val_accuracy: 0.5000\n",
            "Epoch 204 / 1000 - loss: 0.8800 - accuracy: 0.6364 - val_loss: 1.0818 - val_accuracy: 0.5000\n",
            "Epoch 205 / 1000 - loss: 0.8791 - accuracy: 0.6364 - val_loss: 1.0822 - val_accuracy: 0.5000\n",
            "Epoch 206 / 1000 - loss: 0.8785 - accuracy: 0.6498 - val_loss: 1.0807 - val_accuracy: 0.5000\n",
            "Epoch 207 / 1000 - loss: 0.8777 - accuracy: 0.6364 - val_loss: 1.0797 - val_accuracy: 0.5000\n",
            "Epoch 208 / 1000 - loss: 0.8769 - accuracy: 0.6364 - val_loss: 1.0790 - val_accuracy: 0.5000\n",
            "Epoch 209 / 1000 - loss: 0.8760 - accuracy: 0.6364 - val_loss: 1.0780 - val_accuracy: 0.5000\n",
            "Epoch 210 / 1000 - loss: 0.8751 - accuracy: 0.6498 - val_loss: 1.0783 - val_accuracy: 0.5000\n",
            "Epoch 211 / 1000 - loss: 0.8745 - accuracy: 0.6498 - val_loss: 1.0768 - val_accuracy: 0.5000\n",
            "Epoch 212 / 1000 - loss: 0.8736 - accuracy: 0.6498 - val_loss: 1.0764 - val_accuracy: 0.5000\n",
            "Epoch 213 / 1000 - loss: 0.8728 - accuracy: 0.6465 - val_loss: 1.0755 - val_accuracy: 0.5000\n",
            "Epoch 214 / 1000 - loss: 0.8720 - accuracy: 0.6498 - val_loss: 1.0748 - val_accuracy: 0.5000\n",
            "Epoch 215 / 1000 - loss: 0.8713 - accuracy: 0.6498 - val_loss: 1.0741 - val_accuracy: 0.5000\n",
            "Epoch 216 / 1000 - loss: 0.8704 - accuracy: 0.6431 - val_loss: 1.0733 - val_accuracy: 0.5000\n",
            "Epoch 217 / 1000 - loss: 0.8697 - accuracy: 0.6498 - val_loss: 1.0726 - val_accuracy: 0.5000\n",
            "Epoch 218 / 1000 - loss: 0.8693 - accuracy: 0.6498 - val_loss: 1.0706 - val_accuracy: 0.5000\n",
            "Epoch 219 / 1000 - loss: 0.8682 - accuracy: 0.6498 - val_loss: 1.0709 - val_accuracy: 0.5000\n",
            "Epoch 220 / 1000 - loss: 0.8675 - accuracy: 0.6397 - val_loss: 1.0693 - val_accuracy: 0.5000\n",
            "Epoch 221 / 1000 - loss: 0.8666 - accuracy: 0.6465 - val_loss: 1.0689 - val_accuracy: 0.5000\n",
            "Epoch 222 / 1000 - loss: 0.8658 - accuracy: 0.6397 - val_loss: 1.0686 - val_accuracy: 0.5000\n",
            "Epoch 223 / 1000 - loss: 0.8652 - accuracy: 0.6397 - val_loss: 1.0669 - val_accuracy: 0.5000\n",
            "Epoch 224 / 1000 - loss: 0.8643 - accuracy: 0.6397 - val_loss: 1.0673 - val_accuracy: 0.5000\n",
            "Epoch 225 / 1000 - loss: 0.8636 - accuracy: 0.6397 - val_loss: 1.0665 - val_accuracy: 0.5000\n",
            "Epoch 226 / 1000 - loss: 0.8629 - accuracy: 0.6532 - val_loss: 1.0657 - val_accuracy: 0.5000\n",
            "Epoch 227 / 1000 - loss: 0.8624 - accuracy: 0.6532 - val_loss: 1.0640 - val_accuracy: 0.5000\n",
            "Epoch 228 / 1000 - loss: 0.8612 - accuracy: 0.6532 - val_loss: 1.0644 - val_accuracy: 0.5000\n",
            "Epoch 229 / 1000 - loss: 0.8606 - accuracy: 0.6532 - val_loss: 1.0639 - val_accuracy: 0.5000\n",
            "Epoch 230 / 1000 - loss: 0.8599 - accuracy: 0.6532 - val_loss: 1.0624 - val_accuracy: 0.5000\n",
            "Epoch 231 / 1000 - loss: 0.8591 - accuracy: 0.6532 - val_loss: 1.0624 - val_accuracy: 0.5000\n",
            "Epoch 232 / 1000 - loss: 0.8583 - accuracy: 0.6532 - val_loss: 1.0617 - val_accuracy: 0.5000\n",
            "Epoch 233 / 1000 - loss: 0.8575 - accuracy: 0.6532 - val_loss: 1.0610 - val_accuracy: 0.5000\n",
            "Epoch 234 / 1000 - loss: 0.8568 - accuracy: 0.6667 - val_loss: 1.0602 - val_accuracy: 0.5000\n",
            "Epoch 235 / 1000 - loss: 0.8561 - accuracy: 0.6667 - val_loss: 1.0586 - val_accuracy: 0.5000\n",
            "Epoch 236 / 1000 - loss: 0.8554 - accuracy: 0.6667 - val_loss: 1.0583 - val_accuracy: 0.5000\n",
            "Epoch 237 / 1000 - loss: 0.8546 - accuracy: 0.6667 - val_loss: 1.0580 - val_accuracy: 0.5000\n",
            "Epoch 238 / 1000 - loss: 0.8540 - accuracy: 0.6667 - val_loss: 1.0574 - val_accuracy: 0.5000\n",
            "Epoch 239 / 1000 - loss: 0.8532 - accuracy: 0.6667 - val_loss: 1.0567 - val_accuracy: 0.5000\n",
            "Epoch 240 / 1000 - loss: 0.8525 - accuracy: 0.6667 - val_loss: 1.0557 - val_accuracy: 0.5000\n",
            "Epoch 241 / 1000 - loss: 0.8518 - accuracy: 0.6667 - val_loss: 1.0551 - val_accuracy: 0.5000\n",
            "Epoch 242 / 1000 - loss: 0.8510 - accuracy: 0.6667 - val_loss: 1.0551 - val_accuracy: 0.5000\n",
            "Epoch 243 / 1000 - loss: 0.8504 - accuracy: 0.6667 - val_loss: 1.0541 - val_accuracy: 0.5000\n",
            "Epoch 244 / 1000 - loss: 0.8496 - accuracy: 0.6667 - val_loss: 1.0531 - val_accuracy: 0.5000\n",
            "Epoch 245 / 1000 - loss: 0.8490 - accuracy: 0.6667 - val_loss: 1.0517 - val_accuracy: 0.5000\n",
            "Epoch 246 / 1000 - loss: 0.8483 - accuracy: 0.6667 - val_loss: 1.0511 - val_accuracy: 0.5000\n",
            "Epoch 247 / 1000 - loss: 0.8474 - accuracy: 0.6667 - val_loss: 1.0507 - val_accuracy: 0.5000\n",
            "Epoch 248 / 1000 - loss: 0.8466 - accuracy: 0.6667 - val_loss: 1.0499 - val_accuracy: 0.5000\n",
            "Epoch 249 / 1000 - loss: 0.8460 - accuracy: 0.6667 - val_loss: 1.0485 - val_accuracy: 0.5000\n",
            "Epoch 250 / 1000 - loss: 0.8456 - accuracy: 0.6667 - val_loss: 1.0478 - val_accuracy: 0.5000\n",
            "Epoch 251 / 1000 - loss: 0.8449 - accuracy: 0.6667 - val_loss: 1.0472 - val_accuracy: 0.5000\n",
            "Epoch 252 / 1000 - loss: 0.8439 - accuracy: 0.6667 - val_loss: 1.0470 - val_accuracy: 0.5000\n",
            "Epoch 253 / 1000 - loss: 0.8433 - accuracy: 0.6667 - val_loss: 1.0459 - val_accuracy: 0.5000\n",
            "Epoch 254 / 1000 - loss: 0.8425 - accuracy: 0.6667 - val_loss: 1.0455 - val_accuracy: 0.5000\n",
            "Epoch 255 / 1000 - loss: 0.8420 - accuracy: 0.6667 - val_loss: 1.0448 - val_accuracy: 0.5000\n",
            "Epoch 256 / 1000 - loss: 0.8415 - accuracy: 0.6667 - val_loss: 1.0431 - val_accuracy: 0.5000\n",
            "Epoch 257 / 1000 - loss: 0.8405 - accuracy: 0.6667 - val_loss: 1.0429 - val_accuracy: 0.5000\n",
            "Epoch 258 / 1000 - loss: 0.8398 - accuracy: 0.6667 - val_loss: 1.0425 - val_accuracy: 0.5000\n",
            "Epoch 259 / 1000 - loss: 0.8391 - accuracy: 0.6667 - val_loss: 1.0415 - val_accuracy: 0.5000\n",
            "Epoch 260 / 1000 - loss: 0.8385 - accuracy: 0.6667 - val_loss: 1.0407 - val_accuracy: 0.5000\n",
            "Epoch 261 / 1000 - loss: 0.8377 - accuracy: 0.6667 - val_loss: 1.0404 - val_accuracy: 0.5000\n",
            "Epoch 262 / 1000 - loss: 0.8370 - accuracy: 0.6667 - val_loss: 1.0397 - val_accuracy: 0.5000\n",
            "Epoch 263 / 1000 - loss: 0.8364 - accuracy: 0.6667 - val_loss: 1.0387 - val_accuracy: 0.5000\n",
            "Epoch 264 / 1000 - loss: 0.8357 - accuracy: 0.6667 - val_loss: 1.0385 - val_accuracy: 0.5000\n",
            "Epoch 265 / 1000 - loss: 0.8351 - accuracy: 0.6667 - val_loss: 1.0374 - val_accuracy: 0.5000\n",
            "Epoch 266 / 1000 - loss: 0.8343 - accuracy: 0.6633 - val_loss: 1.0368 - val_accuracy: 0.5000\n",
            "Epoch 267 / 1000 - loss: 0.8336 - accuracy: 0.6633 - val_loss: 1.0368 - val_accuracy: 0.5000\n",
            "Epoch 268 / 1000 - loss: 0.8329 - accuracy: 0.6633 - val_loss: 1.0360 - val_accuracy: 0.5000\n",
            "Epoch 269 / 1000 - loss: 0.8323 - accuracy: 0.6667 - val_loss: 1.0350 - val_accuracy: 0.5000\n",
            "Epoch 270 / 1000 - loss: 0.8317 - accuracy: 0.6667 - val_loss: 1.0343 - val_accuracy: 0.5000\n",
            "Epoch 271 / 1000 - loss: 0.8310 - accuracy: 0.6667 - val_loss: 1.0334 - val_accuracy: 0.5000\n",
            "Epoch 272 / 1000 - loss: 0.8302 - accuracy: 0.6667 - val_loss: 1.0334 - val_accuracy: 0.5000\n",
            "Epoch 273 / 1000 - loss: 0.8296 - accuracy: 0.6667 - val_loss: 1.0326 - val_accuracy: 0.5000\n",
            "Epoch 274 / 1000 - loss: 0.8289 - accuracy: 0.6667 - val_loss: 1.0322 - val_accuracy: 0.5000\n",
            "Epoch 275 / 1000 - loss: 0.8282 - accuracy: 0.6667 - val_loss: 1.0324 - val_accuracy: 0.5000\n",
            "Epoch 276 / 1000 - loss: 0.8277 - accuracy: 0.6667 - val_loss: 1.0315 - val_accuracy: 0.5000\n",
            "Epoch 277 / 1000 - loss: 0.8270 - accuracy: 0.6667 - val_loss: 1.0306 - val_accuracy: 0.5000\n",
            "Epoch 278 / 1000 - loss: 0.8264 - accuracy: 0.6667 - val_loss: 1.0298 - val_accuracy: 0.5000\n",
            "Epoch 279 / 1000 - loss: 0.8259 - accuracy: 0.6667 - val_loss: 1.0282 - val_accuracy: 0.5000\n",
            "Epoch 280 / 1000 - loss: 0.8252 - accuracy: 0.6700 - val_loss: 1.0285 - val_accuracy: 0.5000\n",
            "Epoch 281 / 1000 - loss: 0.8245 - accuracy: 0.6835 - val_loss: 1.0281 - val_accuracy: 0.5000\n",
            "Epoch 282 / 1000 - loss: 0.8238 - accuracy: 0.6700 - val_loss: 1.0273 - val_accuracy: 0.5000\n",
            "Epoch 283 / 1000 - loss: 0.8231 - accuracy: 0.6835 - val_loss: 1.0264 - val_accuracy: 0.5000\n",
            "Epoch 284 / 1000 - loss: 0.8226 - accuracy: 0.6835 - val_loss: 1.0255 - val_accuracy: 0.5000\n",
            "Epoch 285 / 1000 - loss: 0.8218 - accuracy: 0.6835 - val_loss: 1.0247 - val_accuracy: 0.5000\n",
            "Epoch 286 / 1000 - loss: 0.8211 - accuracy: 0.6835 - val_loss: 1.0239 - val_accuracy: 0.5000\n",
            "Epoch 287 / 1000 - loss: 0.8205 - accuracy: 0.6835 - val_loss: 1.0231 - val_accuracy: 0.5000\n",
            "Epoch 288 / 1000 - loss: 0.8198 - accuracy: 0.6835 - val_loss: 1.0220 - val_accuracy: 0.5000\n",
            "Epoch 289 / 1000 - loss: 0.8191 - accuracy: 0.6835 - val_loss: 1.0219 - val_accuracy: 0.5000\n",
            "Epoch 290 / 1000 - loss: 0.8185 - accuracy: 0.6835 - val_loss: 1.0213 - val_accuracy: 0.5000\n",
            "Epoch 291 / 1000 - loss: 0.8179 - accuracy: 0.6835 - val_loss: 1.0212 - val_accuracy: 0.5000\n",
            "Epoch 292 / 1000 - loss: 0.8172 - accuracy: 0.6835 - val_loss: 1.0203 - val_accuracy: 0.5000\n",
            "Epoch 293 / 1000 - loss: 0.8166 - accuracy: 0.6835 - val_loss: 1.0197 - val_accuracy: 0.5000\n",
            "Epoch 294 / 1000 - loss: 0.8160 - accuracy: 0.6835 - val_loss: 1.0191 - val_accuracy: 0.5000\n",
            "Epoch 295 / 1000 - loss: 0.8153 - accuracy: 0.6835 - val_loss: 1.0194 - val_accuracy: 0.5000\n",
            "Epoch 296 / 1000 - loss: 0.8148 - accuracy: 0.6835 - val_loss: 1.0176 - val_accuracy: 0.5000\n",
            "Epoch 297 / 1000 - loss: 0.8141 - accuracy: 0.6835 - val_loss: 1.0179 - val_accuracy: 0.5000\n",
            "Epoch 298 / 1000 - loss: 0.8134 - accuracy: 0.6835 - val_loss: 1.0165 - val_accuracy: 0.5000\n",
            "Epoch 299 / 1000 - loss: 0.8127 - accuracy: 0.6835 - val_loss: 1.0159 - val_accuracy: 0.5000\n",
            "Epoch 300 / 1000 - loss: 0.8121 - accuracy: 0.6835 - val_loss: 1.0150 - val_accuracy: 0.5000\n",
            "Epoch 301 / 1000 - loss: 0.8114 - accuracy: 0.6835 - val_loss: 1.0146 - val_accuracy: 0.5000\n",
            "Epoch 302 / 1000 - loss: 0.8110 - accuracy: 0.6835 - val_loss: 1.0135 - val_accuracy: 0.5000\n",
            "Epoch 303 / 1000 - loss: 0.8103 - accuracy: 0.6835 - val_loss: 1.0136 - val_accuracy: 0.5000\n",
            "Epoch 304 / 1000 - loss: 0.8096 - accuracy: 0.6835 - val_loss: 1.0130 - val_accuracy: 0.5000\n",
            "Epoch 305 / 1000 - loss: 0.8089 - accuracy: 0.6835 - val_loss: 1.0123 - val_accuracy: 0.5000\n",
            "Epoch 306 / 1000 - loss: 0.8085 - accuracy: 0.6835 - val_loss: 1.0104 - val_accuracy: 0.5000\n",
            "Epoch 307 / 1000 - loss: 0.8078 - accuracy: 0.6835 - val_loss: 1.0101 - val_accuracy: 0.5000\n",
            "Epoch 308 / 1000 - loss: 0.8071 - accuracy: 0.6835 - val_loss: 1.0102 - val_accuracy: 0.5000\n",
            "Epoch 309 / 1000 - loss: 0.8064 - accuracy: 0.6835 - val_loss: 1.0094 - val_accuracy: 0.5000\n",
            "Epoch 310 / 1000 - loss: 0.8060 - accuracy: 0.6835 - val_loss: 1.0089 - val_accuracy: 0.5000\n",
            "Epoch 311 / 1000 - loss: 0.8052 - accuracy: 0.6835 - val_loss: 1.0081 - val_accuracy: 0.5000\n",
            "Epoch 312 / 1000 - loss: 0.8046 - accuracy: 0.6835 - val_loss: 1.0078 - val_accuracy: 0.5000\n",
            "Epoch 313 / 1000 - loss: 0.8040 - accuracy: 0.6835 - val_loss: 1.0072 - val_accuracy: 0.5000\n",
            "Epoch 314 / 1000 - loss: 0.8034 - accuracy: 0.6835 - val_loss: 1.0062 - val_accuracy: 0.5000\n",
            "Epoch 315 / 1000 - loss: 0.8026 - accuracy: 0.6835 - val_loss: 1.0058 - val_accuracy: 0.5000\n",
            "Epoch 316 / 1000 - loss: 0.8021 - accuracy: 0.6835 - val_loss: 1.0055 - val_accuracy: 0.5000\n",
            "Epoch 317 / 1000 - loss: 0.8016 - accuracy: 0.6869 - val_loss: 1.0043 - val_accuracy: 0.5000\n",
            "Epoch 318 / 1000 - loss: 0.8007 - accuracy: 0.6835 - val_loss: 1.0043 - val_accuracy: 0.5000\n",
            "Epoch 319 / 1000 - loss: 0.8004 - accuracy: 0.6835 - val_loss: 1.0029 - val_accuracy: 0.5000\n",
            "Epoch 320 / 1000 - loss: 0.7996 - accuracy: 0.6869 - val_loss: 1.0023 - val_accuracy: 0.5000\n",
            "Epoch 321 / 1000 - loss: 0.7992 - accuracy: 0.6869 - val_loss: 1.0021 - val_accuracy: 0.5000\n",
            "Epoch 322 / 1000 - loss: 0.7985 - accuracy: 0.6835 - val_loss: 1.0018 - val_accuracy: 0.5000\n",
            "Epoch 323 / 1000 - loss: 0.7978 - accuracy: 0.6835 - val_loss: 1.0011 - val_accuracy: 0.5000\n",
            "Epoch 324 / 1000 - loss: 0.7973 - accuracy: 0.6869 - val_loss: 0.9992 - val_accuracy: 0.5000\n",
            "Epoch 325 / 1000 - loss: 0.7965 - accuracy: 0.6869 - val_loss: 0.9993 - val_accuracy: 0.5000\n",
            "Epoch 326 / 1000 - loss: 0.7959 - accuracy: 0.6869 - val_loss: 0.9994 - val_accuracy: 0.5000\n",
            "Epoch 327 / 1000 - loss: 0.7955 - accuracy: 0.6869 - val_loss: 0.9981 - val_accuracy: 0.5000\n",
            "Epoch 328 / 1000 - loss: 0.7947 - accuracy: 0.6869 - val_loss: 0.9972 - val_accuracy: 0.5000\n",
            "Epoch 329 / 1000 - loss: 0.7944 - accuracy: 0.6869 - val_loss: 0.9959 - val_accuracy: 0.5000\n",
            "Epoch 330 / 1000 - loss: 0.7935 - accuracy: 0.6869 - val_loss: 0.9957 - val_accuracy: 0.5000\n",
            "Epoch 331 / 1000 - loss: 0.7928 - accuracy: 0.6869 - val_loss: 0.9949 - val_accuracy: 0.5000\n",
            "Epoch 332 / 1000 - loss: 0.7925 - accuracy: 0.6869 - val_loss: 0.9952 - val_accuracy: 0.5000\n",
            "Epoch 333 / 1000 - loss: 0.7918 - accuracy: 0.6869 - val_loss: 0.9937 - val_accuracy: 0.5000\n",
            "Epoch 334 / 1000 - loss: 0.7910 - accuracy: 0.6869 - val_loss: 0.9931 - val_accuracy: 0.5000\n",
            "Epoch 335 / 1000 - loss: 0.7904 - accuracy: 0.6869 - val_loss: 0.9922 - val_accuracy: 0.5000\n",
            "Epoch 336 / 1000 - loss: 0.7898 - accuracy: 0.6869 - val_loss: 0.9921 - val_accuracy: 0.5000\n",
            "Epoch 337 / 1000 - loss: 0.7892 - accuracy: 0.6869 - val_loss: 0.9914 - val_accuracy: 0.5000\n",
            "Epoch 338 / 1000 - loss: 0.7887 - accuracy: 0.7003 - val_loss: 0.9903 - val_accuracy: 0.5000\n",
            "Epoch 339 / 1000 - loss: 0.7881 - accuracy: 0.6869 - val_loss: 0.9892 - val_accuracy: 0.5000\n",
            "Epoch 340 / 1000 - loss: 0.7874 - accuracy: 0.7003 - val_loss: 0.9897 - val_accuracy: 0.5000\n",
            "Epoch 341 / 1000 - loss: 0.7867 - accuracy: 0.7003 - val_loss: 0.9892 - val_accuracy: 0.5000\n",
            "Epoch 342 / 1000 - loss: 0.7863 - accuracy: 0.7003 - val_loss: 0.9883 - val_accuracy: 0.5263\n",
            "Epoch 343 / 1000 - loss: 0.7856 - accuracy: 0.7003 - val_loss: 0.9873 - val_accuracy: 0.5000\n",
            "Epoch 344 / 1000 - loss: 0.7854 - accuracy: 0.7003 - val_loss: 0.9855 - val_accuracy: 0.5263\n",
            "Epoch 345 / 1000 - loss: 0.7843 - accuracy: 0.7003 - val_loss: 0.9855 - val_accuracy: 0.5526\n",
            "Epoch 346 / 1000 - loss: 0.7837 - accuracy: 0.7003 - val_loss: 0.9849 - val_accuracy: 0.5526\n",
            "Epoch 347 / 1000 - loss: 0.7832 - accuracy: 0.7003 - val_loss: 0.9850 - val_accuracy: 0.5526\n",
            "Epoch 348 / 1000 - loss: 0.7825 - accuracy: 0.7003 - val_loss: 0.9844 - val_accuracy: 0.5526\n",
            "Epoch 349 / 1000 - loss: 0.7818 - accuracy: 0.7003 - val_loss: 0.9838 - val_accuracy: 0.5526\n",
            "Epoch 350 / 1000 - loss: 0.7814 - accuracy: 0.7003 - val_loss: 0.9830 - val_accuracy: 0.5526\n",
            "Epoch 351 / 1000 - loss: 0.7807 - accuracy: 0.7003 - val_loss: 0.9820 - val_accuracy: 0.5526\n",
            "Epoch 352 / 1000 - loss: 0.7807 - accuracy: 0.7003 - val_loss: 0.9797 - val_accuracy: 0.5526\n",
            "Epoch 353 / 1000 - loss: 0.7795 - accuracy: 0.7003 - val_loss: 0.9813 - val_accuracy: 0.5526\n",
            "Epoch 354 / 1000 - loss: 0.7790 - accuracy: 0.7003 - val_loss: 0.9801 - val_accuracy: 0.5526\n",
            "Epoch 355 / 1000 - loss: 0.7783 - accuracy: 0.7003 - val_loss: 0.9792 - val_accuracy: 0.5526\n",
            "Epoch 356 / 1000 - loss: 0.7777 - accuracy: 0.7003 - val_loss: 0.9787 - val_accuracy: 0.5526\n",
            "Epoch 357 / 1000 - loss: 0.7772 - accuracy: 0.7003 - val_loss: 0.9781 - val_accuracy: 0.5526\n",
            "Epoch 358 / 1000 - loss: 0.7765 - accuracy: 0.7003 - val_loss: 0.9776 - val_accuracy: 0.5526\n",
            "Epoch 359 / 1000 - loss: 0.7761 - accuracy: 0.7003 - val_loss: 0.9768 - val_accuracy: 0.5526\n",
            "Epoch 360 / 1000 - loss: 0.7755 - accuracy: 0.7003 - val_loss: 0.9761 - val_accuracy: 0.5526\n",
            "Epoch 361 / 1000 - loss: 0.7749 - accuracy: 0.7003 - val_loss: 0.9754 - val_accuracy: 0.5526\n",
            "Epoch 362 / 1000 - loss: 0.7745 - accuracy: 0.7003 - val_loss: 0.9740 - val_accuracy: 0.5526\n",
            "Epoch 363 / 1000 - loss: 0.7735 - accuracy: 0.7003 - val_loss: 0.9744 - val_accuracy: 0.5526\n",
            "Epoch 364 / 1000 - loss: 0.7731 - accuracy: 0.7172 - val_loss: 0.9749 - val_accuracy: 0.5526\n",
            "Epoch 365 / 1000 - loss: 0.7726 - accuracy: 0.7037 - val_loss: 0.9730 - val_accuracy: 0.5526\n",
            "Epoch 366 / 1000 - loss: 0.7723 - accuracy: 0.7138 - val_loss: 0.9726 - val_accuracy: 0.5526\n",
            "Epoch 367 / 1000 - loss: 0.7712 - accuracy: 0.7172 - val_loss: 0.9719 - val_accuracy: 0.5526\n",
            "Epoch 368 / 1000 - loss: 0.7709 - accuracy: 0.7037 - val_loss: 0.9704 - val_accuracy: 0.5526\n",
            "Epoch 369 / 1000 - loss: 0.7699 - accuracy: 0.7071 - val_loss: 0.9713 - val_accuracy: 0.5526\n",
            "Epoch 370 / 1000 - loss: 0.7695 - accuracy: 0.7205 - val_loss: 0.9711 - val_accuracy: 0.5526\n",
            "Epoch 371 / 1000 - loss: 0.7690 - accuracy: 0.7071 - val_loss: 0.9698 - val_accuracy: 0.5526\n",
            "Epoch 372 / 1000 - loss: 0.7682 - accuracy: 0.7071 - val_loss: 0.9694 - val_accuracy: 0.5526\n",
            "Epoch 373 / 1000 - loss: 0.7676 - accuracy: 0.7205 - val_loss: 0.9691 - val_accuracy: 0.5526\n",
            "Epoch 374 / 1000 - loss: 0.7671 - accuracy: 0.7205 - val_loss: 0.9689 - val_accuracy: 0.5526\n",
            "Epoch 375 / 1000 - loss: 0.7665 - accuracy: 0.7205 - val_loss: 0.9679 - val_accuracy: 0.5526\n",
            "Epoch 376 / 1000 - loss: 0.7659 - accuracy: 0.7205 - val_loss: 0.9674 - val_accuracy: 0.5526\n",
            "Epoch 377 / 1000 - loss: 0.7653 - accuracy: 0.7205 - val_loss: 0.9673 - val_accuracy: 0.5526\n",
            "Epoch 378 / 1000 - loss: 0.7649 - accuracy: 0.7205 - val_loss: 0.9667 - val_accuracy: 0.5526\n",
            "Epoch 379 / 1000 - loss: 0.7641 - accuracy: 0.7205 - val_loss: 0.9653 - val_accuracy: 0.5526\n",
            "Epoch 380 / 1000 - loss: 0.7635 - accuracy: 0.7205 - val_loss: 0.9644 - val_accuracy: 0.5526\n",
            "Epoch 381 / 1000 - loss: 0.7629 - accuracy: 0.7205 - val_loss: 0.9635 - val_accuracy: 0.5526\n",
            "Epoch 382 / 1000 - loss: 0.7625 - accuracy: 0.7205 - val_loss: 0.9618 - val_accuracy: 0.5526\n",
            "Epoch 383 / 1000 - loss: 0.7617 - accuracy: 0.7205 - val_loss: 0.9613 - val_accuracy: 0.5526\n",
            "Epoch 384 / 1000 - loss: 0.7612 - accuracy: 0.7205 - val_loss: 0.9603 - val_accuracy: 0.5526\n",
            "Epoch 385 / 1000 - loss: 0.7607 - accuracy: 0.7205 - val_loss: 0.9606 - val_accuracy: 0.5526\n",
            "Epoch 386 / 1000 - loss: 0.7600 - accuracy: 0.7205 - val_loss: 0.9595 - val_accuracy: 0.5526\n",
            "Epoch 387 / 1000 - loss: 0.7597 - accuracy: 0.7205 - val_loss: 0.9583 - val_accuracy: 0.5526\n",
            "Epoch 388 / 1000 - loss: 0.7588 - accuracy: 0.7205 - val_loss: 0.9581 - val_accuracy: 0.5526\n",
            "Epoch 389 / 1000 - loss: 0.7581 - accuracy: 0.7205 - val_loss: 0.9574 - val_accuracy: 0.5526\n",
            "Epoch 390 / 1000 - loss: 0.7576 - accuracy: 0.7205 - val_loss: 0.9565 - val_accuracy: 0.5526\n",
            "Epoch 391 / 1000 - loss: 0.7570 - accuracy: 0.7205 - val_loss: 0.9561 - val_accuracy: 0.5526\n",
            "Epoch 392 / 1000 - loss: 0.7565 - accuracy: 0.7205 - val_loss: 0.9546 - val_accuracy: 0.5526\n",
            "Epoch 393 / 1000 - loss: 0.7560 - accuracy: 0.7205 - val_loss: 0.9541 - val_accuracy: 0.5526\n",
            "Epoch 394 / 1000 - loss: 0.7552 - accuracy: 0.7239 - val_loss: 0.9540 - val_accuracy: 0.5526\n",
            "Epoch 395 / 1000 - loss: 0.7549 - accuracy: 0.7205 - val_loss: 0.9531 - val_accuracy: 0.5526\n",
            "Epoch 396 / 1000 - loss: 0.7540 - accuracy: 0.7205 - val_loss: 0.9527 - val_accuracy: 0.5526\n",
            "Epoch 397 / 1000 - loss: 0.7536 - accuracy: 0.7239 - val_loss: 0.9516 - val_accuracy: 0.5526\n",
            "Epoch 398 / 1000 - loss: 0.7528 - accuracy: 0.7239 - val_loss: 0.9516 - val_accuracy: 0.5526\n",
            "Epoch 399 / 1000 - loss: 0.7526 - accuracy: 0.7239 - val_loss: 0.9497 - val_accuracy: 0.5526\n",
            "Epoch 400 / 1000 - loss: 0.7519 - accuracy: 0.7205 - val_loss: 0.9489 - val_accuracy: 0.5526\n",
            "Epoch 401 / 1000 - loss: 0.7514 - accuracy: 0.7239 - val_loss: 0.9495 - val_accuracy: 0.5526\n",
            "Epoch 402 / 1000 - loss: 0.7505 - accuracy: 0.7239 - val_loss: 0.9483 - val_accuracy: 0.5526\n",
            "Epoch 403 / 1000 - loss: 0.7499 - accuracy: 0.7239 - val_loss: 0.9484 - val_accuracy: 0.5526\n",
            "Epoch 404 / 1000 - loss: 0.7493 - accuracy: 0.7239 - val_loss: 0.9479 - val_accuracy: 0.5526\n",
            "Epoch 405 / 1000 - loss: 0.7489 - accuracy: 0.7239 - val_loss: 0.9460 - val_accuracy: 0.5526\n",
            "Epoch 406 / 1000 - loss: 0.7482 - accuracy: 0.7239 - val_loss: 0.9458 - val_accuracy: 0.5526\n",
            "Epoch 407 / 1000 - loss: 0.7476 - accuracy: 0.7239 - val_loss: 0.9448 - val_accuracy: 0.5526\n",
            "Epoch 408 / 1000 - loss: 0.7469 - accuracy: 0.7273 - val_loss: 0.9455 - val_accuracy: 0.5526\n",
            "Epoch 409 / 1000 - loss: 0.7464 - accuracy: 0.7273 - val_loss: 0.9453 - val_accuracy: 0.5526\n",
            "Epoch 410 / 1000 - loss: 0.7458 - accuracy: 0.7273 - val_loss: 0.9436 - val_accuracy: 0.5526\n",
            "Epoch 411 / 1000 - loss: 0.7452 - accuracy: 0.7273 - val_loss: 0.9426 - val_accuracy: 0.5526\n",
            "Epoch 412 / 1000 - loss: 0.7445 - accuracy: 0.7407 - val_loss: 0.9420 - val_accuracy: 0.5526\n",
            "Epoch 413 / 1000 - loss: 0.7440 - accuracy: 0.7273 - val_loss: 0.9414 - val_accuracy: 0.5526\n",
            "Epoch 414 / 1000 - loss: 0.7438 - accuracy: 0.7407 - val_loss: 0.9404 - val_accuracy: 0.5526\n",
            "Epoch 415 / 1000 - loss: 0.7428 - accuracy: 0.7407 - val_loss: 0.9405 - val_accuracy: 0.5526\n",
            "Epoch 416 / 1000 - loss: 0.7422 - accuracy: 0.7407 - val_loss: 0.9390 - val_accuracy: 0.5526\n",
            "Epoch 417 / 1000 - loss: 0.7415 - accuracy: 0.7407 - val_loss: 0.9385 - val_accuracy: 0.5526\n",
            "Epoch 418 / 1000 - loss: 0.7411 - accuracy: 0.7441 - val_loss: 0.9378 - val_accuracy: 0.5526\n",
            "Epoch 419 / 1000 - loss: 0.7404 - accuracy: 0.7441 - val_loss: 0.9373 - val_accuracy: 0.5526\n",
            "Epoch 420 / 1000 - loss: 0.7398 - accuracy: 0.7441 - val_loss: 0.9371 - val_accuracy: 0.5526\n",
            "Epoch 421 / 1000 - loss: 0.7393 - accuracy: 0.7475 - val_loss: 0.9358 - val_accuracy: 0.5526\n",
            "Epoch 422 / 1000 - loss: 0.7388 - accuracy: 0.7475 - val_loss: 0.9352 - val_accuracy: 0.5526\n",
            "Epoch 423 / 1000 - loss: 0.7382 - accuracy: 0.7475 - val_loss: 0.9342 - val_accuracy: 0.5526\n",
            "Epoch 424 / 1000 - loss: 0.7374 - accuracy: 0.7475 - val_loss: 0.9341 - val_accuracy: 0.5526\n",
            "Epoch 425 / 1000 - loss: 0.7369 - accuracy: 0.7475 - val_loss: 0.9343 - val_accuracy: 0.5526\n",
            "Epoch 426 / 1000 - loss: 0.7363 - accuracy: 0.7475 - val_loss: 0.9338 - val_accuracy: 0.5526\n",
            "Epoch 427 / 1000 - loss: 0.7361 - accuracy: 0.7475 - val_loss: 0.9308 - val_accuracy: 0.5526\n",
            "Epoch 428 / 1000 - loss: 0.7354 - accuracy: 0.7475 - val_loss: 0.9302 - val_accuracy: 0.5526\n",
            "Epoch 429 / 1000 - loss: 0.7347 - accuracy: 0.7475 - val_loss: 0.9299 - val_accuracy: 0.5526\n",
            "Epoch 430 / 1000 - loss: 0.7339 - accuracy: 0.7475 - val_loss: 0.9297 - val_accuracy: 0.5526\n",
            "Epoch 431 / 1000 - loss: 0.7334 - accuracy: 0.7475 - val_loss: 0.9294 - val_accuracy: 0.5526\n",
            "Epoch 432 / 1000 - loss: 0.7328 - accuracy: 0.7475 - val_loss: 0.9291 - val_accuracy: 0.5526\n",
            "Epoch 433 / 1000 - loss: 0.7322 - accuracy: 0.7475 - val_loss: 0.9277 - val_accuracy: 0.5526\n",
            "Epoch 434 / 1000 - loss: 0.7317 - accuracy: 0.7609 - val_loss: 0.9267 - val_accuracy: 0.5526\n",
            "Epoch 435 / 1000 - loss: 0.7312 - accuracy: 0.7643 - val_loss: 0.9260 - val_accuracy: 0.5526\n",
            "Epoch 436 / 1000 - loss: 0.7308 - accuracy: 0.7643 - val_loss: 0.9251 - val_accuracy: 0.5526\n",
            "Epoch 437 / 1000 - loss: 0.7300 - accuracy: 0.7508 - val_loss: 0.9255 - val_accuracy: 0.5526\n",
            "Epoch 438 / 1000 - loss: 0.7293 - accuracy: 0.7508 - val_loss: 0.9239 - val_accuracy: 0.5526\n",
            "Epoch 439 / 1000 - loss: 0.7291 - accuracy: 0.7643 - val_loss: 0.9230 - val_accuracy: 0.5526\n",
            "Epoch 440 / 1000 - loss: 0.7281 - accuracy: 0.7643 - val_loss: 0.9231 - val_accuracy: 0.5526\n",
            "Epoch 441 / 1000 - loss: 0.7278 - accuracy: 0.7508 - val_loss: 0.9211 - val_accuracy: 0.5526\n",
            "Epoch 442 / 1000 - loss: 0.7272 - accuracy: 0.7508 - val_loss: 0.9210 - val_accuracy: 0.5526\n",
            "Epoch 443 / 1000 - loss: 0.7267 - accuracy: 0.7643 - val_loss: 0.9196 - val_accuracy: 0.5526\n",
            "Epoch 444 / 1000 - loss: 0.7259 - accuracy: 0.7643 - val_loss: 0.9186 - val_accuracy: 0.5526\n",
            "Epoch 445 / 1000 - loss: 0.7254 - accuracy: 0.7643 - val_loss: 0.9173 - val_accuracy: 0.5526\n",
            "Epoch 446 / 1000 - loss: 0.7249 - accuracy: 0.7643 - val_loss: 0.9180 - val_accuracy: 0.5526\n",
            "Epoch 447 / 1000 - loss: 0.7240 - accuracy: 0.7643 - val_loss: 0.9185 - val_accuracy: 0.5526\n",
            "Epoch 448 / 1000 - loss: 0.7236 - accuracy: 0.7643 - val_loss: 0.9165 - val_accuracy: 0.5526\n",
            "Epoch 449 / 1000 - loss: 0.7229 - accuracy: 0.7643 - val_loss: 0.9164 - val_accuracy: 0.5526\n",
            "Epoch 450 / 1000 - loss: 0.7222 - accuracy: 0.7643 - val_loss: 0.9170 - val_accuracy: 0.5526\n",
            "Epoch 451 / 1000 - loss: 0.7216 - accuracy: 0.7643 - val_loss: 0.9172 - val_accuracy: 0.5526\n",
            "Epoch 452 / 1000 - loss: 0.7211 - accuracy: 0.7643 - val_loss: 0.9159 - val_accuracy: 0.5526\n",
            "Epoch 453 / 1000 - loss: 0.7205 - accuracy: 0.7643 - val_loss: 0.9148 - val_accuracy: 0.5526\n",
            "Epoch 454 / 1000 - loss: 0.7199 - accuracy: 0.7677 - val_loss: 0.9141 - val_accuracy: 0.5526\n",
            "Epoch 455 / 1000 - loss: 0.7194 - accuracy: 0.7643 - val_loss: 0.9136 - val_accuracy: 0.5526\n",
            "Epoch 456 / 1000 - loss: 0.7188 - accuracy: 0.7677 - val_loss: 0.9132 - val_accuracy: 0.5526\n",
            "Epoch 457 / 1000 - loss: 0.7182 - accuracy: 0.7643 - val_loss: 0.9122 - val_accuracy: 0.5526\n",
            "Epoch 458 / 1000 - loss: 0.7177 - accuracy: 0.7643 - val_loss: 0.9105 - val_accuracy: 0.5526\n",
            "Epoch 459 / 1000 - loss: 0.7170 - accuracy: 0.7677 - val_loss: 0.9105 - val_accuracy: 0.5526\n",
            "Epoch 460 / 1000 - loss: 0.7164 - accuracy: 0.7643 - val_loss: 0.9093 - val_accuracy: 0.5526\n",
            "Epoch 461 / 1000 - loss: 0.7159 - accuracy: 0.7643 - val_loss: 0.9081 - val_accuracy: 0.5526\n",
            "Epoch 462 / 1000 - loss: 0.7153 - accuracy: 0.7677 - val_loss: 0.9075 - val_accuracy: 0.5526\n",
            "Epoch 463 / 1000 - loss: 0.7146 - accuracy: 0.7710 - val_loss: 0.9074 - val_accuracy: 0.5526\n",
            "Epoch 464 / 1000 - loss: 0.7140 - accuracy: 0.7710 - val_loss: 0.9072 - val_accuracy: 0.5526\n",
            "Epoch 465 / 1000 - loss: 0.7136 - accuracy: 0.7710 - val_loss: 0.9057 - val_accuracy: 0.5526\n",
            "Epoch 466 / 1000 - loss: 0.7129 - accuracy: 0.7710 - val_loss: 0.9060 - val_accuracy: 0.5526\n",
            "Epoch 467 / 1000 - loss: 0.7124 - accuracy: 0.7710 - val_loss: 0.9054 - val_accuracy: 0.5526\n",
            "Epoch 468 / 1000 - loss: 0.7119 - accuracy: 0.7710 - val_loss: 0.9039 - val_accuracy: 0.5526\n",
            "Epoch 469 / 1000 - loss: 0.7111 - accuracy: 0.7710 - val_loss: 0.9040 - val_accuracy: 0.5526\n",
            "Epoch 470 / 1000 - loss: 0.7107 - accuracy: 0.7710 - val_loss: 0.9031 - val_accuracy: 0.5526\n",
            "Epoch 471 / 1000 - loss: 0.7099 - accuracy: 0.7710 - val_loss: 0.9026 - val_accuracy: 0.5526\n",
            "Epoch 472 / 1000 - loss: 0.7095 - accuracy: 0.7710 - val_loss: 0.9030 - val_accuracy: 0.5526\n",
            "Epoch 473 / 1000 - loss: 0.7090 - accuracy: 0.7710 - val_loss: 0.9030 - val_accuracy: 0.5526\n",
            "Epoch 474 / 1000 - loss: 0.7084 - accuracy: 0.7710 - val_loss: 0.9022 - val_accuracy: 0.5526\n",
            "Epoch 475 / 1000 - loss: 0.7079 - accuracy: 0.7710 - val_loss: 0.9011 - val_accuracy: 0.5526\n",
            "Epoch 476 / 1000 - loss: 0.7072 - accuracy: 0.7710 - val_loss: 0.8996 - val_accuracy: 0.5526\n",
            "Epoch 477 / 1000 - loss: 0.7068 - accuracy: 0.7710 - val_loss: 0.9003 - val_accuracy: 0.5526\n",
            "Epoch 478 / 1000 - loss: 0.7061 - accuracy: 0.7710 - val_loss: 0.8986 - val_accuracy: 0.5526\n",
            "Epoch 479 / 1000 - loss: 0.7058 - accuracy: 0.7710 - val_loss: 0.8991 - val_accuracy: 0.5526\n",
            "Epoch 480 / 1000 - loss: 0.7049 - accuracy: 0.7710 - val_loss: 0.8993 - val_accuracy: 0.5526\n",
            "Epoch 481 / 1000 - loss: 0.7046 - accuracy: 0.7710 - val_loss: 0.8964 - val_accuracy: 0.5526\n",
            "Epoch 482 / 1000 - loss: 0.7041 - accuracy: 0.7710 - val_loss: 0.8953 - val_accuracy: 0.5526\n",
            "Epoch 483 / 1000 - loss: 0.7033 - accuracy: 0.7710 - val_loss: 0.8949 - val_accuracy: 0.5526\n",
            "Epoch 484 / 1000 - loss: 0.7027 - accuracy: 0.7710 - val_loss: 0.8941 - val_accuracy: 0.5526\n",
            "Epoch 485 / 1000 - loss: 0.7022 - accuracy: 0.7710 - val_loss: 0.8944 - val_accuracy: 0.5526\n",
            "Epoch 486 / 1000 - loss: 0.7018 - accuracy: 0.7710 - val_loss: 0.8924 - val_accuracy: 0.5526\n",
            "Epoch 487 / 1000 - loss: 0.7010 - accuracy: 0.7710 - val_loss: 0.8934 - val_accuracy: 0.5526\n",
            "Epoch 488 / 1000 - loss: 0.7005 - accuracy: 0.7710 - val_loss: 0.8910 - val_accuracy: 0.5526\n",
            "Epoch 489 / 1000 - loss: 0.7001 - accuracy: 0.7710 - val_loss: 0.8899 - val_accuracy: 0.5526\n",
            "Epoch 490 / 1000 - loss: 0.6999 - accuracy: 0.7710 - val_loss: 0.8898 - val_accuracy: 0.5526\n",
            "Epoch 491 / 1000 - loss: 0.6996 - accuracy: 0.7710 - val_loss: 0.8883 - val_accuracy: 0.5526\n",
            "Epoch 492 / 1000 - loss: 0.6986 - accuracy: 0.7710 - val_loss: 0.8884 - val_accuracy: 0.5526\n",
            "Epoch 493 / 1000 - loss: 0.6981 - accuracy: 0.7710 - val_loss: 0.8888 - val_accuracy: 0.5526\n",
            "Epoch 494 / 1000 - loss: 0.6972 - accuracy: 0.7710 - val_loss: 0.8888 - val_accuracy: 0.5526\n",
            "Epoch 495 / 1000 - loss: 0.6970 - accuracy: 0.7710 - val_loss: 0.8895 - val_accuracy: 0.5526\n",
            "Epoch 496 / 1000 - loss: 0.6963 - accuracy: 0.7710 - val_loss: 0.8892 - val_accuracy: 0.5526\n",
            "Epoch 497 / 1000 - loss: 0.6959 - accuracy: 0.7710 - val_loss: 0.8884 - val_accuracy: 0.5526\n",
            "Epoch 498 / 1000 - loss: 0.6954 - accuracy: 0.7778 - val_loss: 0.8877 - val_accuracy: 0.5526\n",
            "Epoch 499 / 1000 - loss: 0.6948 - accuracy: 0.7744 - val_loss: 0.8877 - val_accuracy: 0.5526\n",
            "Epoch 500 / 1000 - loss: 0.6943 - accuracy: 0.7778 - val_loss: 0.8858 - val_accuracy: 0.5526\n",
            "Epoch 501 / 1000 - loss: 0.6937 - accuracy: 0.7744 - val_loss: 0.8855 - val_accuracy: 0.5526\n",
            "Epoch 502 / 1000 - loss: 0.6935 - accuracy: 0.7710 - val_loss: 0.8840 - val_accuracy: 0.5526\n",
            "Epoch 503 / 1000 - loss: 0.6929 - accuracy: 0.7710 - val_loss: 0.8826 - val_accuracy: 0.5526\n",
            "Epoch 504 / 1000 - loss: 0.6924 - accuracy: 0.7710 - val_loss: 0.8832 - val_accuracy: 0.5526\n",
            "Epoch 505 / 1000 - loss: 0.6921 - accuracy: 0.7744 - val_loss: 0.8823 - val_accuracy: 0.5526\n",
            "Epoch 506 / 1000 - loss: 0.6912 - accuracy: 0.7778 - val_loss: 0.8831 - val_accuracy: 0.5526\n",
            "Epoch 507 / 1000 - loss: 0.6907 - accuracy: 0.7845 - val_loss: 0.8816 - val_accuracy: 0.5526\n",
            "Epoch 508 / 1000 - loss: 0.6902 - accuracy: 0.7778 - val_loss: 0.8827 - val_accuracy: 0.5526\n",
            "Epoch 509 / 1000 - loss: 0.6897 - accuracy: 0.7811 - val_loss: 0.8814 - val_accuracy: 0.5526\n",
            "Epoch 510 / 1000 - loss: 0.6892 - accuracy: 0.7845 - val_loss: 0.8799 - val_accuracy: 0.5526\n",
            "Epoch 511 / 1000 - loss: 0.6888 - accuracy: 0.7845 - val_loss: 0.8793 - val_accuracy: 0.5526\n",
            "Epoch 512 / 1000 - loss: 0.6882 - accuracy: 0.7845 - val_loss: 0.8781 - val_accuracy: 0.5526\n",
            "Epoch 513 / 1000 - loss: 0.6877 - accuracy: 0.7845 - val_loss: 0.8780 - val_accuracy: 0.5526\n",
            "Epoch 514 / 1000 - loss: 0.6871 - accuracy: 0.7845 - val_loss: 0.8770 - val_accuracy: 0.5526\n",
            "Epoch 515 / 1000 - loss: 0.6866 - accuracy: 0.7845 - val_loss: 0.8770 - val_accuracy: 0.5526\n",
            "Epoch 516 / 1000 - loss: 0.6864 - accuracy: 0.7845 - val_loss: 0.8781 - val_accuracy: 0.5526\n",
            "Epoch 517 / 1000 - loss: 0.6857 - accuracy: 0.7845 - val_loss: 0.8778 - val_accuracy: 0.5526\n",
            "Epoch 518 / 1000 - loss: 0.6852 - accuracy: 0.7845 - val_loss: 0.8773 - val_accuracy: 0.5526\n",
            "Epoch 519 / 1000 - loss: 0.6846 - accuracy: 0.7811 - val_loss: 0.8772 - val_accuracy: 0.5526\n",
            "Epoch 520 / 1000 - loss: 0.6844 - accuracy: 0.7980 - val_loss: 0.8745 - val_accuracy: 0.5526\n",
            "Epoch 521 / 1000 - loss: 0.6836 - accuracy: 0.8013 - val_loss: 0.8737 - val_accuracy: 0.5526\n",
            "Epoch 522 / 1000 - loss: 0.6834 - accuracy: 0.8013 - val_loss: 0.8714 - val_accuracy: 0.5526\n",
            "Epoch 523 / 1000 - loss: 0.6827 - accuracy: 0.7879 - val_loss: 0.8721 - val_accuracy: 0.5526\n",
            "Epoch 524 / 1000 - loss: 0.6825 - accuracy: 0.8013 - val_loss: 0.8718 - val_accuracy: 0.5526\n",
            "Epoch 525 / 1000 - loss: 0.6817 - accuracy: 0.7879 - val_loss: 0.8722 - val_accuracy: 0.5526\n",
            "Epoch 526 / 1000 - loss: 0.6812 - accuracy: 0.7879 - val_loss: 0.8710 - val_accuracy: 0.5526\n",
            "Epoch 527 / 1000 - loss: 0.6810 - accuracy: 0.8013 - val_loss: 0.8691 - val_accuracy: 0.5526\n",
            "Epoch 528 / 1000 - loss: 0.6802 - accuracy: 0.8013 - val_loss: 0.8697 - val_accuracy: 0.5526\n",
            "Epoch 529 / 1000 - loss: 0.6797 - accuracy: 0.8013 - val_loss: 0.8697 - val_accuracy: 0.5526\n",
            "Epoch 530 / 1000 - loss: 0.6795 - accuracy: 0.8013 - val_loss: 0.8684 - val_accuracy: 0.5526\n",
            "Epoch 531 / 1000 - loss: 0.6794 - accuracy: 0.8013 - val_loss: 0.8678 - val_accuracy: 0.5526\n",
            "Epoch 532 / 1000 - loss: 0.6784 - accuracy: 0.8013 - val_loss: 0.8676 - val_accuracy: 0.5526\n",
            "Epoch 533 / 1000 - loss: 0.6779 - accuracy: 0.8013 - val_loss: 0.8666 - val_accuracy: 0.5526\n",
            "Epoch 534 / 1000 - loss: 0.6775 - accuracy: 0.8013 - val_loss: 0.8661 - val_accuracy: 0.5526\n",
            "Epoch 535 / 1000 - loss: 0.6769 - accuracy: 0.8013 - val_loss: 0.8655 - val_accuracy: 0.5526\n",
            "Epoch 536 / 1000 - loss: 0.6762 - accuracy: 0.8013 - val_loss: 0.8654 - val_accuracy: 0.5526\n",
            "Epoch 537 / 1000 - loss: 0.6760 - accuracy: 0.8013 - val_loss: 0.8680 - val_accuracy: 0.5526\n",
            "Epoch 538 / 1000 - loss: 0.6754 - accuracy: 0.8013 - val_loss: 0.8673 - val_accuracy: 0.5789\n",
            "Epoch 539 / 1000 - loss: 0.6749 - accuracy: 0.8013 - val_loss: 0.8658 - val_accuracy: 0.5526\n",
            "Epoch 540 / 1000 - loss: 0.6744 - accuracy: 0.8013 - val_loss: 0.8646 - val_accuracy: 0.5526\n",
            "Epoch 541 / 1000 - loss: 0.6741 - accuracy: 0.8013 - val_loss: 0.8630 - val_accuracy: 0.5526\n",
            "Epoch 542 / 1000 - loss: 0.6734 - accuracy: 0.8013 - val_loss: 0.8621 - val_accuracy: 0.5526\n",
            "Epoch 543 / 1000 - loss: 0.6730 - accuracy: 0.8013 - val_loss: 0.8630 - val_accuracy: 0.5526\n",
            "Epoch 544 / 1000 - loss: 0.6724 - accuracy: 0.8013 - val_loss: 0.8622 - val_accuracy: 0.5789\n",
            "Epoch 545 / 1000 - loss: 0.6719 - accuracy: 0.8013 - val_loss: 0.8613 - val_accuracy: 0.5789\n",
            "Epoch 546 / 1000 - loss: 0.6715 - accuracy: 0.8013 - val_loss: 0.8607 - val_accuracy: 0.5789\n",
            "Epoch 547 / 1000 - loss: 0.6710 - accuracy: 0.8013 - val_loss: 0.8618 - val_accuracy: 0.5789\n",
            "Epoch 548 / 1000 - loss: 0.6707 - accuracy: 0.8013 - val_loss: 0.8610 - val_accuracy: 0.5789\n",
            "Epoch 549 / 1000 - loss: 0.6702 - accuracy: 0.8013 - val_loss: 0.8614 - val_accuracy: 0.5789\n",
            "Epoch 550 / 1000 - loss: 0.6698 - accuracy: 0.8013 - val_loss: 0.8598 - val_accuracy: 0.5789\n",
            "Epoch 551 / 1000 - loss: 0.6691 - accuracy: 0.8013 - val_loss: 0.8585 - val_accuracy: 0.5789\n",
            "Epoch 552 / 1000 - loss: 0.6685 - accuracy: 0.8013 - val_loss: 0.8577 - val_accuracy: 0.5789\n",
            "Epoch 553 / 1000 - loss: 0.6682 - accuracy: 0.8013 - val_loss: 0.8583 - val_accuracy: 0.5789\n",
            "Epoch 554 / 1000 - loss: 0.6676 - accuracy: 0.8013 - val_loss: 0.8577 - val_accuracy: 0.5789\n",
            "Epoch 555 / 1000 - loss: 0.6672 - accuracy: 0.8047 - val_loss: 0.8588 - val_accuracy: 0.5789\n",
            "Epoch 556 / 1000 - loss: 0.6668 - accuracy: 0.8047 - val_loss: 0.8570 - val_accuracy: 0.5789\n",
            "Epoch 557 / 1000 - loss: 0.6665 - accuracy: 0.8013 - val_loss: 0.8547 - val_accuracy: 0.5789\n",
            "Epoch 558 / 1000 - loss: 0.6658 - accuracy: 0.8013 - val_loss: 0.8542 - val_accuracy: 0.5789\n",
            "Epoch 559 / 1000 - loss: 0.6653 - accuracy: 0.8047 - val_loss: 0.8539 - val_accuracy: 0.5789\n",
            "Epoch 560 / 1000 - loss: 0.6648 - accuracy: 0.8047 - val_loss: 0.8531 - val_accuracy: 0.5789\n",
            "Epoch 561 / 1000 - loss: 0.6643 - accuracy: 0.8013 - val_loss: 0.8533 - val_accuracy: 0.5789\n",
            "Epoch 562 / 1000 - loss: 0.6639 - accuracy: 0.8013 - val_loss: 0.8525 - val_accuracy: 0.5789\n",
            "Epoch 563 / 1000 - loss: 0.6636 - accuracy: 0.8047 - val_loss: 0.8522 - val_accuracy: 0.5789\n",
            "Epoch 564 / 1000 - loss: 0.6629 - accuracy: 0.8013 - val_loss: 0.8518 - val_accuracy: 0.5789\n",
            "Epoch 565 / 1000 - loss: 0.6625 - accuracy: 0.8047 - val_loss: 0.8513 - val_accuracy: 0.5789\n",
            "Epoch 566 / 1000 - loss: 0.6622 - accuracy: 0.8114 - val_loss: 0.8503 - val_accuracy: 0.6053\n",
            "Epoch 567 / 1000 - loss: 0.6615 - accuracy: 0.8047 - val_loss: 0.8521 - val_accuracy: 0.5789\n",
            "Epoch 568 / 1000 - loss: 0.6611 - accuracy: 0.8047 - val_loss: 0.8521 - val_accuracy: 0.5789\n",
            "Epoch 569 / 1000 - loss: 0.6608 - accuracy: 0.8148 - val_loss: 0.8491 - val_accuracy: 0.6053\n",
            "Epoch 570 / 1000 - loss: 0.6603 - accuracy: 0.8114 - val_loss: 0.8476 - val_accuracy: 0.6316\n",
            "Epoch 571 / 1000 - loss: 0.6599 - accuracy: 0.8148 - val_loss: 0.8468 - val_accuracy: 0.6316\n",
            "Epoch 572 / 1000 - loss: 0.6595 - accuracy: 0.8047 - val_loss: 0.8494 - val_accuracy: 0.6053\n",
            "Epoch 573 / 1000 - loss: 0.6591 - accuracy: 0.8148 - val_loss: 0.8479 - val_accuracy: 0.6316\n",
            "Epoch 574 / 1000 - loss: 0.6587 - accuracy: 0.8182 - val_loss: 0.8460 - val_accuracy: 0.6316\n",
            "Epoch 575 / 1000 - loss: 0.6577 - accuracy: 0.8148 - val_loss: 0.8472 - val_accuracy: 0.6316\n",
            "Epoch 576 / 1000 - loss: 0.6576 - accuracy: 0.8114 - val_loss: 0.8471 - val_accuracy: 0.6316\n",
            "Epoch 577 / 1000 - loss: 0.6570 - accuracy: 0.8148 - val_loss: 0.8451 - val_accuracy: 0.6316\n",
            "Epoch 578 / 1000 - loss: 0.6565 - accuracy: 0.8182 - val_loss: 0.8445 - val_accuracy: 0.6316\n",
            "Epoch 579 / 1000 - loss: 0.6560 - accuracy: 0.8148 - val_loss: 0.8448 - val_accuracy: 0.6316\n",
            "Epoch 580 / 1000 - loss: 0.6556 - accuracy: 0.8182 - val_loss: 0.8436 - val_accuracy: 0.6316\n",
            "Epoch 581 / 1000 - loss: 0.6554 - accuracy: 0.8215 - val_loss: 0.8423 - val_accuracy: 0.6316\n",
            "Epoch 582 / 1000 - loss: 0.6545 - accuracy: 0.8148 - val_loss: 0.8437 - val_accuracy: 0.6579\n",
            "Epoch 583 / 1000 - loss: 0.6542 - accuracy: 0.8148 - val_loss: 0.8446 - val_accuracy: 0.6579\n",
            "Epoch 584 / 1000 - loss: 0.6536 - accuracy: 0.8148 - val_loss: 0.8446 - val_accuracy: 0.6579\n",
            "Epoch 585 / 1000 - loss: 0.6531 - accuracy: 0.8148 - val_loss: 0.8436 - val_accuracy: 0.6579\n",
            "Epoch 586 / 1000 - loss: 0.6531 - accuracy: 0.8182 - val_loss: 0.8421 - val_accuracy: 0.6579\n",
            "Epoch 587 / 1000 - loss: 0.6525 - accuracy: 0.8148 - val_loss: 0.8420 - val_accuracy: 0.6579\n",
            "Epoch 588 / 1000 - loss: 0.6519 - accuracy: 0.8182 - val_loss: 0.8406 - val_accuracy: 0.6579\n",
            "Epoch 589 / 1000 - loss: 0.6516 - accuracy: 0.8114 - val_loss: 0.8418 - val_accuracy: 0.6579\n",
            "Epoch 590 / 1000 - loss: 0.6510 - accuracy: 0.8215 - val_loss: 0.8391 - val_accuracy: 0.6579\n",
            "Epoch 591 / 1000 - loss: 0.6507 - accuracy: 0.8215 - val_loss: 0.8393 - val_accuracy: 0.6579\n",
            "Epoch 592 / 1000 - loss: 0.6500 - accuracy: 0.8215 - val_loss: 0.8391 - val_accuracy: 0.6579\n",
            "Epoch 593 / 1000 - loss: 0.6495 - accuracy: 0.8215 - val_loss: 0.8382 - val_accuracy: 0.6579\n",
            "Epoch 594 / 1000 - loss: 0.6494 - accuracy: 0.8148 - val_loss: 0.8409 - val_accuracy: 0.6579\n",
            "Epoch 595 / 1000 - loss: 0.6488 - accuracy: 0.8215 - val_loss: 0.8372 - val_accuracy: 0.6579\n",
            "Epoch 596 / 1000 - loss: 0.6485 - accuracy: 0.8215 - val_loss: 0.8380 - val_accuracy: 0.6579\n",
            "Epoch 597 / 1000 - loss: 0.6478 - accuracy: 0.8215 - val_loss: 0.8380 - val_accuracy: 0.6579\n",
            "Epoch 598 / 1000 - loss: 0.6474 - accuracy: 0.8182 - val_loss: 0.8375 - val_accuracy: 0.6579\n",
            "Epoch 599 / 1000 - loss: 0.6475 - accuracy: 0.8384 - val_loss: 0.8365 - val_accuracy: 0.6579\n",
            "Epoch 600 / 1000 - loss: 0.6466 - accuracy: 0.8182 - val_loss: 0.8373 - val_accuracy: 0.6579\n",
            "Epoch 601 / 1000 - loss: 0.6463 - accuracy: 0.8249 - val_loss: 0.8338 - val_accuracy: 0.6842\n",
            "Epoch 602 / 1000 - loss: 0.6456 - accuracy: 0.8249 - val_loss: 0.8344 - val_accuracy: 0.6579\n",
            "Epoch 603 / 1000 - loss: 0.6451 - accuracy: 0.8283 - val_loss: 0.8324 - val_accuracy: 0.6842\n",
            "Epoch 604 / 1000 - loss: 0.6445 - accuracy: 0.8283 - val_loss: 0.8318 - val_accuracy: 0.6842\n",
            "Epoch 605 / 1000 - loss: 0.6441 - accuracy: 0.8283 - val_loss: 0.8321 - val_accuracy: 0.6842\n",
            "Epoch 606 / 1000 - loss: 0.6439 - accuracy: 0.8418 - val_loss: 0.8320 - val_accuracy: 0.6842\n",
            "Epoch 607 / 1000 - loss: 0.6438 - accuracy: 0.8283 - val_loss: 0.8283 - val_accuracy: 0.6842\n",
            "Epoch 608 / 1000 - loss: 0.6428 - accuracy: 0.8283 - val_loss: 0.8314 - val_accuracy: 0.6842\n",
            "Epoch 609 / 1000 - loss: 0.6426 - accuracy: 0.8283 - val_loss: 0.8290 - val_accuracy: 0.6842\n",
            "Epoch 610 / 1000 - loss: 0.6422 - accuracy: 0.8283 - val_loss: 0.8283 - val_accuracy: 0.6842\n",
            "Epoch 611 / 1000 - loss: 0.6418 - accuracy: 0.8283 - val_loss: 0.8298 - val_accuracy: 0.6842\n",
            "Epoch 612 / 1000 - loss: 0.6412 - accuracy: 0.8283 - val_loss: 0.8290 - val_accuracy: 0.6842\n",
            "Epoch 613 / 1000 - loss: 0.6406 - accuracy: 0.8283 - val_loss: 0.8279 - val_accuracy: 0.6842\n",
            "Epoch 614 / 1000 - loss: 0.6403 - accuracy: 0.8283 - val_loss: 0.8287 - val_accuracy: 0.6842\n",
            "Epoch 615 / 1000 - loss: 0.6399 - accuracy: 0.8283 - val_loss: 0.8282 - val_accuracy: 0.6842\n",
            "Epoch 616 / 1000 - loss: 0.6395 - accuracy: 0.8283 - val_loss: 0.8274 - val_accuracy: 0.6842\n",
            "Epoch 617 / 1000 - loss: 0.6388 - accuracy: 0.8283 - val_loss: 0.8276 - val_accuracy: 0.6842\n",
            "Epoch 618 / 1000 - loss: 0.6386 - accuracy: 0.8283 - val_loss: 0.8261 - val_accuracy: 0.6842\n",
            "Epoch 619 / 1000 - loss: 0.6385 - accuracy: 0.8451 - val_loss: 0.8253 - val_accuracy: 0.6842\n",
            "Epoch 620 / 1000 - loss: 0.6382 - accuracy: 0.8418 - val_loss: 0.8240 - val_accuracy: 0.6842\n",
            "Epoch 621 / 1000 - loss: 0.6371 - accuracy: 0.8283 - val_loss: 0.8244 - val_accuracy: 0.6842\n",
            "Epoch 622 / 1000 - loss: 0.6367 - accuracy: 0.8283 - val_loss: 0.8255 - val_accuracy: 0.6842\n",
            "Epoch 623 / 1000 - loss: 0.6363 - accuracy: 0.8283 - val_loss: 0.8247 - val_accuracy: 0.6842\n",
            "Epoch 624 / 1000 - loss: 0.6358 - accuracy: 0.8283 - val_loss: 0.8228 - val_accuracy: 0.6842\n",
            "Epoch 625 / 1000 - loss: 0.6354 - accuracy: 0.8283 - val_loss: 0.8231 - val_accuracy: 0.6842\n",
            "Epoch 626 / 1000 - loss: 0.6349 - accuracy: 0.8283 - val_loss: 0.8231 - val_accuracy: 0.6842\n",
            "Epoch 627 / 1000 - loss: 0.6347 - accuracy: 0.8451 - val_loss: 0.8214 - val_accuracy: 0.6842\n",
            "Epoch 628 / 1000 - loss: 0.6341 - accuracy: 0.8316 - val_loss: 0.8212 - val_accuracy: 0.6842\n",
            "Epoch 629 / 1000 - loss: 0.6336 - accuracy: 0.8316 - val_loss: 0.8212 - val_accuracy: 0.6842\n",
            "Epoch 630 / 1000 - loss: 0.6332 - accuracy: 0.8316 - val_loss: 0.8194 - val_accuracy: 0.6842\n",
            "Epoch 631 / 1000 - loss: 0.6333 - accuracy: 0.8451 - val_loss: 0.8162 - val_accuracy: 0.6842\n",
            "Epoch 632 / 1000 - loss: 0.6323 - accuracy: 0.8316 - val_loss: 0.8186 - val_accuracy: 0.6842\n",
            "Epoch 633 / 1000 - loss: 0.6320 - accuracy: 0.8316 - val_loss: 0.8160 - val_accuracy: 0.6842\n",
            "Epoch 634 / 1000 - loss: 0.6316 - accuracy: 0.8451 - val_loss: 0.8189 - val_accuracy: 0.6842\n",
            "Epoch 635 / 1000 - loss: 0.6312 - accuracy: 0.8316 - val_loss: 0.8182 - val_accuracy: 0.6842\n",
            "Epoch 636 / 1000 - loss: 0.6306 - accuracy: 0.8316 - val_loss: 0.8171 - val_accuracy: 0.6842\n",
            "Epoch 637 / 1000 - loss: 0.6302 - accuracy: 0.8451 - val_loss: 0.8177 - val_accuracy: 0.7105\n",
            "Epoch 638 / 1000 - loss: 0.6298 - accuracy: 0.8451 - val_loss: 0.8177 - val_accuracy: 0.6842\n",
            "Epoch 639 / 1000 - loss: 0.6295 - accuracy: 0.8451 - val_loss: 0.8173 - val_accuracy: 0.7105\n",
            "Epoch 640 / 1000 - loss: 0.6289 - accuracy: 0.8316 - val_loss: 0.8165 - val_accuracy: 0.7105\n",
            "Epoch 641 / 1000 - loss: 0.6286 - accuracy: 0.8316 - val_loss: 0.8159 - val_accuracy: 0.7105\n",
            "Epoch 642 / 1000 - loss: 0.6279 - accuracy: 0.8451 - val_loss: 0.8156 - val_accuracy: 0.7105\n",
            "Epoch 643 / 1000 - loss: 0.6276 - accuracy: 0.8316 - val_loss: 0.8140 - val_accuracy: 0.6842\n",
            "Epoch 644 / 1000 - loss: 0.6273 - accuracy: 0.8451 - val_loss: 0.8132 - val_accuracy: 0.6842\n",
            "Epoch 645 / 1000 - loss: 0.6270 - accuracy: 0.8451 - val_loss: 0.8126 - val_accuracy: 0.6842\n",
            "Epoch 646 / 1000 - loss: 0.6266 - accuracy: 0.8451 - val_loss: 0.8134 - val_accuracy: 0.7105\n",
            "Epoch 647 / 1000 - loss: 0.6260 - accuracy: 0.8451 - val_loss: 0.8129 - val_accuracy: 0.7105\n",
            "Epoch 648 / 1000 - loss: 0.6255 - accuracy: 0.8451 - val_loss: 0.8129 - val_accuracy: 0.7105\n",
            "Epoch 649 / 1000 - loss: 0.6252 - accuracy: 0.8451 - val_loss: 0.8137 - val_accuracy: 0.7105\n",
            "Epoch 650 / 1000 - loss: 0.6247 - accuracy: 0.8451 - val_loss: 0.8105 - val_accuracy: 0.7105\n",
            "Epoch 651 / 1000 - loss: 0.6242 - accuracy: 0.8451 - val_loss: 0.8116 - val_accuracy: 0.7105\n",
            "Epoch 652 / 1000 - loss: 0.6238 - accuracy: 0.8451 - val_loss: 0.8107 - val_accuracy: 0.7105\n",
            "Epoch 653 / 1000 - loss: 0.6233 - accuracy: 0.8451 - val_loss: 0.8100 - val_accuracy: 0.7105\n",
            "Epoch 654 / 1000 - loss: 0.6231 - accuracy: 0.8451 - val_loss: 0.8100 - val_accuracy: 0.7105\n",
            "Epoch 655 / 1000 - loss: 0.6226 - accuracy: 0.8316 - val_loss: 0.8104 - val_accuracy: 0.7105\n",
            "Epoch 656 / 1000 - loss: 0.6222 - accuracy: 0.8316 - val_loss: 0.8113 - val_accuracy: 0.7105\n",
            "Epoch 657 / 1000 - loss: 0.6218 - accuracy: 0.8316 - val_loss: 0.8111 - val_accuracy: 0.7105\n",
            "Epoch 658 / 1000 - loss: 0.6216 - accuracy: 0.8451 - val_loss: 0.8064 - val_accuracy: 0.7105\n",
            "Epoch 659 / 1000 - loss: 0.6209 - accuracy: 0.8316 - val_loss: 0.8084 - val_accuracy: 0.7105\n",
            "Epoch 660 / 1000 - loss: 0.6204 - accuracy: 0.8451 - val_loss: 0.8075 - val_accuracy: 0.7105\n",
            "Epoch 661 / 1000 - loss: 0.6201 - accuracy: 0.8451 - val_loss: 0.8076 - val_accuracy: 0.7105\n",
            "Epoch 662 / 1000 - loss: 0.6197 - accuracy: 0.8451 - val_loss: 0.8075 - val_accuracy: 0.7105\n",
            "Epoch 663 / 1000 - loss: 0.6192 - accuracy: 0.8451 - val_loss: 0.8074 - val_accuracy: 0.7105\n",
            "Epoch 664 / 1000 - loss: 0.6189 - accuracy: 0.8451 - val_loss: 0.8057 - val_accuracy: 0.7105\n",
            "Epoch 665 / 1000 - loss: 0.6185 - accuracy: 0.8316 - val_loss: 0.8059 - val_accuracy: 0.7105\n",
            "Epoch 666 / 1000 - loss: 0.6179 - accuracy: 0.8451 - val_loss: 0.8035 - val_accuracy: 0.7105\n",
            "Epoch 667 / 1000 - loss: 0.6178 - accuracy: 0.8451 - val_loss: 0.8012 - val_accuracy: 0.6842\n",
            "Epoch 668 / 1000 - loss: 0.6173 - accuracy: 0.8451 - val_loss: 0.8052 - val_accuracy: 0.7105\n",
            "Epoch 669 / 1000 - loss: 0.6168 - accuracy: 0.8451 - val_loss: 0.8015 - val_accuracy: 0.7105\n",
            "Epoch 670 / 1000 - loss: 0.6163 - accuracy: 0.8451 - val_loss: 0.8021 - val_accuracy: 0.7105\n",
            "Epoch 671 / 1000 - loss: 0.6161 - accuracy: 0.8451 - val_loss: 0.8001 - val_accuracy: 0.7105\n",
            "Epoch 672 / 1000 - loss: 0.6156 - accuracy: 0.8451 - val_loss: 0.7995 - val_accuracy: 0.7105\n",
            "Epoch 673 / 1000 - loss: 0.6151 - accuracy: 0.8451 - val_loss: 0.7988 - val_accuracy: 0.7105\n",
            "Epoch 674 / 1000 - loss: 0.6146 - accuracy: 0.8451 - val_loss: 0.8011 - val_accuracy: 0.7105\n",
            "Epoch 675 / 1000 - loss: 0.6145 - accuracy: 0.8451 - val_loss: 0.7984 - val_accuracy: 0.7105\n",
            "Epoch 676 / 1000 - loss: 0.6139 - accuracy: 0.8451 - val_loss: 0.7990 - val_accuracy: 0.7105\n",
            "Epoch 677 / 1000 - loss: 0.6138 - accuracy: 0.8485 - val_loss: 0.7948 - val_accuracy: 0.7105\n",
            "Epoch 678 / 1000 - loss: 0.6132 - accuracy: 0.8451 - val_loss: 0.7978 - val_accuracy: 0.7105\n",
            "Epoch 679 / 1000 - loss: 0.6129 - accuracy: 0.8485 - val_loss: 0.7959 - val_accuracy: 0.7105\n",
            "Epoch 680 / 1000 - loss: 0.6125 - accuracy: 0.8451 - val_loss: 0.7958 - val_accuracy: 0.7105\n",
            "Epoch 681 / 1000 - loss: 0.6126 - accuracy: 0.8485 - val_loss: 0.7929 - val_accuracy: 0.7105\n",
            "Epoch 682 / 1000 - loss: 0.6114 - accuracy: 0.8451 - val_loss: 0.7979 - val_accuracy: 0.7105\n",
            "Epoch 683 / 1000 - loss: 0.6110 - accuracy: 0.8451 - val_loss: 0.7982 - val_accuracy: 0.7105\n",
            "Epoch 684 / 1000 - loss: 0.6109 - accuracy: 0.8451 - val_loss: 0.7945 - val_accuracy: 0.7105\n",
            "Epoch 685 / 1000 - loss: 0.6100 - accuracy: 0.8451 - val_loss: 0.7953 - val_accuracy: 0.7105\n",
            "Epoch 686 / 1000 - loss: 0.6098 - accuracy: 0.8451 - val_loss: 0.7956 - val_accuracy: 0.7105\n",
            "Epoch 687 / 1000 - loss: 0.6095 - accuracy: 0.8451 - val_loss: 0.7964 - val_accuracy: 0.7105\n",
            "Epoch 688 / 1000 - loss: 0.6092 - accuracy: 0.8451 - val_loss: 0.7966 - val_accuracy: 0.7105\n",
            "Epoch 689 / 1000 - loss: 0.6090 - accuracy: 0.8451 - val_loss: 0.7943 - val_accuracy: 0.7105\n",
            "Epoch 690 / 1000 - loss: 0.6083 - accuracy: 0.8485 - val_loss: 0.7919 - val_accuracy: 0.7105\n",
            "Epoch 691 / 1000 - loss: 0.6083 - accuracy: 0.8485 - val_loss: 0.7896 - val_accuracy: 0.7105\n",
            "Epoch 692 / 1000 - loss: 0.6075 - accuracy: 0.8485 - val_loss: 0.7923 - val_accuracy: 0.7105\n",
            "Epoch 693 / 1000 - loss: 0.6074 - accuracy: 0.8485 - val_loss: 0.7925 - val_accuracy: 0.7105\n",
            "Epoch 694 / 1000 - loss: 0.6066 - accuracy: 0.8485 - val_loss: 0.7920 - val_accuracy: 0.7105\n",
            "Epoch 695 / 1000 - loss: 0.6063 - accuracy: 0.8451 - val_loss: 0.7927 - val_accuracy: 0.7105\n",
            "Epoch 696 / 1000 - loss: 0.6060 - accuracy: 0.8485 - val_loss: 0.7902 - val_accuracy: 0.7105\n",
            "Epoch 697 / 1000 - loss: 0.6055 - accuracy: 0.8485 - val_loss: 0.7907 - val_accuracy: 0.7105\n",
            "Epoch 698 / 1000 - loss: 0.6052 - accuracy: 0.8316 - val_loss: 0.7916 - val_accuracy: 0.7105\n",
            "Epoch 699 / 1000 - loss: 0.6047 - accuracy: 0.8485 - val_loss: 0.7911 - val_accuracy: 0.7105\n",
            "Epoch 700 / 1000 - loss: 0.6042 - accuracy: 0.8485 - val_loss: 0.7893 - val_accuracy: 0.7105\n",
            "Epoch 701 / 1000 - loss: 0.6039 - accuracy: 0.8485 - val_loss: 0.7862 - val_accuracy: 0.7105\n",
            "Epoch 702 / 1000 - loss: 0.6035 - accuracy: 0.8485 - val_loss: 0.7902 - val_accuracy: 0.7105\n",
            "Epoch 703 / 1000 - loss: 0.6036 - accuracy: 0.8485 - val_loss: 0.7874 - val_accuracy: 0.7105\n",
            "Epoch 704 / 1000 - loss: 0.6028 - accuracy: 0.8485 - val_loss: 0.7863 - val_accuracy: 0.7105\n",
            "Epoch 705 / 1000 - loss: 0.6022 - accuracy: 0.8485 - val_loss: 0.7861 - val_accuracy: 0.7105\n",
            "Epoch 706 / 1000 - loss: 0.6019 - accuracy: 0.8485 - val_loss: 0.7876 - val_accuracy: 0.7105\n",
            "Epoch 707 / 1000 - loss: 0.6016 - accuracy: 0.8485 - val_loss: 0.7885 - val_accuracy: 0.7105\n",
            "Epoch 708 / 1000 - loss: 0.6010 - accuracy: 0.8485 - val_loss: 0.7873 - val_accuracy: 0.7105\n",
            "Epoch 709 / 1000 - loss: 0.6008 - accuracy: 0.8485 - val_loss: 0.7866 - val_accuracy: 0.7105\n",
            "Epoch 710 / 1000 - loss: 0.6003 - accuracy: 0.8485 - val_loss: 0.7852 - val_accuracy: 0.7105\n",
            "Epoch 711 / 1000 - loss: 0.6000 - accuracy: 0.8485 - val_loss: 0.7831 - val_accuracy: 0.7105\n",
            "Epoch 712 / 1000 - loss: 0.5994 - accuracy: 0.8485 - val_loss: 0.7838 - val_accuracy: 0.7105\n",
            "Epoch 713 / 1000 - loss: 0.5990 - accuracy: 0.8485 - val_loss: 0.7842 - val_accuracy: 0.7105\n",
            "Epoch 714 / 1000 - loss: 0.5989 - accuracy: 0.8485 - val_loss: 0.7834 - val_accuracy: 0.7105\n",
            "Epoch 715 / 1000 - loss: 0.5983 - accuracy: 0.8485 - val_loss: 0.7839 - val_accuracy: 0.7105\n",
            "Epoch 716 / 1000 - loss: 0.5979 - accuracy: 0.8485 - val_loss: 0.7820 - val_accuracy: 0.7105\n",
            "Epoch 717 / 1000 - loss: 0.5976 - accuracy: 0.8485 - val_loss: 0.7823 - val_accuracy: 0.7105\n",
            "Epoch 718 / 1000 - loss: 0.5970 - accuracy: 0.8485 - val_loss: 0.7836 - val_accuracy: 0.7105\n",
            "Epoch 719 / 1000 - loss: 0.5968 - accuracy: 0.8485 - val_loss: 0.7804 - val_accuracy: 0.7105\n",
            "Epoch 720 / 1000 - loss: 0.5964 - accuracy: 0.8485 - val_loss: 0.7816 - val_accuracy: 0.7105\n",
            "Epoch 721 / 1000 - loss: 0.5963 - accuracy: 0.8519 - val_loss: 0.7765 - val_accuracy: 0.7105\n",
            "Epoch 722 / 1000 - loss: 0.5956 - accuracy: 0.8485 - val_loss: 0.7796 - val_accuracy: 0.7105\n",
            "Epoch 723 / 1000 - loss: 0.5952 - accuracy: 0.8485 - val_loss: 0.7787 - val_accuracy: 0.7105\n",
            "Epoch 724 / 1000 - loss: 0.5952 - accuracy: 0.8485 - val_loss: 0.7779 - val_accuracy: 0.7105\n",
            "Epoch 725 / 1000 - loss: 0.5944 - accuracy: 0.8485 - val_loss: 0.7781 - val_accuracy: 0.7105\n",
            "Epoch 726 / 1000 - loss: 0.5942 - accuracy: 0.8485 - val_loss: 0.7773 - val_accuracy: 0.7105\n",
            "Epoch 727 / 1000 - loss: 0.5936 - accuracy: 0.8485 - val_loss: 0.7756 - val_accuracy: 0.7105\n",
            "Epoch 728 / 1000 - loss: 0.5933 - accuracy: 0.8485 - val_loss: 0.7767 - val_accuracy: 0.7105\n",
            "Epoch 729 / 1000 - loss: 0.5928 - accuracy: 0.8485 - val_loss: 0.7760 - val_accuracy: 0.7105\n",
            "Epoch 730 / 1000 - loss: 0.5926 - accuracy: 0.8485 - val_loss: 0.7764 - val_accuracy: 0.7105\n",
            "Epoch 731 / 1000 - loss: 0.5922 - accuracy: 0.8519 - val_loss: 0.7760 - val_accuracy: 0.7105\n",
            "Epoch 732 / 1000 - loss: 0.5918 - accuracy: 0.8519 - val_loss: 0.7741 - val_accuracy: 0.7105\n",
            "Epoch 733 / 1000 - loss: 0.5914 - accuracy: 0.8519 - val_loss: 0.7728 - val_accuracy: 0.7105\n",
            "Epoch 734 / 1000 - loss: 0.5910 - accuracy: 0.8519 - val_loss: 0.7726 - val_accuracy: 0.7105\n",
            "Epoch 735 / 1000 - loss: 0.5906 - accuracy: 0.8552 - val_loss: 0.7760 - val_accuracy: 0.7105\n",
            "Epoch 736 / 1000 - loss: 0.5902 - accuracy: 0.8519 - val_loss: 0.7747 - val_accuracy: 0.7105\n",
            "Epoch 737 / 1000 - loss: 0.5899 - accuracy: 0.8552 - val_loss: 0.7739 - val_accuracy: 0.7105\n",
            "Epoch 738 / 1000 - loss: 0.5895 - accuracy: 0.8519 - val_loss: 0.7761 - val_accuracy: 0.7105\n",
            "Epoch 739 / 1000 - loss: 0.5891 - accuracy: 0.8552 - val_loss: 0.7732 - val_accuracy: 0.7105\n",
            "Epoch 740 / 1000 - loss: 0.5892 - accuracy: 0.8552 - val_loss: 0.7704 - val_accuracy: 0.7105\n",
            "Epoch 741 / 1000 - loss: 0.5882 - accuracy: 0.8552 - val_loss: 0.7707 - val_accuracy: 0.7105\n",
            "Epoch 742 / 1000 - loss: 0.5880 - accuracy: 0.8552 - val_loss: 0.7721 - val_accuracy: 0.7105\n",
            "Epoch 743 / 1000 - loss: 0.5877 - accuracy: 0.8519 - val_loss: 0.7740 - val_accuracy: 0.7105\n",
            "Epoch 744 / 1000 - loss: 0.5872 - accuracy: 0.8552 - val_loss: 0.7712 - val_accuracy: 0.7105\n",
            "Epoch 745 / 1000 - loss: 0.5867 - accuracy: 0.8552 - val_loss: 0.7715 - val_accuracy: 0.7105\n",
            "Epoch 746 / 1000 - loss: 0.5863 - accuracy: 0.8552 - val_loss: 0.7707 - val_accuracy: 0.7105\n",
            "Epoch 747 / 1000 - loss: 0.5863 - accuracy: 0.8552 - val_loss: 0.7695 - val_accuracy: 0.7105\n",
            "Epoch 748 / 1000 - loss: 0.5859 - accuracy: 0.8552 - val_loss: 0.7691 - val_accuracy: 0.7105\n",
            "Epoch 749 / 1000 - loss: 0.5854 - accuracy: 0.8552 - val_loss: 0.7680 - val_accuracy: 0.7105\n",
            "Epoch 750 / 1000 - loss: 0.5850 - accuracy: 0.8552 - val_loss: 0.7671 - val_accuracy: 0.7105\n",
            "Epoch 751 / 1000 - loss: 0.5848 - accuracy: 0.8552 - val_loss: 0.7687 - val_accuracy: 0.7105\n",
            "Epoch 752 / 1000 - loss: 0.5841 - accuracy: 0.8552 - val_loss: 0.7690 - val_accuracy: 0.7105\n",
            "Epoch 753 / 1000 - loss: 0.5839 - accuracy: 0.8552 - val_loss: 0.7656 - val_accuracy: 0.7105\n",
            "Epoch 754 / 1000 - loss: 0.5834 - accuracy: 0.8586 - val_loss: 0.7644 - val_accuracy: 0.7105\n",
            "Epoch 755 / 1000 - loss: 0.5833 - accuracy: 0.8586 - val_loss: 0.7642 - val_accuracy: 0.7105\n",
            "Epoch 756 / 1000 - loss: 0.5826 - accuracy: 0.8552 - val_loss: 0.7659 - val_accuracy: 0.7105\n",
            "Epoch 757 / 1000 - loss: 0.5824 - accuracy: 0.8552 - val_loss: 0.7607 - val_accuracy: 0.7105\n",
            "Epoch 758 / 1000 - loss: 0.5823 - accuracy: 0.8586 - val_loss: 0.7635 - val_accuracy: 0.7105\n",
            "Epoch 759 / 1000 - loss: 0.5817 - accuracy: 0.8552 - val_loss: 0.7651 - val_accuracy: 0.7105\n",
            "Epoch 760 / 1000 - loss: 0.5814 - accuracy: 0.8552 - val_loss: 0.7663 - val_accuracy: 0.7105\n",
            "Epoch 761 / 1000 - loss: 0.5809 - accuracy: 0.8552 - val_loss: 0.7653 - val_accuracy: 0.7105\n",
            "Epoch 762 / 1000 - loss: 0.5804 - accuracy: 0.8586 - val_loss: 0.7617 - val_accuracy: 0.7105\n",
            "Epoch 763 / 1000 - loss: 0.5803 - accuracy: 0.8586 - val_loss: 0.7605 - val_accuracy: 0.7105\n",
            "Epoch 764 / 1000 - loss: 0.5797 - accuracy: 0.8586 - val_loss: 0.7637 - val_accuracy: 0.7105\n",
            "Epoch 765 / 1000 - loss: 0.5794 - accuracy: 0.8552 - val_loss: 0.7590 - val_accuracy: 0.7105\n",
            "Epoch 766 / 1000 - loss: 0.5789 - accuracy: 0.8586 - val_loss: 0.7586 - val_accuracy: 0.7105\n",
            "Epoch 767 / 1000 - loss: 0.5789 - accuracy: 0.8586 - val_loss: 0.7584 - val_accuracy: 0.7105\n",
            "Epoch 768 / 1000 - loss: 0.5782 - accuracy: 0.8620 - val_loss: 0.7575 - val_accuracy: 0.7105\n",
            "Epoch 769 / 1000 - loss: 0.5778 - accuracy: 0.8586 - val_loss: 0.7597 - val_accuracy: 0.7105\n",
            "Epoch 770 / 1000 - loss: 0.5776 - accuracy: 0.8586 - val_loss: 0.7571 - val_accuracy: 0.7105\n",
            "Epoch 771 / 1000 - loss: 0.5772 - accuracy: 0.8586 - val_loss: 0.7586 - val_accuracy: 0.7105\n",
            "Epoch 772 / 1000 - loss: 0.5767 - accuracy: 0.8586 - val_loss: 0.7603 - val_accuracy: 0.7105\n",
            "Epoch 773 / 1000 - loss: 0.5763 - accuracy: 0.8586 - val_loss: 0.7597 - val_accuracy: 0.7105\n",
            "Epoch 774 / 1000 - loss: 0.5760 - accuracy: 0.8586 - val_loss: 0.7599 - val_accuracy: 0.7105\n",
            "Epoch 775 / 1000 - loss: 0.5758 - accuracy: 0.8586 - val_loss: 0.7588 - val_accuracy: 0.7105\n",
            "Epoch 776 / 1000 - loss: 0.5753 - accuracy: 0.8586 - val_loss: 0.7594 - val_accuracy: 0.7105\n",
            "Epoch 777 / 1000 - loss: 0.5748 - accuracy: 0.8586 - val_loss: 0.7583 - val_accuracy: 0.7105\n",
            "Epoch 778 / 1000 - loss: 0.5745 - accuracy: 0.8586 - val_loss: 0.7555 - val_accuracy: 0.7105\n",
            "Epoch 779 / 1000 - loss: 0.5740 - accuracy: 0.8586 - val_loss: 0.7549 - val_accuracy: 0.7105\n",
            "Epoch 780 / 1000 - loss: 0.5737 - accuracy: 0.8586 - val_loss: 0.7554 - val_accuracy: 0.7105\n",
            "Epoch 781 / 1000 - loss: 0.5733 - accuracy: 0.8586 - val_loss: 0.7555 - val_accuracy: 0.7105\n",
            "Epoch 782 / 1000 - loss: 0.5733 - accuracy: 0.8586 - val_loss: 0.7549 - val_accuracy: 0.7105\n",
            "Epoch 783 / 1000 - loss: 0.5727 - accuracy: 0.8586 - val_loss: 0.7569 - val_accuracy: 0.7105\n",
            "Epoch 784 / 1000 - loss: 0.5724 - accuracy: 0.8586 - val_loss: 0.7546 - val_accuracy: 0.7105\n",
            "Epoch 785 / 1000 - loss: 0.5721 - accuracy: 0.8586 - val_loss: 0.7527 - val_accuracy: 0.7105\n",
            "Epoch 786 / 1000 - loss: 0.5717 - accuracy: 0.8620 - val_loss: 0.7513 - val_accuracy: 0.7105\n",
            "Epoch 787 / 1000 - loss: 0.5714 - accuracy: 0.8620 - val_loss: 0.7522 - val_accuracy: 0.7105\n",
            "Epoch 788 / 1000 - loss: 0.5709 - accuracy: 0.8620 - val_loss: 0.7513 - val_accuracy: 0.7105\n",
            "Epoch 789 / 1000 - loss: 0.5705 - accuracy: 0.8586 - val_loss: 0.7541 - val_accuracy: 0.7105\n",
            "Epoch 790 / 1000 - loss: 0.5702 - accuracy: 0.8620 - val_loss: 0.7504 - val_accuracy: 0.7105\n",
            "Epoch 791 / 1000 - loss: 0.5699 - accuracy: 0.8620 - val_loss: 0.7502 - val_accuracy: 0.7105\n",
            "Epoch 792 / 1000 - loss: 0.5694 - accuracy: 0.8586 - val_loss: 0.7524 - val_accuracy: 0.7105\n",
            "Epoch 793 / 1000 - loss: 0.5691 - accuracy: 0.8620 - val_loss: 0.7502 - val_accuracy: 0.7105\n",
            "Epoch 794 / 1000 - loss: 0.5692 - accuracy: 0.8822 - val_loss: 0.7461 - val_accuracy: 0.7368\n",
            "Epoch 795 / 1000 - loss: 0.5685 - accuracy: 0.8620 - val_loss: 0.7475 - val_accuracy: 0.7105\n",
            "Epoch 796 / 1000 - loss: 0.5681 - accuracy: 0.8620 - val_loss: 0.7487 - val_accuracy: 0.7105\n",
            "Epoch 797 / 1000 - loss: 0.5677 - accuracy: 0.8620 - val_loss: 0.7488 - val_accuracy: 0.7105\n",
            "Epoch 798 / 1000 - loss: 0.5675 - accuracy: 0.8620 - val_loss: 0.7459 - val_accuracy: 0.7368\n",
            "Epoch 799 / 1000 - loss: 0.5671 - accuracy: 0.8687 - val_loss: 0.7461 - val_accuracy: 0.7368\n",
            "Epoch 800 / 1000 - loss: 0.5665 - accuracy: 0.8620 - val_loss: 0.7461 - val_accuracy: 0.7105\n",
            "Epoch 801 / 1000 - loss: 0.5661 - accuracy: 0.8620 - val_loss: 0.7450 - val_accuracy: 0.7368\n",
            "Epoch 802 / 1000 - loss: 0.5659 - accuracy: 0.8620 - val_loss: 0.7447 - val_accuracy: 0.7368\n",
            "Epoch 803 / 1000 - loss: 0.5655 - accuracy: 0.8687 - val_loss: 0.7446 - val_accuracy: 0.7368\n",
            "Epoch 804 / 1000 - loss: 0.5653 - accuracy: 0.8620 - val_loss: 0.7489 - val_accuracy: 0.7105\n",
            "Epoch 805 / 1000 - loss: 0.5649 - accuracy: 0.8620 - val_loss: 0.7460 - val_accuracy: 0.7368\n",
            "Epoch 806 / 1000 - loss: 0.5643 - accuracy: 0.8620 - val_loss: 0.7462 - val_accuracy: 0.7105\n",
            "Epoch 807 / 1000 - loss: 0.5641 - accuracy: 0.8620 - val_loss: 0.7442 - val_accuracy: 0.7368\n",
            "Epoch 808 / 1000 - loss: 0.5638 - accuracy: 0.8653 - val_loss: 0.7415 - val_accuracy: 0.7368\n",
            "Epoch 809 / 1000 - loss: 0.5633 - accuracy: 0.8620 - val_loss: 0.7449 - val_accuracy: 0.7105\n",
            "Epoch 810 / 1000 - loss: 0.5633 - accuracy: 0.8620 - val_loss: 0.7457 - val_accuracy: 0.7105\n",
            "Epoch 811 / 1000 - loss: 0.5627 - accuracy: 0.8754 - val_loss: 0.7428 - val_accuracy: 0.7368\n",
            "Epoch 812 / 1000 - loss: 0.5626 - accuracy: 0.8620 - val_loss: 0.7443 - val_accuracy: 0.7105\n",
            "Epoch 813 / 1000 - loss: 0.5619 - accuracy: 0.8620 - val_loss: 0.7410 - val_accuracy: 0.7368\n",
            "Epoch 814 / 1000 - loss: 0.5619 - accuracy: 0.8620 - val_loss: 0.7394 - val_accuracy: 0.7368\n",
            "Epoch 815 / 1000 - loss: 0.5617 - accuracy: 0.8653 - val_loss: 0.7394 - val_accuracy: 0.7368\n",
            "Epoch 816 / 1000 - loss: 0.5609 - accuracy: 0.8620 - val_loss: 0.7425 - val_accuracy: 0.7368\n",
            "Epoch 817 / 1000 - loss: 0.5606 - accuracy: 0.8653 - val_loss: 0.7388 - val_accuracy: 0.7368\n",
            "Epoch 818 / 1000 - loss: 0.5604 - accuracy: 0.8620 - val_loss: 0.7399 - val_accuracy: 0.7368\n",
            "Epoch 819 / 1000 - loss: 0.5602 - accuracy: 0.8788 - val_loss: 0.7374 - val_accuracy: 0.7368\n",
            "Epoch 820 / 1000 - loss: 0.5596 - accuracy: 0.8754 - val_loss: 0.7399 - val_accuracy: 0.7368\n",
            "Epoch 821 / 1000 - loss: 0.5593 - accuracy: 0.8620 - val_loss: 0.7383 - val_accuracy: 0.7368\n",
            "Epoch 822 / 1000 - loss: 0.5590 - accuracy: 0.8620 - val_loss: 0.7413 - val_accuracy: 0.7368\n",
            "Epoch 823 / 1000 - loss: 0.5586 - accuracy: 0.8620 - val_loss: 0.7397 - val_accuracy: 0.7368\n",
            "Epoch 824 / 1000 - loss: 0.5582 - accuracy: 0.8788 - val_loss: 0.7395 - val_accuracy: 0.7368\n",
            "Epoch 825 / 1000 - loss: 0.5581 - accuracy: 0.8687 - val_loss: 0.7340 - val_accuracy: 0.7368\n",
            "Epoch 826 / 1000 - loss: 0.5576 - accuracy: 0.8687 - val_loss: 0.7333 - val_accuracy: 0.7368\n",
            "Epoch 827 / 1000 - loss: 0.5572 - accuracy: 0.8788 - val_loss: 0.7390 - val_accuracy: 0.7368\n",
            "Epoch 828 / 1000 - loss: 0.5569 - accuracy: 0.8754 - val_loss: 0.7350 - val_accuracy: 0.7368\n",
            "Epoch 829 / 1000 - loss: 0.5566 - accuracy: 0.8788 - val_loss: 0.7335 - val_accuracy: 0.7368\n",
            "Epoch 830 / 1000 - loss: 0.5562 - accuracy: 0.8620 - val_loss: 0.7363 - val_accuracy: 0.7368\n",
            "Epoch 831 / 1000 - loss: 0.5559 - accuracy: 0.8754 - val_loss: 0.7339 - val_accuracy: 0.7368\n",
            "Epoch 832 / 1000 - loss: 0.5555 - accuracy: 0.8788 - val_loss: 0.7338 - val_accuracy: 0.7368\n",
            "Epoch 833 / 1000 - loss: 0.5552 - accuracy: 0.8822 - val_loss: 0.7335 - val_accuracy: 0.7368\n",
            "Epoch 834 / 1000 - loss: 0.5555 - accuracy: 0.8822 - val_loss: 0.7284 - val_accuracy: 0.7368\n",
            "Epoch 835 / 1000 - loss: 0.5549 - accuracy: 0.8822 - val_loss: 0.7307 - val_accuracy: 0.7368\n",
            "Epoch 836 / 1000 - loss: 0.5541 - accuracy: 0.8822 - val_loss: 0.7324 - val_accuracy: 0.7368\n",
            "Epoch 837 / 1000 - loss: 0.5539 - accuracy: 0.8788 - val_loss: 0.7346 - val_accuracy: 0.7368\n",
            "Epoch 838 / 1000 - loss: 0.5535 - accuracy: 0.8653 - val_loss: 0.7318 - val_accuracy: 0.7368\n",
            "Epoch 839 / 1000 - loss: 0.5531 - accuracy: 0.8620 - val_loss: 0.7331 - val_accuracy: 0.7368\n",
            "Epoch 840 / 1000 - loss: 0.5528 - accuracy: 0.8653 - val_loss: 0.7344 - val_accuracy: 0.7368\n",
            "Epoch 841 / 1000 - loss: 0.5526 - accuracy: 0.8653 - val_loss: 0.7304 - val_accuracy: 0.7368\n",
            "Epoch 842 / 1000 - loss: 0.5520 - accuracy: 0.8687 - val_loss: 0.7294 - val_accuracy: 0.7368\n",
            "Epoch 843 / 1000 - loss: 0.5518 - accuracy: 0.8687 - val_loss: 0.7296 - val_accuracy: 0.7368\n",
            "Epoch 844 / 1000 - loss: 0.5515 - accuracy: 0.8653 - val_loss: 0.7311 - val_accuracy: 0.7368\n",
            "Epoch 845 / 1000 - loss: 0.5513 - accuracy: 0.8653 - val_loss: 0.7314 - val_accuracy: 0.7368\n",
            "Epoch 846 / 1000 - loss: 0.5508 - accuracy: 0.8653 - val_loss: 0.7308 - val_accuracy: 0.7368\n",
            "Epoch 847 / 1000 - loss: 0.5507 - accuracy: 0.8620 - val_loss: 0.7284 - val_accuracy: 0.7368\n",
            "Epoch 848 / 1000 - loss: 0.5502 - accuracy: 0.8687 - val_loss: 0.7270 - val_accuracy: 0.7368\n",
            "Epoch 849 / 1000 - loss: 0.5499 - accuracy: 0.8822 - val_loss: 0.7255 - val_accuracy: 0.7368\n",
            "Epoch 850 / 1000 - loss: 0.5496 - accuracy: 0.8822 - val_loss: 0.7260 - val_accuracy: 0.7368\n",
            "Epoch 851 / 1000 - loss: 0.5498 - accuracy: 0.8721 - val_loss: 0.7210 - val_accuracy: 0.7368\n",
            "Epoch 852 / 1000 - loss: 0.5487 - accuracy: 0.8687 - val_loss: 0.7251 - val_accuracy: 0.7368\n",
            "Epoch 853 / 1000 - loss: 0.5488 - accuracy: 0.8653 - val_loss: 0.7303 - val_accuracy: 0.7368\n",
            "Epoch 854 / 1000 - loss: 0.5482 - accuracy: 0.8653 - val_loss: 0.7281 - val_accuracy: 0.7368\n",
            "Epoch 855 / 1000 - loss: 0.5479 - accuracy: 0.8687 - val_loss: 0.7234 - val_accuracy: 0.7368\n",
            "Epoch 856 / 1000 - loss: 0.5478 - accuracy: 0.8822 - val_loss: 0.7247 - val_accuracy: 0.7368\n",
            "Epoch 857 / 1000 - loss: 0.5471 - accuracy: 0.8822 - val_loss: 0.7234 - val_accuracy: 0.7368\n",
            "Epoch 858 / 1000 - loss: 0.5469 - accuracy: 0.8687 - val_loss: 0.7222 - val_accuracy: 0.7368\n",
            "Epoch 859 / 1000 - loss: 0.5464 - accuracy: 0.8822 - val_loss: 0.7226 - val_accuracy: 0.7368\n",
            "Epoch 860 / 1000 - loss: 0.5459 - accuracy: 0.8788 - val_loss: 0.7242 - val_accuracy: 0.7368\n",
            "Epoch 861 / 1000 - loss: 0.5458 - accuracy: 0.8788 - val_loss: 0.7226 - val_accuracy: 0.7368\n",
            "Epoch 862 / 1000 - loss: 0.5455 - accuracy: 0.8822 - val_loss: 0.7223 - val_accuracy: 0.7368\n",
            "Epoch 863 / 1000 - loss: 0.5451 - accuracy: 0.8822 - val_loss: 0.7200 - val_accuracy: 0.7368\n",
            "Epoch 864 / 1000 - loss: 0.5451 - accuracy: 0.8687 - val_loss: 0.7204 - val_accuracy: 0.7368\n",
            "Epoch 865 / 1000 - loss: 0.5447 - accuracy: 0.8822 - val_loss: 0.7196 - val_accuracy: 0.7368\n",
            "Epoch 866 / 1000 - loss: 0.5443 - accuracy: 0.8653 - val_loss: 0.7236 - val_accuracy: 0.7368\n",
            "Epoch 867 / 1000 - loss: 0.5440 - accuracy: 0.8653 - val_loss: 0.7242 - val_accuracy: 0.7368\n",
            "Epoch 868 / 1000 - loss: 0.5435 - accuracy: 0.8653 - val_loss: 0.7231 - val_accuracy: 0.7368\n",
            "Epoch 869 / 1000 - loss: 0.5438 - accuracy: 0.8889 - val_loss: 0.7171 - val_accuracy: 0.7368\n",
            "Epoch 870 / 1000 - loss: 0.5430 - accuracy: 0.8653 - val_loss: 0.7227 - val_accuracy: 0.7368\n",
            "Epoch 871 / 1000 - loss: 0.5424 - accuracy: 0.8788 - val_loss: 0.7220 - val_accuracy: 0.7368\n",
            "Epoch 872 / 1000 - loss: 0.5419 - accuracy: 0.8788 - val_loss: 0.7217 - val_accuracy: 0.7368\n",
            "Epoch 873 / 1000 - loss: 0.5418 - accuracy: 0.8855 - val_loss: 0.7199 - val_accuracy: 0.7368\n",
            "Epoch 874 / 1000 - loss: 0.5414 - accuracy: 0.8855 - val_loss: 0.7153 - val_accuracy: 0.7368\n",
            "Epoch 875 / 1000 - loss: 0.5412 - accuracy: 0.8855 - val_loss: 0.7191 - val_accuracy: 0.7368\n",
            "Epoch 876 / 1000 - loss: 0.5410 - accuracy: 0.8855 - val_loss: 0.7193 - val_accuracy: 0.7368\n",
            "Epoch 877 / 1000 - loss: 0.5406 - accuracy: 0.8822 - val_loss: 0.7173 - val_accuracy: 0.7368\n",
            "Epoch 878 / 1000 - loss: 0.5400 - accuracy: 0.8855 - val_loss: 0.7178 - val_accuracy: 0.7368\n",
            "Epoch 879 / 1000 - loss: 0.5399 - accuracy: 0.8822 - val_loss: 0.7163 - val_accuracy: 0.7368\n",
            "Epoch 880 / 1000 - loss: 0.5394 - accuracy: 0.8855 - val_loss: 0.7181 - val_accuracy: 0.7368\n",
            "Epoch 881 / 1000 - loss: 0.5392 - accuracy: 0.8653 - val_loss: 0.7204 - val_accuracy: 0.7368\n",
            "Epoch 882 / 1000 - loss: 0.5390 - accuracy: 0.8788 - val_loss: 0.7220 - val_accuracy: 0.7368\n",
            "Epoch 883 / 1000 - loss: 0.5387 - accuracy: 0.8855 - val_loss: 0.7161 - val_accuracy: 0.7368\n",
            "Epoch 884 / 1000 - loss: 0.5387 - accuracy: 0.8754 - val_loss: 0.7132 - val_accuracy: 0.7368\n",
            "Epoch 885 / 1000 - loss: 0.5380 - accuracy: 0.8889 - val_loss: 0.7105 - val_accuracy: 0.7368\n",
            "Epoch 886 / 1000 - loss: 0.5378 - accuracy: 0.8889 - val_loss: 0.7102 - val_accuracy: 0.7368\n",
            "Epoch 887 / 1000 - loss: 0.5372 - accuracy: 0.8822 - val_loss: 0.7166 - val_accuracy: 0.7368\n",
            "Epoch 888 / 1000 - loss: 0.5369 - accuracy: 0.8855 - val_loss: 0.7152 - val_accuracy: 0.7368\n",
            "Epoch 889 / 1000 - loss: 0.5367 - accuracy: 0.8788 - val_loss: 0.7160 - val_accuracy: 0.7368\n",
            "Epoch 890 / 1000 - loss: 0.5364 - accuracy: 0.8889 - val_loss: 0.7106 - val_accuracy: 0.7368\n",
            "Epoch 891 / 1000 - loss: 0.5358 - accuracy: 0.8822 - val_loss: 0.7155 - val_accuracy: 0.7368\n",
            "Epoch 892 / 1000 - loss: 0.5360 - accuracy: 0.8889 - val_loss: 0.7102 - val_accuracy: 0.7368\n",
            "Epoch 893 / 1000 - loss: 0.5363 - accuracy: 0.8889 - val_loss: 0.7044 - val_accuracy: 0.7368\n",
            "Epoch 894 / 1000 - loss: 0.5350 - accuracy: 0.8788 - val_loss: 0.7124 - val_accuracy: 0.7368\n",
            "Epoch 895 / 1000 - loss: 0.5348 - accuracy: 0.8822 - val_loss: 0.7151 - val_accuracy: 0.7368\n",
            "Epoch 896 / 1000 - loss: 0.5348 - accuracy: 0.8889 - val_loss: 0.7080 - val_accuracy: 0.7368\n",
            "Epoch 897 / 1000 - loss: 0.5341 - accuracy: 0.8788 - val_loss: 0.7108 - val_accuracy: 0.7368\n",
            "Epoch 898 / 1000 - loss: 0.5343 - accuracy: 0.8923 - val_loss: 0.7090 - val_accuracy: 0.7368\n",
            "Epoch 899 / 1000 - loss: 0.5337 - accuracy: 0.8889 - val_loss: 0.7062 - val_accuracy: 0.7368\n",
            "Epoch 900 / 1000 - loss: 0.5331 - accuracy: 0.8923 - val_loss: 0.7096 - val_accuracy: 0.7368\n",
            "Epoch 901 / 1000 - loss: 0.5331 - accuracy: 0.8923 - val_loss: 0.7073 - val_accuracy: 0.7368\n",
            "Epoch 902 / 1000 - loss: 0.5328 - accuracy: 0.8754 - val_loss: 0.7102 - val_accuracy: 0.7368\n",
            "Epoch 903 / 1000 - loss: 0.5324 - accuracy: 0.8788 - val_loss: 0.7070 - val_accuracy: 0.7368\n",
            "Epoch 904 / 1000 - loss: 0.5320 - accuracy: 0.8788 - val_loss: 0.7068 - val_accuracy: 0.7368\n",
            "Epoch 905 / 1000 - loss: 0.5318 - accuracy: 0.8923 - val_loss: 0.7074 - val_accuracy: 0.7368\n",
            "Epoch 906 / 1000 - loss: 0.5313 - accuracy: 0.8855 - val_loss: 0.7116 - val_accuracy: 0.7368\n",
            "Epoch 907 / 1000 - loss: 0.5315 - accuracy: 0.8923 - val_loss: 0.7064 - val_accuracy: 0.7368\n",
            "Epoch 908 / 1000 - loss: 0.5307 - accuracy: 0.8855 - val_loss: 0.7106 - val_accuracy: 0.7368\n",
            "Epoch 909 / 1000 - loss: 0.5303 - accuracy: 0.8923 - val_loss: 0.7068 - val_accuracy: 0.7368\n",
            "Epoch 910 / 1000 - loss: 0.5301 - accuracy: 0.8923 - val_loss: 0.7080 - val_accuracy: 0.7368\n",
            "Epoch 911 / 1000 - loss: 0.5297 - accuracy: 0.8889 - val_loss: 0.7093 - val_accuracy: 0.7368\n",
            "Epoch 912 / 1000 - loss: 0.5293 - accuracy: 0.8923 - val_loss: 0.7040 - val_accuracy: 0.7368\n",
            "Epoch 913 / 1000 - loss: 0.5292 - accuracy: 0.8923 - val_loss: 0.7056 - val_accuracy: 0.7368\n",
            "Epoch 914 / 1000 - loss: 0.5289 - accuracy: 0.8923 - val_loss: 0.7062 - val_accuracy: 0.7368\n",
            "Epoch 915 / 1000 - loss: 0.5286 - accuracy: 0.8923 - val_loss: 0.7057 - val_accuracy: 0.7368\n",
            "Epoch 916 / 1000 - loss: 0.5284 - accuracy: 0.8923 - val_loss: 0.7060 - val_accuracy: 0.7368\n",
            "Epoch 917 / 1000 - loss: 0.5279 - accuracy: 0.8889 - val_loss: 0.7054 - val_accuracy: 0.7368\n",
            "Epoch 918 / 1000 - loss: 0.5275 - accuracy: 0.8923 - val_loss: 0.7014 - val_accuracy: 0.7368\n",
            "Epoch 919 / 1000 - loss: 0.5273 - accuracy: 0.8923 - val_loss: 0.7007 - val_accuracy: 0.7368\n",
            "Epoch 920 / 1000 - loss: 0.5270 - accuracy: 0.8923 - val_loss: 0.7022 - val_accuracy: 0.7368\n",
            "Epoch 921 / 1000 - loss: 0.5268 - accuracy: 0.8923 - val_loss: 0.7016 - val_accuracy: 0.7368\n",
            "Epoch 922 / 1000 - loss: 0.5263 - accuracy: 0.8889 - val_loss: 0.7066 - val_accuracy: 0.7368\n",
            "Epoch 923 / 1000 - loss: 0.5258 - accuracy: 0.8923 - val_loss: 0.7022 - val_accuracy: 0.7368\n",
            "Epoch 924 / 1000 - loss: 0.5257 - accuracy: 0.8923 - val_loss: 0.7024 - val_accuracy: 0.7368\n",
            "Epoch 925 / 1000 - loss: 0.5253 - accuracy: 0.8923 - val_loss: 0.6996 - val_accuracy: 0.7368\n",
            "Epoch 926 / 1000 - loss: 0.5250 - accuracy: 0.8923 - val_loss: 0.7022 - val_accuracy: 0.7368\n",
            "Epoch 927 / 1000 - loss: 0.5247 - accuracy: 0.8889 - val_loss: 0.7030 - val_accuracy: 0.7368\n",
            "Epoch 928 / 1000 - loss: 0.5245 - accuracy: 0.8855 - val_loss: 0.7072 - val_accuracy: 0.7368\n",
            "Epoch 929 / 1000 - loss: 0.5241 - accuracy: 0.8923 - val_loss: 0.6984 - val_accuracy: 0.7368\n",
            "Epoch 930 / 1000 - loss: 0.5241 - accuracy: 0.8923 - val_loss: 0.6962 - val_accuracy: 0.7368\n",
            "Epoch 931 / 1000 - loss: 0.5239 - accuracy: 0.8923 - val_loss: 0.6981 - val_accuracy: 0.7368\n",
            "Epoch 932 / 1000 - loss: 0.5235 - accuracy: 0.8923 - val_loss: 0.6964 - val_accuracy: 0.7368\n",
            "Epoch 933 / 1000 - loss: 0.5231 - accuracy: 0.8889 - val_loss: 0.6999 - val_accuracy: 0.7368\n",
            "Epoch 934 / 1000 - loss: 0.5225 - accuracy: 0.8889 - val_loss: 0.7027 - val_accuracy: 0.7368\n",
            "Epoch 935 / 1000 - loss: 0.5225 - accuracy: 0.8923 - val_loss: 0.6953 - val_accuracy: 0.7368\n",
            "Epoch 936 / 1000 - loss: 0.5220 - accuracy: 0.8923 - val_loss: 0.6945 - val_accuracy: 0.7368\n",
            "Epoch 937 / 1000 - loss: 0.5218 - accuracy: 0.8889 - val_loss: 0.6990 - val_accuracy: 0.7368\n",
            "Epoch 938 / 1000 - loss: 0.5214 - accuracy: 0.8889 - val_loss: 0.7001 - val_accuracy: 0.7368\n",
            "Epoch 939 / 1000 - loss: 0.5213 - accuracy: 0.8923 - val_loss: 0.6943 - val_accuracy: 0.7368\n",
            "Epoch 940 / 1000 - loss: 0.5211 - accuracy: 0.8923 - val_loss: 0.6976 - val_accuracy: 0.7368\n",
            "Epoch 941 / 1000 - loss: 0.5204 - accuracy: 0.8923 - val_loss: 0.6958 - val_accuracy: 0.7368\n",
            "Epoch 942 / 1000 - loss: 0.5202 - accuracy: 0.8923 - val_loss: 0.6957 - val_accuracy: 0.7368\n",
            "Epoch 943 / 1000 - loss: 0.5203 - accuracy: 0.8788 - val_loss: 0.6946 - val_accuracy: 0.7368\n",
            "Epoch 944 / 1000 - loss: 0.5197 - accuracy: 0.8889 - val_loss: 0.6962 - val_accuracy: 0.7368\n",
            "Epoch 945 / 1000 - loss: 0.5193 - accuracy: 0.8923 - val_loss: 0.6982 - val_accuracy: 0.7368\n",
            "Epoch 946 / 1000 - loss: 0.5192 - accuracy: 0.8889 - val_loss: 0.6972 - val_accuracy: 0.7368\n",
            "Epoch 947 / 1000 - loss: 0.5188 - accuracy: 0.8923 - val_loss: 0.6954 - val_accuracy: 0.7368\n",
            "Epoch 948 / 1000 - loss: 0.5185 - accuracy: 0.8889 - val_loss: 0.6952 - val_accuracy: 0.7368\n",
            "Epoch 949 / 1000 - loss: 0.5189 - accuracy: 0.8889 - val_loss: 0.6872 - val_accuracy: 0.7368\n",
            "Epoch 950 / 1000 - loss: 0.5180 - accuracy: 0.8788 - val_loss: 0.6891 - val_accuracy: 0.7368\n",
            "Epoch 951 / 1000 - loss: 0.5177 - accuracy: 0.8923 - val_loss: 0.6897 - val_accuracy: 0.7368\n",
            "Epoch 952 / 1000 - loss: 0.5172 - accuracy: 0.8923 - val_loss: 0.6904 - val_accuracy: 0.7368\n",
            "Epoch 953 / 1000 - loss: 0.5168 - accuracy: 0.8923 - val_loss: 0.6912 - val_accuracy: 0.7368\n",
            "Epoch 954 / 1000 - loss: 0.5164 - accuracy: 0.8923 - val_loss: 0.6934 - val_accuracy: 0.7368\n",
            "Epoch 955 / 1000 - loss: 0.5166 - accuracy: 0.8923 - val_loss: 0.6913 - val_accuracy: 0.7368\n",
            "Epoch 956 / 1000 - loss: 0.5165 - accuracy: 0.8889 - val_loss: 0.6975 - val_accuracy: 0.7368\n",
            "Epoch 957 / 1000 - loss: 0.5161 - accuracy: 0.8923 - val_loss: 0.6894 - val_accuracy: 0.7368\n",
            "Epoch 958 / 1000 - loss: 0.5155 - accuracy: 0.8923 - val_loss: 0.6895 - val_accuracy: 0.7368\n",
            "Epoch 959 / 1000 - loss: 0.5152 - accuracy: 0.8923 - val_loss: 0.6897 - val_accuracy: 0.7368\n",
            "Epoch 960 / 1000 - loss: 0.5150 - accuracy: 0.8923 - val_loss: 0.6853 - val_accuracy: 0.7368\n",
            "Epoch 961 / 1000 - loss: 0.5147 - accuracy: 0.8923 - val_loss: 0.6855 - val_accuracy: 0.7368\n",
            "Epoch 962 / 1000 - loss: 0.5142 - accuracy: 0.8923 - val_loss: 0.6890 - val_accuracy: 0.7368\n",
            "Epoch 963 / 1000 - loss: 0.5140 - accuracy: 0.8923 - val_loss: 0.6878 - val_accuracy: 0.7368\n",
            "Epoch 964 / 1000 - loss: 0.5137 - accuracy: 0.8923 - val_loss: 0.6875 - val_accuracy: 0.7368\n",
            "Epoch 965 / 1000 - loss: 0.5134 - accuracy: 0.8923 - val_loss: 0.6924 - val_accuracy: 0.7368\n",
            "Epoch 966 / 1000 - loss: 0.5132 - accuracy: 0.8923 - val_loss: 0.6865 - val_accuracy: 0.7368\n",
            "Epoch 967 / 1000 - loss: 0.5129 - accuracy: 0.8923 - val_loss: 0.6826 - val_accuracy: 0.7368\n",
            "Epoch 968 / 1000 - loss: 0.5123 - accuracy: 0.8923 - val_loss: 0.6886 - val_accuracy: 0.7368\n",
            "Epoch 969 / 1000 - loss: 0.5122 - accuracy: 0.8923 - val_loss: 0.6845 - val_accuracy: 0.7368\n",
            "Epoch 970 / 1000 - loss: 0.5121 - accuracy: 0.8923 - val_loss: 0.6874 - val_accuracy: 0.7368\n",
            "Epoch 971 / 1000 - loss: 0.5115 - accuracy: 0.8923 - val_loss: 0.6855 - val_accuracy: 0.7368\n",
            "Epoch 972 / 1000 - loss: 0.5114 - accuracy: 0.8923 - val_loss: 0.6909 - val_accuracy: 0.7368\n",
            "Epoch 973 / 1000 - loss: 0.5109 - accuracy: 0.8923 - val_loss: 0.6868 - val_accuracy: 0.7368\n",
            "Epoch 974 / 1000 - loss: 0.5109 - accuracy: 0.8923 - val_loss: 0.6848 - val_accuracy: 0.7368\n",
            "Epoch 975 / 1000 - loss: 0.5107 - accuracy: 0.8788 - val_loss: 0.6856 - val_accuracy: 0.7368\n",
            "Epoch 976 / 1000 - loss: 0.5105 - accuracy: 0.8923 - val_loss: 0.6796 - val_accuracy: 0.7368\n",
            "Epoch 977 / 1000 - loss: 0.5100 - accuracy: 0.8923 - val_loss: 0.6808 - val_accuracy: 0.7368\n",
            "Epoch 978 / 1000 - loss: 0.5097 - accuracy: 0.8923 - val_loss: 0.6848 - val_accuracy: 0.7368\n",
            "Epoch 979 / 1000 - loss: 0.5093 - accuracy: 0.8923 - val_loss: 0.6800 - val_accuracy: 0.7368\n",
            "Epoch 980 / 1000 - loss: 0.5095 - accuracy: 0.8956 - val_loss: 0.6765 - val_accuracy: 0.7368\n",
            "Epoch 981 / 1000 - loss: 0.5090 - accuracy: 0.8923 - val_loss: 0.6783 - val_accuracy: 0.7368\n",
            "Epoch 982 / 1000 - loss: 0.5083 - accuracy: 0.8923 - val_loss: 0.6836 - val_accuracy: 0.7368\n",
            "Epoch 983 / 1000 - loss: 0.5082 - accuracy: 0.8923 - val_loss: 0.6861 - val_accuracy: 0.7368\n",
            "Epoch 984 / 1000 - loss: 0.5078 - accuracy: 0.8923 - val_loss: 0.6822 - val_accuracy: 0.7368\n",
            "Epoch 985 / 1000 - loss: 0.5075 - accuracy: 0.8923 - val_loss: 0.6860 - val_accuracy: 0.7368\n",
            "Epoch 986 / 1000 - loss: 0.5072 - accuracy: 0.8923 - val_loss: 0.6823 - val_accuracy: 0.7368\n",
            "Epoch 987 / 1000 - loss: 0.5069 - accuracy: 0.8923 - val_loss: 0.6827 - val_accuracy: 0.7368\n",
            "Epoch 988 / 1000 - loss: 0.5067 - accuracy: 0.8923 - val_loss: 0.6793 - val_accuracy: 0.7368\n",
            "Epoch 989 / 1000 - loss: 0.5065 - accuracy: 0.8923 - val_loss: 0.6802 - val_accuracy: 0.7368\n",
            "Epoch 990 / 1000 - loss: 0.5063 - accuracy: 0.8889 - val_loss: 0.6854 - val_accuracy: 0.7368\n",
            "Epoch 991 / 1000 - loss: 0.5059 - accuracy: 0.8923 - val_loss: 0.6829 - val_accuracy: 0.7368\n",
            "Epoch 992 / 1000 - loss: 0.5057 - accuracy: 0.8923 - val_loss: 0.6807 - val_accuracy: 0.7368\n",
            "Epoch 993 / 1000 - loss: 0.5054 - accuracy: 0.8923 - val_loss: 0.6828 - val_accuracy: 0.7368\n",
            "Epoch 994 / 1000 - loss: 0.5049 - accuracy: 0.8923 - val_loss: 0.6798 - val_accuracy: 0.7368\n",
            "Epoch 995 / 1000 - loss: 0.5048 - accuracy: 0.8923 - val_loss: 0.6777 - val_accuracy: 0.7368\n",
            "Epoch 996 / 1000 - loss: 0.5046 - accuracy: 0.8923 - val_loss: 0.6812 - val_accuracy: 0.7368\n",
            "Epoch 997 / 1000 - loss: 0.5042 - accuracy: 0.8923 - val_loss: 0.6801 - val_accuracy: 0.7368\n",
            "Epoch 998 / 1000 - loss: 0.5037 - accuracy: 0.8923 - val_loss: 0.6789 - val_accuracy: 0.7368\n",
            "Epoch 999 / 1000 - loss: 0.5045 - accuracy: 0.8990 - val_loss: 0.6699 - val_accuracy: 0.7368\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1152x720 with 4 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "id": "LeMEfwATIG8R",
        "outputId": "2d8c782e-913a-425f-b6c6-d14f26171196"
      },
      "source": [
        "plt.title('Weights change')\n",
        "plt.ylabel('Weights')\n",
        "plt.xlabel('Iterations')\n",
        "plt.plot(np.array(all_w).reshape((len(all_w),all_w[0].shape[0]*all_w[0].shape[1])))\n",
        "plt.show()"
      ],
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "U0doFXubIHtJ"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}