Detect2class.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "ajHeD1nTIB1j"
   },
   "source": [
    "# 1DCNN code for damage detection"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "Mjdojg-1ILA4"
   },
   "source": [
    "## Importing APIs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "IkeIr5YWbktM"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "#imports\n",
    "import os\n",
    "import zipfile\n",
    "%matplotlib inline\n",
    "import random \n",
    "import tensorflow as tf\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "from tensorflow.keras.models import Sequential\n",
    "from tensorflow.keras.layers import Dense, Flatten, Activation, Conv1D, MaxPooling1D, Dropout, Lambda, BatchNormalization\n",
    "from tensorflow.keras.optimizers import SGD, Adam, RMSprop\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn import preprocessing\n",
    "from keras.utils import to_categorical,plot_model\n",
    "from sklearn.metrics import confusion_matrix, classification_report\n",
    "from scipy import stats\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "from tensorflow.compat.v1 import ConfigProto\n",
    "from tensorflow.compat.v1 import InteractiveSession\n",
    "config = ConfigProto()\n",
    "config.gpu_options.allow_growth = True\n",
    "session = InteractiveSession(config=config)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "7ildUOZpINr6"
   },
   "source": [
    "## Callbacks"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "H-lBgT16UzvV"
   },
   "outputs": [],
   "source": [
    "class myCallback(tf.keras.callbacks.Callback):\n",
    "  def on_epoch_end(self, epoch, logs={}):\n",
    "    if (logs.get('val_acc')>0.99) and (logs.get('acc')>0.99) and (logs.get('val_loss')<0.05) and (logs.get('loss')<0.05):\n",
    "      print(\"\\nReached perfect accuracy so cancelling training!\")\n",
    "      self.model.stop_training = True\n",
    "\n",
    "epoch_schedule = myCallback()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "50uuiQSSyqyj"
   },
   "outputs": [],
   "source": [
    "lr_schedule = tf.keras.callbacks.LearningRateScheduler(\n",
    "    lambda epoch: 1e-6 * 10**(epoch / 5))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Importing data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "gPvQTFpd5iZv"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(22176, 13108)\n"
     ]
    }
   ],
   "source": [
    "df_base = pd.read_csv('E:/PhD-MSR/OGWdataset/Guided_wave_basic_measurement_data/dataset/AllDatasets/combinedata/SHM_Baseline-1_filter.txt',header=None)\n",
    "print(df_base.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "gD3j7iBZ5jN1"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(22176, 13108)\n"
     ]
    }
   ],
   "source": [
    "df_dam = pd.read_csv('E:/PhD-MSR/OGWdataset/Guided_wave_basic_measurement_data/dataset/AllDatasets/combinedata/SHM_Damage_filter.txt',header=None)\n",
    "print(df_dam.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_base = np.array(df_base)\n",
    "df_dam = np.array(df_dam)\n",
    "df = np.concatenate([df_base,df_dam], axis=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Function for RMSD (Root Mean Squared Difference)\n",
    "def RMSD(base,dam):\n",
    "    num = np.sum((base-dam)**2,axis=1)\n",
    "    den = np.sum(base**2,axis=1)\n",
    "    rmsd = np.sqrt(np.divide(num,den))\n",
    "    return rmsd"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Call RMSD function\n",
    "rmsd = RMSD(df_base,df_dam)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "t9btG2kdIU6H"
   },
   "source": [
    "## Plotting dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(13108,)\n"
     ]
    }
   ],
   "source": [
    "seqlen = df.shape[1]\n",
    "dT = 1e-7\n",
    "time = np.arange(0,(seqlen)*dT,dT)\n",
    "print(time.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "sigwindow = 1e-3\n",
    "idx1 = 1000\n",
    "idx2 = 13108"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "14727\n",
      "0.03452000015664699\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "Text(0, 0.5, 'Norm Amplitude')"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(12,4))\n",
    "nn = random.randint(1,df_base.shape[0])\n",
    "#nn=51\n",
    "print(nn)\n",
    "print(rmsd[nn])\n",
    "plt.plot(time[idx1:idx2], df_base[nn,idx1: idx2])\n",
    "plt.plot(time[idx1:idx2], df_dam[nn,idx1: idx2])\n",
    "plt.title('Training example',fontsize=25)\n",
    "plt.legend(['base', 'damage'], loc='upper right',fontsize=18)\n",
    "plt.xticks(fontsize=25)\n",
    "plt.yticks(fontsize=25)\n",
    "plt.xlabel('Time is seconds',fontsize=25)\n",
    "plt.ylabel('Norm Amplitude',fontsize=25)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "CXy8CBXkIY0p"
   },
   "source": [
    "## Deciding Features and Labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(3024,)\n",
      "[    6     7     8 ... 21565 21566 21567]\n"
     ]
    }
   ],
   "source": [
    "# Using only transmission channels\n",
    "sch1 = np.arange(6,11+1,1)                   # sub channels with 1 as actuator and other as sensors\n",
    "sch2 = np.arange(16,21+1,1)                  # sub channels with 2 as actuator and other as sensors\n",
    "sch3 = np.arange(25,30+1,1)\n",
    "sch4 = np.arange(33,38+1,1)\n",
    "sch5 = np.arange(40,45+1,1)\n",
    "sch6 = np.arange(46,51+1,1)\n",
    "transch = np.concatenate([sch1,sch2,sch3,sch4,sch5,sch6], axis=0)\n",
    "freqtransch = np.concatenate([transch,transch+66,transch+66*2], axis=0)\n",
    "\n",
    "A = []\n",
    "for i in range(1,28,1):                       # we are incorporating all damages, i = 1 here means D2\n",
    "    damtransch = 792*i + freqtransch      \n",
    "    A.append(damtransch)\n",
    "\n",
    "A = np.concatenate(A)                         # concatenate all appended vectors\n",
    "B = np.concatenate([freqtransch,A],axis=0)    # concatenate D1 with D2 to D28\n",
    "print(B.shape)\n",
    "print(B)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "#del df_base, df_dam"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Using only 'B' training samples\n",
    "df_base = df_base[B, idx1:idx2]\n",
    "df_dam = df_dam[B, idx1:idx2]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(6048, 12108)\n"
     ]
    }
   ],
   "source": [
    "# Concatenate baseline and damage signals\n",
    "df = np.concatenate([df_base,df_dam], axis=0)\n",
    "X = np.array(df)\n",
    "print(X.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(6048, 1)\n"
     ]
    }
   ],
   "source": [
    "# Input/Features and labels extraction\n",
    "# Baseline = 0\n",
    "# Damage = 1\n",
    "mclass = df_base.shape[0]\n",
    "y1 = np.zeros((mclass,1), dtype=int)\n",
    "y2 = np.ones((mclass,1), dtype=int)\n",
    "\n",
    "y = np.concatenate([y1,y2], axis=0)\n",
    "y = np.array(y)\n",
    "print(y.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Add Noise into the dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#del Xn1,n1,r,maxX"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(1, 12108)\n",
      "0.664957667855495\n",
      "(1, 12108)\n",
      "(6048, 12108)\n",
      "0.0012543115301581937\n",
      "4.023462953566065e-06\n",
      "SNR :  24.938054076337423\n"
     ]
    }
   ],
   "source": [
    "# Random gaussian noise parameter\n",
    "beta1 = 0.0025\n",
    "beta2 = 0.003\n",
    "\n",
    "mu = 0\n",
    "sigma = 1\n",
    "\n",
    "#random parameter with standard normal distribution\n",
    "r = sigma*np.random.randn(X.shape[1],1) + mu   \n",
    "r = np.transpose(r)\n",
    "print(r.shape)\n",
    "\n",
    "# Noisy signal\n",
    "maxX = np.max(X)\n",
    "maxX = np.array(maxX)\n",
    "print(maxX)\n",
    "\n",
    "n1 = beta1*r*maxX\n",
    "n2 = beta2*r*maxX\n",
    "print(n2.shape)\n",
    "\n",
    "Xn1 = X + n1\n",
    "Xn2 = X + n2\n",
    "print(Xn2.shape)\n",
    "\n",
    "# Signal to noise ratio\n",
    "import math\n",
    "\n",
    "rms_Xn = np.sqrt(np.mean(Xn2**2))\n",
    "Power_Xn = rms_Xn**2\n",
    "print(Power_Xn)\n",
    "\n",
    "rms_n = np.sqrt(np.mean(n2**2))\n",
    "Power_n = rms_n**2\n",
    "print(Power_n)\n",
    "\n",
    "SNR_dB = 10*math.log10(Power_Xn/Power_n)\n",
    "print(\"SNR : \",SNR_dB)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(18144, 12108)\n",
      "(18144, 1)\n"
     ]
    }
   ],
   "source": [
    "# Concatenate noisy and non-noisy samples\n",
    "Xn = np.concatenate([X,Xn1,Xn2], axis=0)\n",
    "yn = np.concatenate([y,y,y], axis=0)\n",
    "print(Xn.shape)\n",
    "print(yn.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "4736\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "Text(0.5, 1.0, 'Training sample')"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot signals\n",
    "D = random.randint(1, Xn.shape[0])\n",
    "print(D)\n",
    "\n",
    "plt.figure(figsize=(10,4))\n",
    "plt.plot(time[idx1:idx2], Xn[D, 0 : Xn.shape[1]])\n",
    "plt.title('Training sample')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "_Ga43nEOId8F"
   },
   "source": [
    "## Splitting testing and test set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# RERUN the model ==> Delete the model\n",
    "del model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "GXBYgxIbVdaH"
   },
   "outputs": [],
   "source": [
    "# split into train and test\n",
    "# Run this line of code for every run of the code if some parameters are changed\n",
    "X1, X_test, y1, y_test = train_test_split(Xn,yn, test_size = 0.005, random_state = 42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train, X_valid, y_train, y_valid = train_test_split(X1,y1, test_size = 0.1, random_state = 42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 87
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 7372,
     "status": "ok",
     "timestamp": 1571626984037,
     "user": {
      "displayName": "Mahindra Singh Rautela",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mBTSnpeHr1j42u0w7ABMmG3BYbWyCtFQVbOvr6sAA=s64",
      "userId": "15859880813264051870"
     },
     "user_tz": -330
    },
    "id": "qtkixnevVfr4",
    "outputId": "3c7ae0b4-d2fb-4bd8-b7e1-7c6ed8a0dedb"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(16247, 12108)\n",
      "(1806, 12108)\n",
      "(91, 12108)\n",
      "(16247, 1)\n",
      "(1806, 1)\n",
      "(91, 1)\n"
     ]
    }
   ],
   "source": [
    "# Shapes of training and test tests\n",
    "X_train = np.array(X_train)\n",
    "X_valid = np.array(X_valid)\n",
    "X_test = np.array(X_test)\n",
    "y_train = np.array(y_train)\n",
    "y_train = np.array(y_train)\n",
    "y_train = np.array(y_train)\n",
    "print(X_train.shape)\n",
    "print(X_valid.shape)\n",
    "print(X_test.shape)\n",
    "print(y_train.shape)\n",
    "print(y_valid.shape)\n",
    "print(y_test.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "N3mc165NIib0"
   },
   "source": [
    "## Reshape the arrays"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 212
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 6545,
     "status": "ok",
     "timestamp": 1571626984335,
     "user": {
      "displayName": "Mahindra Singh Rautela",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mBTSnpeHr1j42u0w7ABMmG3BYbWyCtFQVbOvr6sAA=s64",
      "userId": "15859880813264051870"
     },
     "user_tz": -330
    },
    "id": "S0GkyLE2ViB4",
    "outputId": "bd475f73-8cdc-444e-e576-87df1bd9d07c"
   },
   "outputs": [],
   "source": [
    "# reshape the arrays\n",
    "X_train = np.reshape(X_train, (X_train.shape[0],X_train.shape[1],1))\n",
    "X_valid = np.reshape(X_valid, (X_valid.shape[0],X_valid.shape[1],1))\n",
    "X_test  = np.reshape(X_test,  (X_test.shape[0],X_test.shape[1],1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(16247, 12108, 1)\n",
      "(1806, 12108, 1)\n",
      "(16247, 1)\n",
      "(1806, 1)\n"
     ]
    }
   ],
   "source": [
    "print(X_train.shape)\n",
    "print(X_valid.shape)\n",
    "print(y_train.shape)\n",
    "print(y_valid.shape)\n",
    "# print(y_train)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "XW7VmHZfInYs"
   },
   "source": [
    "## Model Architecture"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 90
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 825,
     "status": "ok",
     "timestamp": 1571626986180,
     "user": {
      "displayName": "Mahindra Singh Rautela",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mBTSnpeHr1j42u0w7ABMmG3BYbWyCtFQVbOvr6sAA=s64",
      "userId": "15859880813264051870"
     },
     "user_tz": -330
    },
    "id": "azmX3H0DVkkJ",
    "outputId": "73d34f6a-75d7-423e-a653-87a636ea6ef8"
   },
   "outputs": [],
   "source": [
    "# define model architecture : 1DCNN-classification\n",
    "from keras.layers import Input\n",
    "from keras.models import Model\n",
    "from keras.layers import concatenate\n",
    "from keras.models import Sequential\n",
    "from keras.layers.normalization import BatchNormalization\n",
    "from keras.layers.convolutional import Conv1D\n",
    "from keras.layers.convolutional import MaxPooling1D\n",
    "from keras.layers.core import Activation\n",
    "from keras.layers.core import Dropout\n",
    "from keras.layers.core import Dense\n",
    "from keras.layers import Flatten\n",
    "from keras.layers import Input\n",
    "from keras.models import Model\n",
    "\n",
    "Inp = Input(shape=(X.shape[1],1))\n",
    "\n",
    "x1 = Sequential()\n",
    "\n",
    "x1 = Conv1D(filters=16, kernel_size=3, input_shape=(X.shape[1],1))(Inp)\n",
    "#x1 = BatchNormalization()(x1)\n",
    "x1 = MaxPooling1D(pool_size=2)(x1)\n",
    "\n",
    "x1 = Conv1D(filters=32, kernel_size=3, activation='relu')(x1)\n",
    "#x1 = BatchNormalization()(x1)\n",
    "x1 = MaxPooling1D(pool_size=2)(x1)\n",
    "\n",
    "x1 = Conv1D(filters=64, kernel_size=3, activation='relu')(x1)\n",
    "#x1 = BatchNormalization()(x1)\n",
    "x1 = MaxPooling1D(pool_size=2)(x1)\n",
    "\n",
    "x1 = Conv1D(filters=128, kernel_size=3, activation='relu')(x1)\n",
    "#x1 = BatchNormalization()(x1)\n",
    "x1 = MaxPooling1D(pool_size=2)(x1)\n",
    "\n",
    "x1 = Conv1D(filters=256, kernel_size=3, activation='relu')(x1)\n",
    "#x1 = BatchNormalization()(x1)\n",
    "x1 = MaxPooling1D(pool_size=2)(x1)\n",
    "\n",
    "x1 = Flatten()(x1)\n",
    "\n",
    "cnn1 = Model(Inp,x1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 478
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 559,
     "status": "ok",
     "timestamp": 1571626986708,
     "user": {
      "displayName": "Mahindra Singh Rautela",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mBTSnpeHr1j42u0w7ABMmG3BYbWyCtFQVbOvr6sAA=s64",
      "userId": "15859880813264051870"
     },
     "user_tz": -330
    },
    "id": "JMZwl4oJVm_s",
    "outputId": "049265c3-1b4f-4ff1-a307-c5862f1ffd09"
   },
   "outputs": [],
   "source": [
    "from keras.layers import concatenate\n",
    "#x = concatenate([cnn1.output,cnn2.output])\n",
    "x = cnn1.output\n",
    "#---------------------------------\n",
    "x = Dense(128, activation='relu')(x)\n",
    "x = Dropout(0.25)(x)\n",
    "#x = Dense(16, activation='relu')(x)\n",
    "#x = Dropout(0.2)(x)\n",
    "x = Dense(1, activation='sigmoid')(x)\n",
    "#---------------------------------\n",
    "model = Model(inputs = Inp, outputs=x)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model_2\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_1 (InputLayer)         (None, 12108, 1)          0         \n",
      "_________________________________________________________________\n",
      "conv1d_1 (Conv1D)            (None, 12106, 16)         64        \n",
      "_________________________________________________________________\n",
      "max_pooling1d_1 (MaxPooling1 (None, 6053, 16)          0         \n",
      "_________________________________________________________________\n",
      "conv1d_2 (Conv1D)            (None, 6051, 32)          1568      \n",
      "_________________________________________________________________\n",
      "max_pooling1d_2 (MaxPooling1 (None, 3025, 32)          0         \n",
      "_________________________________________________________________\n",
      "conv1d_3 (Conv1D)            (None, 3023, 64)          6208      \n",
      "_________________________________________________________________\n",
      "max_pooling1d_3 (MaxPooling1 (None, 1511, 64)          0         \n",
      "_________________________________________________________________\n",
      "conv1d_4 (Conv1D)            (None, 1509, 128)         24704     \n",
      "_________________________________________________________________\n",
      "max_pooling1d_4 (MaxPooling1 (None, 754, 128)          0         \n",
      "_________________________________________________________________\n",
      "conv1d_5 (Conv1D)            (None, 752, 256)          98560     \n",
      "_________________________________________________________________\n",
      "max_pooling1d_5 (MaxPooling1 (None, 376, 256)          0         \n",
      "_________________________________________________________________\n",
      "flatten_1 (Flatten)          (None, 96256)             0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 128)               12320896  \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 128)               0         \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 1)                 129       \n",
      "=================================================================\n",
      "Total params: 12,452,129\n",
      "Trainable params: 12,452,129\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "# summary of the model\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "# save model.summary()\n",
    "from contextlib import redirect_stdout\n",
    "\n",
    "with open('modelsummary.txt', 'w') as f:\n",
    "    with redirect_stdout(f):\n",
    "        model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import keras\n",
    "keras.utils.plot_model(model, \"detect1DCNN.png\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Functions for recall, precision and f1 score\n",
    "from keras import backend as K\n",
    "\n",
    "def recall_m(y_true, y_pred):\n",
    "    true_positives = K.sum(K.round(K.clip(y_true * y_pred, 0, 1)))\n",
    "    possible_positives = K.sum(K.round(K.clip(y_true, 0, 1)))\n",
    "    recall = true_positives / (possible_positives + K.epsilon())\n",
    "    return recall\n",
    "\n",
    "def precision_m(y_true, y_pred):\n",
    "    true_positives = K.sum(K.round(K.clip(y_true * y_pred, 0, 1)))\n",
    "    predicted_positives = K.sum(K.round(K.clip(y_pred, 0, 1)))\n",
    "    precision = true_positives / (predicted_positives + K.epsilon())\n",
    "    return precision\n",
    "\n",
    "def f1_m(y_true, y_pred):\n",
    "    precision = precision_m(y_true, y_pred)\n",
    "    recall = recall_m(y_true, y_pred)\n",
    "    return 2*((precision*recall)/(precision+recall+K.epsilon()))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "8tiR31NyIqjJ"
   },
   "source": [
    "## Choose Hyperparameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 90
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 821,
     "status": "ok",
     "timestamp": 1571626989175,
     "user": {
      "displayName": "Mahindra Singh Rautela",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mBTSnpeHr1j42u0w7ABMmG3BYbWyCtFQVbOvr6sAA=s64",
      "userId": "15859880813264051870"
     },
     "user_tz": -330
    },
    "id": "3TxHGquwVpZO",
    "outputId": "f8f6c7a4-e85c-4c76-c5da-71a7589ab847"
   },
   "outputs": [],
   "source": [
    "# compile model\n",
    "model.compile(loss='binary_crossentropy', optimizer=Adam(lr=1e-5), metrics=['acc', precision_m, recall_m])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "SAqS8ppy-13X"
   },
   "source": [
    "## Training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 887
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 73050,
     "status": "ok",
     "timestamp": 1571627063426,
     "user": {
      "displayName": "Mahindra Singh Rautela",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mBTSnpeHr1j42u0w7ABMmG3BYbWyCtFQVbOvr6sAA=s64",
      "userId": "15859880813264051870"
     },
     "user_tz": -330
    },
    "id": "v6kF1cwfVu0n",
    "outputId": "ada6c377-ce02-4132-a01a-17832050313d"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 16247 samples, validate on 1806 samples\n",
      "Epoch 1/5000\n",
      " - 13s - loss: 0.6932 - acc: 0.4944 - precision_m: 0.4373 - recall_m: 0.0891 - val_loss: 0.6932 - val_acc: 0.4884 - val_precision_m: 0.2333 - val_recall_m: 0.0052\n",
      "Epoch 2/5000\n",
      " - 9s - loss: 0.6932 - acc: 0.4984 - precision_m: 0.3088 - recall_m: 0.0879 - val_loss: 0.6933 - val_acc: 0.4762 - val_precision_m: 0.3874 - val_recall_m: 0.0441\n",
      "Epoch 3/5000\n",
      " - 9s - loss: 0.6932 - acc: 0.4973 - precision_m: 0.4861 - recall_m: 0.0677 - val_loss: 0.6933 - val_acc: 0.4729 - val_precision_m: 0.4355 - val_recall_m: 0.0897\n",
      "Epoch 4/5000\n",
      " - 9s - loss: 0.6931 - acc: 0.4997 - precision_m: 0.5006 - recall_m: 0.1674 - val_loss: 0.6934 - val_acc: 0.4762 - val_precision_m: 0.4709 - val_recall_m: 0.1098\n",
      "Epoch 5/5000\n",
      " - 9s - loss: 0.6932 - acc: 0.4986 - precision_m: 0.4696 - recall_m: 0.1806 - val_loss: 0.6934 - val_acc: 0.4779 - val_precision_m: 0.4481 - val_recall_m: 0.0704\n",
      "Epoch 6/5000\n",
      " - 9s - loss: 0.6931 - acc: 0.5059 - precision_m: 0.5274 - recall_m: 0.1900 - val_loss: 0.6935 - val_acc: 0.4806 - val_precision_m: 0.4171 - val_recall_m: 0.0492\n",
      "Epoch 7/5000\n",
      " - 9s - loss: 0.6931 - acc: 0.5036 - precision_m: 0.5194 - recall_m: 0.2446 - val_loss: 0.6936 - val_acc: 0.4834 - val_precision_m: 0.4022 - val_recall_m: 0.0298\n",
      "Epoch 8/5000\n",
      " - 9s - loss: 0.6931 - acc: 0.5062 - precision_m: 0.5162 - recall_m: 0.1589 - val_loss: 0.6936 - val_acc: 0.4651 - val_precision_m: 0.4370 - val_recall_m: 0.1402\n",
      "Epoch 9/5000\n",
      " - 9s - loss: 0.6930 - acc: 0.5128 - precision_m: 0.5287 - recall_m: 0.3334 - val_loss: 0.6937 - val_acc: 0.4834 - val_precision_m: 0.4049 - val_recall_m: 0.0288\n",
      "Epoch 10/5000\n",
      " - 9s - loss: 0.6931 - acc: 0.5075 - precision_m: 0.5201 - recall_m: 0.1898 - val_loss: 0.6937 - val_acc: 0.4862 - val_precision_m: 0.5010 - val_recall_m: 0.5851\n",
      "Epoch 11/5000\n",
      " - 10s - loss: 0.6930 - acc: 0.5116 - precision_m: 0.5187 - recall_m: 0.5355 - val_loss: 0.6938 - val_acc: 0.4845 - val_precision_m: 0.5011 - val_recall_m: 0.4780\n",
      "Epoch 12/5000\n",
      " - 10s - loss: 0.6930 - acc: 0.5033 - precision_m: 0.5018 - recall_m: 0.3642 - val_loss: 0.6940 - val_acc: 0.4762 - val_precision_m: 0.4472 - val_recall_m: 0.0765\n",
      "Epoch 13/5000\n",
      " - 10s - loss: 0.6929 - acc: 0.5101 - precision_m: 0.5175 - recall_m: 0.4387 - val_loss: 0.6941 - val_acc: 0.4701 - val_precision_m: 0.4426 - val_recall_m: 0.1159\n",
      "Epoch 14/5000\n",
      " - 10s - loss: 0.6928 - acc: 0.5122 - precision_m: 0.5345 - recall_m: 0.3033 - val_loss: 0.6939 - val_acc: 0.4956 - val_precision_m: 0.5087 - val_recall_m: 0.8996\n",
      "Epoch 15/5000\n",
      " - 10s - loss: 0.6929 - acc: 0.5125 - precision_m: 0.5168 - recall_m: 0.5552 - val_loss: 0.6942 - val_acc: 0.4767 - val_precision_m: 0.4879 - val_recall_m: 0.2920\n",
      "Epoch 16/5000\n",
      " - 10s - loss: 0.6929 - acc: 0.5050 - precision_m: 0.5050 - recall_m: 0.4057 - val_loss: 0.6940 - val_acc: 0.4994 - val_precision_m: 0.5150 - val_recall_m: 0.6550\n",
      "Epoch 17/5000\n",
      " - 10s - loss: 0.6928 - acc: 0.5171 - precision_m: 0.5257 - recall_m: 0.5663 - val_loss: 0.6942 - val_acc: 0.4773 - val_precision_m: 0.5002 - val_recall_m: 0.4857\n",
      "Epoch 18/5000\n",
      " - 10s - loss: 0.6927 - acc: 0.5115 - precision_m: 0.5195 - recall_m: 0.3820 - val_loss: 0.6942 - val_acc: 0.4895 - val_precision_m: 0.5071 - val_recall_m: 0.5720\n",
      "Epoch 19/5000\n",
      " - 10s - loss: 0.6926 - acc: 0.5121 - precision_m: 0.5222 - recall_m: 0.5742 - val_loss: 0.6952 - val_acc: 0.4845 - val_precision_m: 0.3493 - val_recall_m: 0.0227\n",
      "Epoch 20/5000\n",
      " - 10s - loss: 0.6927 - acc: 0.5099 - precision_m: 0.5306 - recall_m: 0.4791 - val_loss: 0.6946 - val_acc: 0.4845 - val_precision_m: 0.4973 - val_recall_m: 0.2911\n",
      "Epoch 21/5000\n",
      " - 10s - loss: 0.6925 - acc: 0.5126 - precision_m: 0.5177 - recall_m: 0.4172 - val_loss: 0.6945 - val_acc: 0.5011 - val_precision_m: 0.5114 - val_recall_m: 0.8927\n",
      "Epoch 22/5000\n",
      " - 10s - loss: 0.6925 - acc: 0.5122 - precision_m: 0.5169 - recall_m: 0.4998 - val_loss: 0.6946 - val_acc: 0.4740 - val_precision_m: 0.4927 - val_recall_m: 0.4151\n",
      "Epoch 23/5000\n",
      " - 10s - loss: 0.6925 - acc: 0.5184 - precision_m: 0.5283 - recall_m: 0.5012 - val_loss: 0.6950 - val_acc: 0.4839 - val_precision_m: 0.5098 - val_recall_m: 0.1404\n",
      "Epoch 24/5000\n",
      " - 10s - loss: 0.6923 - acc: 0.5166 - precision_m: 0.5265 - recall_m: 0.4558 - val_loss: 0.6944 - val_acc: 0.4917 - val_precision_m: 0.5111 - val_recall_m: 0.6794\n",
      "Epoch 25/5000\n",
      " - 10s - loss: 0.6923 - acc: 0.5202 - precision_m: 0.5222 - recall_m: 0.5110 - val_loss: 0.6953 - val_acc: 0.4823 - val_precision_m: 0.4875 - val_recall_m: 0.1539\n",
      "Epoch 26/5000\n",
      " - 10s - loss: 0.6922 - acc: 0.5172 - precision_m: 0.5282 - recall_m: 0.4603 - val_loss: 0.6950 - val_acc: 0.4740 - val_precision_m: 0.4993 - val_recall_m: 0.4887\n",
      "Epoch 27/5000\n",
      " - 10s - loss: 0.6923 - acc: 0.5191 - precision_m: 0.5259 - recall_m: 0.5698 - val_loss: 0.6957 - val_acc: 0.4817 - val_precision_m: 0.4390 - val_recall_m: 0.0877\n",
      "Epoch 28/5000\n",
      " - 10s - loss: 0.6918 - acc: 0.5231 - precision_m: 0.5212 - recall_m: 0.4946 - val_loss: 0.6962 - val_acc: 0.4834 - val_precision_m: 0.3760 - val_recall_m: 0.0239\n",
      "Epoch 29/5000\n",
      " - 10s - loss: 0.6921 - acc: 0.5165 - precision_m: 0.5339 - recall_m: 0.4561 - val_loss: 0.6948 - val_acc: 0.4889 - val_precision_m: 0.5089 - val_recall_m: 0.6329\n",
      "Epoch 30/5000\n",
      " - 10s - loss: 0.6921 - acc: 0.5217 - precision_m: 0.5302 - recall_m: 0.5514 - val_loss: 0.6949 - val_acc: 0.4673 - val_precision_m: 0.4898 - val_recall_m: 0.4198\n",
      "Epoch 31/5000\n",
      " - 10s - loss: 0.6916 - acc: 0.5249 - precision_m: 0.5322 - recall_m: 0.5195 - val_loss: 0.6962 - val_acc: 0.4856 - val_precision_m: 0.4502 - val_recall_m: 0.0714\n",
      "Epoch 32/5000\n",
      " - 10s - loss: 0.6918 - acc: 0.5247 - precision_m: 0.5395 - recall_m: 0.4060 - val_loss: 0.6948 - val_acc: 0.4911 - val_precision_m: 0.5113 - val_recall_m: 0.6929\n",
      "Epoch 33/5000\n",
      " - 10s - loss: 0.6916 - acc: 0.5275 - precision_m: 0.5298 - recall_m: 0.5513 - val_loss: 0.6951 - val_acc: 0.4873 - val_precision_m: 0.5134 - val_recall_m: 0.3416\n",
      "Epoch 34/5000\n",
      " - 10s - loss: 0.6915 - acc: 0.5265 - precision_m: 0.5339 - recall_m: 0.4770 - val_loss: 0.6954 - val_acc: 0.4845 - val_precision_m: 0.4990 - val_recall_m: 0.3258\n",
      "Epoch 35/5000\n",
      " - 10s - loss: 0.6914 - acc: 0.5308 - precision_m: 0.5336 - recall_m: 0.5149 - val_loss: 0.6952 - val_acc: 0.4895 - val_precision_m: 0.5121 - val_recall_m: 0.3295\n",
      "Epoch 36/5000\n",
      " - 10s - loss: 0.6912 - acc: 0.5271 - precision_m: 0.5411 - recall_m: 0.4620 - val_loss: 0.6949 - val_acc: 0.5039 - val_precision_m: 0.5201 - val_recall_m: 0.7819\n",
      "Epoch 37/5000\n",
      " - 10s - loss: 0.6912 - acc: 0.5256 - precision_m: 0.5284 - recall_m: 0.4835 - val_loss: 0.6948 - val_acc: 0.4828 - val_precision_m: 0.5052 - val_recall_m: 0.6611\n",
      "Epoch 38/5000\n",
      " - 10s - loss: 0.6914 - acc: 0.5293 - precision_m: 0.5350 - recall_m: 0.5229 - val_loss: 0.6948 - val_acc: 0.4983 - val_precision_m: 0.5163 - val_recall_m: 0.5668\n",
      "Epoch 39/5000\n",
      " - 10s - loss: 0.6907 - acc: 0.5293 - precision_m: 0.5364 - recall_m: 0.5179 - val_loss: 0.6951 - val_acc: 0.4989 - val_precision_m: 0.5174 - val_recall_m: 0.7681\n",
      "Epoch 40/5000\n",
      " - 10s - loss: 0.6912 - acc: 0.5243 - precision_m: 0.5348 - recall_m: 0.4911 - val_loss: 0.6947 - val_acc: 0.4850 - val_precision_m: 0.5122 - val_recall_m: 0.3976\n",
      "Epoch 41/5000\n",
      " - 10s - loss: 0.6906 - acc: 0.5301 - precision_m: 0.5370 - recall_m: 0.4777 - val_loss: 0.6945 - val_acc: 0.4922 - val_precision_m: 0.5133 - val_recall_m: 0.6947\n",
      "Epoch 42/5000\n",
      " - 10s - loss: 0.6905 - acc: 0.5347 - precision_m: 0.5384 - recall_m: 0.5263 - val_loss: 0.6946 - val_acc: 0.5011 - val_precision_m: 0.5204 - val_recall_m: 0.5743\n",
      "Epoch 43/5000\n",
      " - 10s - loss: 0.6905 - acc: 0.5323 - precision_m: 0.5427 - recall_m: 0.4913 - val_loss: 0.6948 - val_acc: 0.4994 - val_precision_m: 0.5105 - val_recall_m: 0.8992\n",
      "Epoch 44/5000\n",
      " - 10s - loss: 0.6905 - acc: 0.5330 - precision_m: 0.5384 - recall_m: 0.5547 - val_loss: 0.6965 - val_acc: 0.4983 - val_precision_m: 0.5309 - val_recall_m: 0.0630\n",
      "Epoch 45/5000\n",
      " - 10s - loss: 0.6903 - acc: 0.5371 - precision_m: 0.5523 - recall_m: 0.4745 - val_loss: 0.6940 - val_acc: 0.5111 - val_precision_m: 0.5294 - val_recall_m: 0.6388\n",
      "Epoch 46/5000\n",
      " - 10s - loss: 0.6902 - acc: 0.5301 - precision_m: 0.5384 - recall_m: 0.5195 - val_loss: 0.6944 - val_acc: 0.4967 - val_precision_m: 0.5186 - val_recall_m: 0.4935\n",
      "Epoch 47/5000\n",
      " - 10s - loss: 0.6901 - acc: 0.5389 - precision_m: 0.5494 - recall_m: 0.5197 - val_loss: 0.6943 - val_acc: 0.4972 - val_precision_m: 0.5182 - val_recall_m: 0.4423\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 48/5000\n",
      " - 10s - loss: 0.6896 - acc: 0.5401 - precision_m: 0.5584 - recall_m: 0.5244 - val_loss: 0.6943 - val_acc: 0.5100 - val_precision_m: 0.5196 - val_recall_m: 0.8421\n",
      "Epoch 49/5000\n",
      " - 10s - loss: 0.6894 - acc: 0.5376 - precision_m: 0.5474 - recall_m: 0.5189 - val_loss: 0.6942 - val_acc: 0.5199 - val_precision_m: 0.5293 - val_recall_m: 0.8348\n",
      "Epoch 50/5000\n",
      " - 10s - loss: 0.6893 - acc: 0.5434 - precision_m: 0.5505 - recall_m: 0.5319 - val_loss: 0.6944 - val_acc: 0.5022 - val_precision_m: 0.5252 - val_recall_m: 0.3845\n",
      "Epoch 51/5000\n",
      " - 10s - loss: 0.6893 - acc: 0.5419 - precision_m: 0.5564 - recall_m: 0.5023 - val_loss: 0.6950 - val_acc: 0.5089 - val_precision_m: 0.5681 - val_recall_m: 0.2305\n",
      "Epoch 52/5000\n",
      " - 10s - loss: 0.6886 - acc: 0.5460 - precision_m: 0.5609 - recall_m: 0.5064 - val_loss: 0.6936 - val_acc: 0.5133 - val_precision_m: 0.5306 - val_recall_m: 0.5832\n",
      "Epoch 53/5000\n",
      " - 10s - loss: 0.6893 - acc: 0.5384 - precision_m: 0.5526 - recall_m: 0.5017 - val_loss: 0.6938 - val_acc: 0.5216 - val_precision_m: 0.5309 - val_recall_m: 0.8121\n",
      "Epoch 54/5000\n",
      " - 10s - loss: 0.6886 - acc: 0.5442 - precision_m: 0.5551 - recall_m: 0.5084 - val_loss: 0.6933 - val_acc: 0.5044 - val_precision_m: 0.5223 - val_recall_m: 0.6974\n",
      "Epoch 55/5000\n",
      " - 10s - loss: 0.6881 - acc: 0.5483 - precision_m: 0.5546 - recall_m: 0.5403 - val_loss: 0.6938 - val_acc: 0.5011 - val_precision_m: 0.5213 - val_recall_m: 0.3123\n",
      "Epoch 56/5000\n",
      " - 10s - loss: 0.6878 - acc: 0.5528 - precision_m: 0.5659 - recall_m: 0.5191 - val_loss: 0.6937 - val_acc: 0.5083 - val_precision_m: 0.5594 - val_recall_m: 0.3025\n",
      "Epoch 57/5000\n",
      " - 10s - loss: 0.6879 - acc: 0.5471 - precision_m: 0.5583 - recall_m: 0.5101 - val_loss: 0.6925 - val_acc: 0.5205 - val_precision_m: 0.5265 - val_recall_m: 0.8030\n",
      "Epoch 58/5000\n",
      " - 10s - loss: 0.6881 - acc: 0.5424 - precision_m: 0.5658 - recall_m: 0.5472 - val_loss: 0.6949 - val_acc: 0.5006 - val_precision_m: 0.5013 - val_recall_m: 0.1141\n",
      "Epoch 59/5000\n",
      " - 10s - loss: 0.6873 - acc: 0.5535 - precision_m: 0.5699 - recall_m: 0.5070 - val_loss: 0.6925 - val_acc: 0.5177 - val_precision_m: 0.5294 - val_recall_m: 0.7162\n",
      "Epoch 60/5000\n",
      " - 10s - loss: 0.6870 - acc: 0.5554 - precision_m: 0.5670 - recall_m: 0.5567 - val_loss: 0.6917 - val_acc: 0.5238 - val_precision_m: 0.5561 - val_recall_m: 0.4286\n",
      "Epoch 61/5000\n",
      " - 10s - loss: 0.6866 - acc: 0.5561 - precision_m: 0.5677 - recall_m: 0.5262 - val_loss: 0.6931 - val_acc: 0.5055 - val_precision_m: 0.5563 - val_recall_m: 0.2365\n",
      "Epoch 62/5000\n",
      " - 10s - loss: 0.6864 - acc: 0.5613 - precision_m: 0.5757 - recall_m: 0.5294 - val_loss: 0.6936 - val_acc: 0.5127 - val_precision_m: 0.6044 - val_recall_m: 0.1634\n",
      "Epoch 63/5000\n",
      " - 10s - loss: 0.6865 - acc: 0.5589 - precision_m: 0.5777 - recall_m: 0.5051 - val_loss: 0.6915 - val_acc: 0.5210 - val_precision_m: 0.5346 - val_recall_m: 0.7020\n",
      "Epoch 64/5000\n",
      " - 10s - loss: 0.6861 - acc: 0.5605 - precision_m: 0.5703 - recall_m: 0.5602 - val_loss: 0.6914 - val_acc: 0.5255 - val_precision_m: 0.5400 - val_recall_m: 0.6017\n",
      "Epoch 65/5000\n",
      " - 10s - loss: 0.6860 - acc: 0.5569 - precision_m: 0.5680 - recall_m: 0.5354 - val_loss: 0.6908 - val_acc: 0.5382 - val_precision_m: 0.5710 - val_recall_m: 0.4593\n",
      "Epoch 66/5000\n",
      " - 10s - loss: 0.6854 - acc: 0.5644 - precision_m: 0.5786 - recall_m: 0.5316 - val_loss: 0.6899 - val_acc: 0.5227 - val_precision_m: 0.5437 - val_recall_m: 0.4524\n",
      "Epoch 67/5000\n",
      " - 10s - loss: 0.6847 - acc: 0.5716 - precision_m: 0.5833 - recall_m: 0.5467 - val_loss: 0.6902 - val_acc: 0.5360 - val_precision_m: 0.5714 - val_recall_m: 0.3686\n",
      "Epoch 68/5000\n",
      " - 10s - loss: 0.6848 - acc: 0.5660 - precision_m: 0.5855 - recall_m: 0.5530 - val_loss: 0.6924 - val_acc: 0.5177 - val_precision_m: 0.6302 - val_recall_m: 0.1455\n",
      "Epoch 69/5000\n",
      " - 10s - loss: 0.6845 - acc: 0.5656 - precision_m: 0.5882 - recall_m: 0.5518 - val_loss: 0.6894 - val_acc: 0.5371 - val_precision_m: 0.5556 - val_recall_m: 0.5633\n",
      "Epoch 70/5000\n",
      " - 10s - loss: 0.6837 - acc: 0.5674 - precision_m: 0.5844 - recall_m: 0.5451 - val_loss: 0.6893 - val_acc: 0.5471 - val_precision_m: 0.5979 - val_recall_m: 0.4416\n",
      "Epoch 71/5000\n",
      " - 10s - loss: 0.6833 - acc: 0.5712 - precision_m: 0.5785 - recall_m: 0.5760 - val_loss: 0.6888 - val_acc: 0.5576 - val_precision_m: 0.5997 - val_recall_m: 0.4543\n",
      "Epoch 72/5000\n",
      " - 10s - loss: 0.6826 - acc: 0.5730 - precision_m: 0.5934 - recall_m: 0.5570 - val_loss: 0.6881 - val_acc: 0.5637 - val_precision_m: 0.6133 - val_recall_m: 0.4817\n",
      "Epoch 73/5000\n",
      " - 10s - loss: 0.6833 - acc: 0.5712 - precision_m: 0.5888 - recall_m: 0.5481 - val_loss: 0.6885 - val_acc: 0.5443 - val_precision_m: 0.6096 - val_recall_m: 0.3589\n",
      "Epoch 74/5000\n",
      " - 10s - loss: 0.6822 - acc: 0.5762 - precision_m: 0.5908 - recall_m: 0.5524 - val_loss: 0.6881 - val_acc: 0.5554 - val_precision_m: 0.6062 - val_recall_m: 0.4545\n",
      "Epoch 75/5000\n",
      " - 10s - loss: 0.6820 - acc: 0.5741 - precision_m: 0.5934 - recall_m: 0.5557 - val_loss: 0.6868 - val_acc: 0.5570 - val_precision_m: 0.5960 - val_recall_m: 0.4576\n",
      "Epoch 76/5000\n",
      " - 10s - loss: 0.6811 - acc: 0.5781 - precision_m: 0.5968 - recall_m: 0.5340 - val_loss: 0.6867 - val_acc: 0.5670 - val_precision_m: 0.6178 - val_recall_m: 0.4854\n",
      "Epoch 77/5000\n",
      " - 10s - loss: 0.6819 - acc: 0.5767 - precision_m: 0.6007 - recall_m: 0.5754 - val_loss: 0.6868 - val_acc: 0.5504 - val_precision_m: 0.5423 - val_recall_m: 0.8484\n",
      "Epoch 78/5000\n",
      " - 10s - loss: 0.6798 - acc: 0.5911 - precision_m: 0.6077 - recall_m: 0.5625 - val_loss: 0.6851 - val_acc: 0.5664 - val_precision_m: 0.6233 - val_recall_m: 0.4430\n",
      "Epoch 79/5000\n",
      " - 10s - loss: 0.6803 - acc: 0.5841 - precision_m: 0.6043 - recall_m: 0.5583 - val_loss: 0.6846 - val_acc: 0.5709 - val_precision_m: 0.6311 - val_recall_m: 0.4454\n",
      "Epoch 80/5000\n",
      " - 10s - loss: 0.6795 - acc: 0.5840 - precision_m: 0.6079 - recall_m: 0.5565 - val_loss: 0.6862 - val_acc: 0.5465 - val_precision_m: 0.6454 - val_recall_m: 0.2834\n",
      "Epoch 81/5000\n",
      " - 10s - loss: 0.6792 - acc: 0.5879 - precision_m: 0.6011 - recall_m: 0.5884 - val_loss: 0.6834 - val_acc: 0.5681 - val_precision_m: 0.5727 - val_recall_m: 0.6860\n",
      "Epoch 82/5000\n",
      " - 10s - loss: 0.6779 - acc: 0.5917 - precision_m: 0.6069 - recall_m: 0.5657 - val_loss: 0.6831 - val_acc: 0.5681 - val_precision_m: 0.5897 - val_recall_m: 0.6237\n",
      "Epoch 83/5000\n",
      " - 10s - loss: 0.6775 - acc: 0.5957 - precision_m: 0.6104 - recall_m: 0.5823 - val_loss: 0.6823 - val_acc: 0.5709 - val_precision_m: 0.5645 - val_recall_m: 0.7852\n",
      "Epoch 84/5000\n",
      " - 10s - loss: 0.6769 - acc: 0.5924 - precision_m: 0.6103 - recall_m: 0.5876 - val_loss: 0.6831 - val_acc: 0.5676 - val_precision_m: 0.6607 - val_recall_m: 0.3594\n",
      "Epoch 85/5000\n",
      " - 10s - loss: 0.6769 - acc: 0.5917 - precision_m: 0.6118 - recall_m: 0.5854 - val_loss: 0.6835 - val_acc: 0.5493 - val_precision_m: 0.6619 - val_recall_m: 0.2696\n",
      "Epoch 86/5000\n",
      " - 10s - loss: 0.6753 - acc: 0.6002 - precision_m: 0.6180 - recall_m: 0.5823 - val_loss: 0.6812 - val_acc: 0.5836 - val_precision_m: 0.6697 - val_recall_m: 0.4030\n",
      "Epoch 87/5000\n",
      " - 10s - loss: 0.6752 - acc: 0.5997 - precision_m: 0.6139 - recall_m: 0.5856 - val_loss: 0.6818 - val_acc: 0.5570 - val_precision_m: 0.6687 - val_recall_m: 0.2854\n",
      "Epoch 88/5000\n",
      " - 10s - loss: 0.6754 - acc: 0.5896 - precision_m: 0.6174 - recall_m: 0.5676 - val_loss: 0.6821 - val_acc: 0.5449 - val_precision_m: 0.5377 - val_recall_m: 0.9240\n",
      "Epoch 89/5000\n",
      " - 10s - loss: 0.6746 - acc: 0.5947 - precision_m: 0.6112 - recall_m: 0.5862 - val_loss: 0.6781 - val_acc: 0.6024 - val_precision_m: 0.6792 - val_recall_m: 0.4512\n",
      "Epoch 90/5000\n",
      " - 10s - loss: 0.6734 - acc: 0.6084 - precision_m: 0.6223 - recall_m: 0.6024 - val_loss: 0.6786 - val_acc: 0.5637 - val_precision_m: 0.5564 - val_recall_m: 0.7793\n",
      "Epoch 91/5000\n",
      " - 10s - loss: 0.6727 - acc: 0.6019 - precision_m: 0.6234 - recall_m: 0.5994 - val_loss: 0.6767 - val_acc: 0.5875 - val_precision_m: 0.5792 - val_recall_m: 0.7585\n",
      "Epoch 92/5000\n",
      " - 10s - loss: 0.6719 - acc: 0.6116 - precision_m: 0.6240 - recall_m: 0.5963 - val_loss: 0.6759 - val_acc: 0.5831 - val_precision_m: 0.5771 - val_recall_m: 0.7733\n",
      "Epoch 93/5000\n",
      " - 10s - loss: 0.6705 - acc: 0.6107 - precision_m: 0.6241 - recall_m: 0.6070 - val_loss: 0.6756 - val_acc: 0.5969 - val_precision_m: 0.5990 - val_recall_m: 0.6926\n",
      "Epoch 94/5000\n",
      " - 10s - loss: 0.6703 - acc: 0.6114 - precision_m: 0.6265 - recall_m: 0.6168 - val_loss: 0.6745 - val_acc: 0.6069 - val_precision_m: 0.6818 - val_recall_m: 0.4649\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 95/5000\n",
      " - 10s - loss: 0.6714 - acc: 0.6039 - precision_m: 0.6316 - recall_m: 0.5938 - val_loss: 0.6739 - val_acc: 0.5952 - val_precision_m: 0.5975 - val_recall_m: 0.6557\n",
      "Epoch 96/5000\n",
      " - 10s - loss: 0.6684 - acc: 0.6203 - precision_m: 0.6307 - recall_m: 0.6085 - val_loss: 0.6737 - val_acc: 0.5753 - val_precision_m: 0.5675 - val_recall_m: 0.7693\n",
      "Epoch 97/5000\n",
      " - 10s - loss: 0.6688 - acc: 0.6164 - precision_m: 0.6358 - recall_m: 0.6060 - val_loss: 0.6706 - val_acc: 0.6030 - val_precision_m: 0.5940 - val_recall_m: 0.7646\n",
      "Epoch 98/5000\n",
      " - 11s - loss: 0.6676 - acc: 0.6121 - precision_m: 0.6325 - recall_m: 0.6120 - val_loss: 0.6786 - val_acc: 0.5543 - val_precision_m: 0.7222 - val_recall_m: 0.2149\n",
      "Epoch 99/5000\n",
      " - 10s - loss: 0.6674 - acc: 0.6141 - precision_m: 0.6366 - recall_m: 0.6039 - val_loss: 0.6706 - val_acc: 0.6080 - val_precision_m: 0.6063 - val_recall_m: 0.7242\n",
      "Epoch 100/5000\n",
      " - 10s - loss: 0.6666 - acc: 0.6185 - precision_m: 0.6450 - recall_m: 0.6127 - val_loss: 0.6709 - val_acc: 0.6163 - val_precision_m: 0.6989 - val_recall_m: 0.4675\n",
      "Epoch 101/5000\n",
      " - 10s - loss: 0.6647 - acc: 0.6289 - precision_m: 0.6498 - recall_m: 0.6163 - val_loss: 0.6684 - val_acc: 0.6035 - val_precision_m: 0.5876 - val_recall_m: 0.8209\n",
      "Epoch 102/5000\n",
      " - 10s - loss: 0.6645 - acc: 0.6229 - precision_m: 0.6497 - recall_m: 0.6193 - val_loss: 0.6691 - val_acc: 0.6085 - val_precision_m: 0.7194 - val_recall_m: 0.4056\n",
      "Epoch 103/5000\n",
      " - 10s - loss: 0.6635 - acc: 0.6289 - precision_m: 0.6545 - recall_m: 0.6250 - val_loss: 0.6687 - val_acc: 0.5792 - val_precision_m: 0.5630 - val_recall_m: 0.9011\n",
      "Epoch 104/5000\n",
      " - 10s - loss: 0.6624 - acc: 0.6282 - precision_m: 0.6451 - recall_m: 0.6239 - val_loss: 0.6681 - val_acc: 0.6096 - val_precision_m: 0.7408 - val_recall_m: 0.3722\n",
      "Epoch 105/5000\n",
      " - 10s - loss: 0.6621 - acc: 0.6324 - precision_m: 0.6576 - recall_m: 0.6230 - val_loss: 0.6643 - val_acc: 0.6412 - val_precision_m: 0.6916 - val_recall_m: 0.5885\n",
      "Epoch 106/5000\n",
      " - 10s - loss: 0.6599 - acc: 0.6352 - precision_m: 0.6549 - recall_m: 0.6248 - val_loss: 0.6636 - val_acc: 0.6445 - val_precision_m: 0.6640 - val_recall_m: 0.6551\n",
      "Epoch 107/5000\n",
      " - 10s - loss: 0.6594 - acc: 0.6409 - precision_m: 0.6584 - recall_m: 0.6280 - val_loss: 0.6621 - val_acc: 0.6423 - val_precision_m: 0.6519 - val_recall_m: 0.6490\n",
      "Epoch 108/5000\n",
      " - 10s - loss: 0.6585 - acc: 0.6393 - precision_m: 0.6586 - recall_m: 0.6289 - val_loss: 0.6612 - val_acc: 0.6512 - val_precision_m: 0.7056 - val_recall_m: 0.5812\n",
      "Epoch 109/5000\n",
      " - 10s - loss: 0.6576 - acc: 0.6446 - precision_m: 0.6566 - recall_m: 0.6497 - val_loss: 0.6731 - val_acc: 0.5653 - val_precision_m: 0.7989 - val_recall_m: 0.1979\n",
      "Epoch 110/5000\n",
      " - 10s - loss: 0.6578 - acc: 0.6281 - precision_m: 0.6772 - recall_m: 0.6116 - val_loss: 0.6590 - val_acc: 0.6561 - val_precision_m: 0.6542 - val_recall_m: 0.6942\n",
      "Epoch 111/5000\n",
      " - 10s - loss: 0.6552 - acc: 0.6487 - precision_m: 0.6638 - recall_m: 0.6505 - val_loss: 0.6587 - val_acc: 0.6512 - val_precision_m: 0.6633 - val_recall_m: 0.6941\n",
      "Epoch 112/5000\n",
      " - 10s - loss: 0.6550 - acc: 0.6464 - precision_m: 0.6688 - recall_m: 0.6353 - val_loss: 0.6585 - val_acc: 0.6190 - val_precision_m: 0.6109 - val_recall_m: 0.7310\n",
      "Epoch 113/5000\n",
      " - 10s - loss: 0.6539 - acc: 0.6462 - precision_m: 0.6723 - recall_m: 0.6383 - val_loss: 0.6556 - val_acc: 0.6617 - val_precision_m: 0.6895 - val_recall_m: 0.6121\n",
      "Epoch 114/5000\n",
      " - 10s - loss: 0.6512 - acc: 0.6559 - precision_m: 0.6741 - recall_m: 0.6486 - val_loss: 0.6582 - val_acc: 0.6473 - val_precision_m: 0.7804 - val_recall_m: 0.4438\n",
      "Epoch 115/5000\n",
      " - 10s - loss: 0.6509 - acc: 0.6602 - precision_m: 0.6706 - recall_m: 0.6609 - val_loss: 0.6620 - val_acc: 0.6196 - val_precision_m: 0.8007 - val_recall_m: 0.3446\n",
      "Epoch 116/5000\n",
      " - 10s - loss: 0.6497 - acc: 0.6553 - precision_m: 0.6807 - recall_m: 0.6341 - val_loss: 0.6533 - val_acc: 0.6434 - val_precision_m: 0.6761 - val_recall_m: 0.6092\n",
      "Epoch 117/5000\n",
      " - 10s - loss: 0.6501 - acc: 0.6497 - precision_m: 0.6834 - recall_m: 0.6367 - val_loss: 0.6538 - val_acc: 0.6074 - val_precision_m: 0.5798 - val_recall_m: 0.8827\n",
      "Epoch 118/5000\n",
      " - 10s - loss: 0.6476 - acc: 0.6616 - precision_m: 0.6806 - recall_m: 0.6508 - val_loss: 0.6531 - val_acc: 0.5947 - val_precision_m: 0.5729 - val_recall_m: 0.9042\n",
      "Epoch 119/5000\n",
      " - 10s - loss: 0.6460 - acc: 0.6622 - precision_m: 0.6914 - recall_m: 0.6512 - val_loss: 0.6492 - val_acc: 0.6595 - val_precision_m: 0.6512 - val_recall_m: 0.6920\n",
      "Epoch 120/5000\n",
      " - 10s - loss: 0.6466 - acc: 0.6613 - precision_m: 0.6998 - recall_m: 0.6486 - val_loss: 0.6478 - val_acc: 0.6611 - val_precision_m: 0.7431 - val_recall_m: 0.5113\n",
      "Epoch 121/5000\n",
      " - 10s - loss: 0.6442 - acc: 0.6604 - precision_m: 0.7004 - recall_m: 0.6517 - val_loss: 0.6445 - val_acc: 0.6855 - val_precision_m: 0.6737 - val_recall_m: 0.7404\n",
      "Epoch 122/5000\n",
      " - 10s - loss: 0.6432 - acc: 0.6666 - precision_m: 0.6948 - recall_m: 0.6499 - val_loss: 0.6460 - val_acc: 0.6722 - val_precision_m: 0.6976 - val_recall_m: 0.6408\n",
      "Epoch 123/5000\n",
      " - 10s - loss: 0.6407 - acc: 0.6838 - precision_m: 0.7021 - recall_m: 0.6656 - val_loss: 0.6421 - val_acc: 0.6772 - val_precision_m: 0.6802 - val_recall_m: 0.6916\n",
      "Epoch 124/5000\n",
      " - 10s - loss: 0.6405 - acc: 0.6722 - precision_m: 0.6967 - recall_m: 0.6637 - val_loss: 0.6402 - val_acc: 0.6872 - val_precision_m: 0.7199 - val_recall_m: 0.6526\n",
      "Epoch 125/5000\n",
      " - 10s - loss: 0.6393 - acc: 0.6743 - precision_m: 0.7081 - recall_m: 0.6544 - val_loss: 0.6522 - val_acc: 0.6218 - val_precision_m: 0.8943 - val_recall_m: 0.2982\n",
      "Epoch 126/5000\n",
      " - 10s - loss: 0.6388 - acc: 0.6708 - precision_m: 0.7092 - recall_m: 0.6513 - val_loss: 0.6455 - val_acc: 0.6578 - val_precision_m: 0.8703 - val_recall_m: 0.3859\n",
      "Epoch 127/5000\n",
      " - 10s - loss: 0.6371 - acc: 0.6829 - precision_m: 0.7190 - recall_m: 0.6635 - val_loss: 0.6371 - val_acc: 0.6921 - val_precision_m: 0.7586 - val_recall_m: 0.5861\n",
      "Epoch 128/5000\n",
      " - 10s - loss: 0.6353 - acc: 0.6826 - precision_m: 0.7090 - recall_m: 0.6742 - val_loss: 0.6387 - val_acc: 0.7004 - val_precision_m: 0.7838 - val_recall_m: 0.5834\n",
      "Epoch 129/5000\n",
      " - 10s - loss: 0.6331 - acc: 0.6870 - precision_m: 0.7077 - recall_m: 0.6721 - val_loss: 0.6341 - val_acc: 0.6645 - val_precision_m: 0.6333 - val_recall_m: 0.8120\n",
      "Epoch 130/5000\n",
      " - 10s - loss: 0.6325 - acc: 0.6858 - precision_m: 0.7145 - recall_m: 0.6766 - val_loss: 0.6323 - val_acc: 0.7182 - val_precision_m: 0.8148 - val_recall_m: 0.5932\n",
      "Epoch 131/5000\n",
      " - 10s - loss: 0.6342 - acc: 0.6741 - precision_m: 0.7333 - recall_m: 0.6516 - val_loss: 0.6295 - val_acc: 0.7032 - val_precision_m: 0.7189 - val_recall_m: 0.6871\n",
      "Epoch 132/5000\n",
      " - 10s - loss: 0.6276 - acc: 0.7084 - precision_m: 0.7270 - recall_m: 0.6903 - val_loss: 0.6286 - val_acc: 0.7143 - val_precision_m: 0.7737 - val_recall_m: 0.6232\n",
      "Epoch 133/5000\n",
      " - 10s - loss: 0.6293 - acc: 0.6806 - precision_m: 0.7229 - recall_m: 0.6668 - val_loss: 0.6365 - val_acc: 0.6096 - val_precision_m: 0.5797 - val_recall_m: 0.8983\n",
      "Epoch 134/5000\n",
      " - 10s - loss: 0.6261 - acc: 0.7022 - precision_m: 0.7263 - recall_m: 0.6854 - val_loss: 0.6266 - val_acc: 0.6927 - val_precision_m: 0.6751 - val_recall_m: 0.7238\n",
      "Epoch 135/5000\n",
      " - 10s - loss: 0.6250 - acc: 0.7030 - precision_m: 0.7261 - recall_m: 0.6897 - val_loss: 0.6233 - val_acc: 0.7259 - val_precision_m: 0.7416 - val_recall_m: 0.7105\n",
      "Epoch 136/5000\n",
      " - 10s - loss: 0.6231 - acc: 0.7047 - precision_m: 0.7316 - recall_m: 0.6863 - val_loss: 0.6270 - val_acc: 0.7182 - val_precision_m: 0.8669 - val_recall_m: 0.5379\n",
      "Epoch 137/5000\n",
      " - 10s - loss: 0.6222 - acc: 0.7090 - precision_m: 0.7456 - recall_m: 0.6791 - val_loss: 0.6232 - val_acc: 0.7076 - val_precision_m: 0.6941 - val_recall_m: 0.7530\n",
      "Epoch 138/5000\n",
      " - 10s - loss: 0.6202 - acc: 0.7081 - precision_m: 0.7377 - recall_m: 0.6890 - val_loss: 0.6212 - val_acc: 0.7226 - val_precision_m: 0.7569 - val_recall_m: 0.6866\n",
      "Epoch 139/5000\n",
      " - 10s - loss: 0.6190 - acc: 0.7062 - precision_m: 0.7425 - recall_m: 0.6839 - val_loss: 0.6184 - val_acc: 0.7176 - val_precision_m: 0.7205 - val_recall_m: 0.7091\n",
      "Epoch 140/5000\n",
      " - 10s - loss: 0.6198 - acc: 0.6934 - precision_m: 0.7448 - recall_m: 0.6807 - val_loss: 0.6152 - val_acc: 0.7558 - val_precision_m: 0.8247 - val_recall_m: 0.6797\n",
      "Epoch 141/5000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " - 10s - loss: 0.6134 - acc: 0.7318 - precision_m: 0.7573 - recall_m: 0.7019 - val_loss: 0.6191 - val_acc: 0.7154 - val_precision_m: 0.8974 - val_recall_m: 0.4987\n",
      "Epoch 142/5000\n",
      " - 10s - loss: 0.6142 - acc: 0.7151 - precision_m: 0.7537 - recall_m: 0.6889 - val_loss: 0.6140 - val_acc: 0.7403 - val_precision_m: 0.8320 - val_recall_m: 0.6173\n",
      "Epoch 143/5000\n",
      " - 10s - loss: 0.6119 - acc: 0.7231 - precision_m: 0.7538 - recall_m: 0.7041 - val_loss: 0.6184 - val_acc: 0.7099 - val_precision_m: 0.9227 - val_recall_m: 0.4797\n",
      "Epoch 144/5000\n",
      " - 10s - loss: 0.6134 - acc: 0.7038 - precision_m: 0.7551 - recall_m: 0.6785 - val_loss: 0.6094 - val_acc: 0.7575 - val_precision_m: 0.8069 - val_recall_m: 0.6853\n",
      "Epoch 145/5000\n",
      " - 10s - loss: 0.6096 - acc: 0.7278 - precision_m: 0.7598 - recall_m: 0.7004 - val_loss: 0.6103 - val_acc: 0.7447 - val_precision_m: 0.8418 - val_recall_m: 0.6313\n",
      "Epoch 146/5000\n",
      " - 10s - loss: 0.6071 - acc: 0.7314 - precision_m: 0.7720 - recall_m: 0.6970 - val_loss: 0.6071 - val_acc: 0.7115 - val_precision_m: 0.6833 - val_recall_m: 0.8084\n",
      "Epoch 147/5000\n",
      " - 10s - loss: 0.6048 - acc: 0.7361 - precision_m: 0.7652 - recall_m: 0.7080 - val_loss: 0.6034 - val_acc: 0.7486 - val_precision_m: 0.7508 - val_recall_m: 0.7567\n",
      "Epoch 148/5000\n",
      " - 10s - loss: 0.6052 - acc: 0.7292 - precision_m: 0.7657 - recall_m: 0.7095 - val_loss: 0.6071 - val_acc: 0.7392 - val_precision_m: 0.9250 - val_recall_m: 0.5308\n",
      "Epoch 149/5000\n",
      " - 10s - loss: 0.6017 - acc: 0.7374 - precision_m: 0.7828 - recall_m: 0.7008 - val_loss: 0.5993 - val_acc: 0.7425 - val_precision_m: 0.7413 - val_recall_m: 0.7449\n",
      "Epoch 150/5000\n",
      " - 10s - loss: 0.6015 - acc: 0.7302 - precision_m: 0.7738 - recall_m: 0.7029 - val_loss: 0.5991 - val_acc: 0.7614 - val_precision_m: 0.7520 - val_recall_m: 0.7801\n",
      "Epoch 151/5000\n",
      " - 10s - loss: 0.6003 - acc: 0.7336 - precision_m: 0.7742 - recall_m: 0.7061 - val_loss: 0.6036 - val_acc: 0.7364 - val_precision_m: 0.8502 - val_recall_m: 0.5960\n",
      "Epoch 152/5000\n",
      " - 10s - loss: 0.5976 - acc: 0.7378 - precision_m: 0.7698 - recall_m: 0.7084 - val_loss: 0.5991 - val_acc: 0.7586 - val_precision_m: 0.9028 - val_recall_m: 0.5952\n",
      "Epoch 153/5000\n",
      " - 10s - loss: 0.5945 - acc: 0.7430 - precision_m: 0.7807 - recall_m: 0.7106 - val_loss: 0.5929 - val_acc: 0.7575 - val_precision_m: 0.7489 - val_recall_m: 0.7764\n",
      "Epoch 154/5000\n",
      " - 10s - loss: 0.5968 - acc: 0.7363 - precision_m: 0.7795 - recall_m: 0.7033 - val_loss: 0.5916 - val_acc: 0.7680 - val_precision_m: 0.7854 - val_recall_m: 0.7459\n",
      "Epoch 155/5000\n",
      " - 10s - loss: 0.5916 - acc: 0.7494 - precision_m: 0.7815 - recall_m: 0.7207 - val_loss: 0.5908 - val_acc: 0.7636 - val_precision_m: 0.8747 - val_recall_m: 0.6329\n",
      "Epoch 156/5000\n",
      " - 10s - loss: 0.5899 - acc: 0.7529 - precision_m: 0.7996 - recall_m: 0.7107 - val_loss: 0.6002 - val_acc: 0.6390 - val_precision_m: 0.6002 - val_recall_m: 0.9120\n",
      "Epoch 157/5000\n",
      " - 10s - loss: 0.5881 - acc: 0.7507 - precision_m: 0.7889 - recall_m: 0.7178 - val_loss: 0.5884 - val_acc: 0.7685 - val_precision_m: 0.8665 - val_recall_m: 0.6296\n",
      "Epoch 158/5000\n",
      " - 10s - loss: 0.5875 - acc: 0.7475 - precision_m: 0.7881 - recall_m: 0.7121 - val_loss: 0.5867 - val_acc: 0.7436 - val_precision_m: 0.7133 - val_recall_m: 0.8406\n",
      "Epoch 159/5000\n",
      " - 10s - loss: 0.5847 - acc: 0.7573 - precision_m: 0.7968 - recall_m: 0.7208 - val_loss: 0.5821 - val_acc: 0.7896 - val_precision_m: 0.8661 - val_recall_m: 0.6866\n",
      "Epoch 160/5000\n",
      " - 10s - loss: 0.5837 - acc: 0.7596 - precision_m: 0.7935 - recall_m: 0.7270 - val_loss: 0.5835 - val_acc: 0.7674 - val_precision_m: 0.8684 - val_recall_m: 0.6460\n",
      "Epoch 161/5000\n",
      " - 10s - loss: 0.5812 - acc: 0.7654 - precision_m: 0.8032 - recall_m: 0.7249 - val_loss: 0.5928 - val_acc: 0.6434 - val_precision_m: 0.6020 - val_recall_m: 0.9204\n",
      "Epoch 162/5000\n",
      " - 10s - loss: 0.5807 - acc: 0.7530 - precision_m: 0.8045 - recall_m: 0.7231 - val_loss: 0.5775 - val_acc: 0.7946 - val_precision_m: 0.8287 - val_recall_m: 0.7452\n",
      "Epoch 163/5000\n",
      " - 10s - loss: 0.5778 - acc: 0.7638 - precision_m: 0.8028 - recall_m: 0.7252 - val_loss: 0.5728 - val_acc: 0.7973 - val_precision_m: 0.8230 - val_recall_m: 0.7575\n",
      "Epoch 164/5000\n",
      " - 10s - loss: 0.5790 - acc: 0.7542 - precision_m: 0.8022 - recall_m: 0.7263 - val_loss: 0.5739 - val_acc: 0.7868 - val_precision_m: 0.9303 - val_recall_m: 0.6236\n",
      "Epoch 165/5000\n",
      " - 10s - loss: 0.5762 - acc: 0.7620 - precision_m: 0.8073 - recall_m: 0.7278 - val_loss: 0.5793 - val_acc: 0.7530 - val_precision_m: 0.9553 - val_recall_m: 0.5386\n",
      "Epoch 166/5000\n",
      " - 10s - loss: 0.5754 - acc: 0.7608 - precision_m: 0.8043 - recall_m: 0.7266 - val_loss: 0.5717 - val_acc: 0.7824 - val_precision_m: 0.9162 - val_recall_m: 0.6325\n",
      "Epoch 167/5000\n",
      " - 10s - loss: 0.5707 - acc: 0.7717 - precision_m: 0.8141 - recall_m: 0.7315 - val_loss: 0.5685 - val_acc: 0.7719 - val_precision_m: 0.7338 - val_recall_m: 0.8345\n",
      "Epoch 168/5000\n",
      " - 10s - loss: 0.5682 - acc: 0.7769 - precision_m: 0.8195 - recall_m: 0.7350 - val_loss: 0.5661 - val_acc: 0.8012 - val_precision_m: 0.8741 - val_recall_m: 0.7160\n",
      "Epoch 169/5000\n",
      " - 10s - loss: 0.5647 - acc: 0.7785 - precision_m: 0.8210 - recall_m: 0.7347 - val_loss: 0.5717 - val_acc: 0.7259 - val_precision_m: 0.6774 - val_recall_m: 0.8817\n",
      "Epoch 170/5000\n",
      " - 10s - loss: 0.5662 - acc: 0.7745 - precision_m: 0.8188 - recall_m: 0.7328 - val_loss: 0.5621 - val_acc: 0.8140 - val_precision_m: 0.8956 - val_recall_m: 0.7204\n",
      "Epoch 171/5000\n",
      " - 10s - loss: 0.5624 - acc: 0.7866 - precision_m: 0.8231 - recall_m: 0.7449 - val_loss: 0.5627 - val_acc: 0.7935 - val_precision_m: 0.8901 - val_recall_m: 0.6822\n",
      "Epoch 172/5000\n",
      " - 10s - loss: 0.5600 - acc: 0.7889 - precision_m: 0.8322 - recall_m: 0.7412 - val_loss: 0.5645 - val_acc: 0.7575 - val_precision_m: 0.7186 - val_recall_m: 0.8799\n",
      "Epoch 173/5000\n",
      " - 10s - loss: 0.5597 - acc: 0.7793 - precision_m: 0.8210 - recall_m: 0.7407 - val_loss: 0.5752 - val_acc: 0.6589 - val_precision_m: 0.6128 - val_recall_m: 0.9234\n",
      "Epoch 174/5000\n",
      " - 10s - loss: 0.5604 - acc: 0.7691 - precision_m: 0.8223 - recall_m: 0.7298 - val_loss: 0.5565 - val_acc: 0.7968 - val_precision_m: 0.7864 - val_recall_m: 0.8075\n",
      "Epoch 175/5000\n",
      " - 10s - loss: 0.5572 - acc: 0.7829 - precision_m: 0.8285 - recall_m: 0.7434 - val_loss: 0.5558 - val_acc: 0.7946 - val_precision_m: 0.9666 - val_recall_m: 0.6191\n",
      "Epoch 176/5000\n",
      " - 10s - loss: 0.5575 - acc: 0.7791 - precision_m: 0.8299 - recall_m: 0.7382 - val_loss: 0.5502 - val_acc: 0.8261 - val_precision_m: 0.8954 - val_recall_m: 0.7460\n",
      "Epoch 177/5000\n",
      " - 10s - loss: 0.5531 - acc: 0.7867 - precision_m: 0.8296 - recall_m: 0.7458 - val_loss: 0.5559 - val_acc: 0.7835 - val_precision_m: 0.9557 - val_recall_m: 0.5958\n",
      "Epoch 178/5000\n",
      " - 10s - loss: 0.5520 - acc: 0.7891 - precision_m: 0.8345 - recall_m: 0.7444 - val_loss: 0.5486 - val_acc: 0.8112 - val_precision_m: 0.8329 - val_recall_m: 0.7779\n",
      "Epoch 179/5000\n",
      " - 10s - loss: 0.5523 - acc: 0.7789 - precision_m: 0.8319 - recall_m: 0.7416 - val_loss: 0.5472 - val_acc: 0.8106 - val_precision_m: 0.9343 - val_recall_m: 0.6760\n",
      "Epoch 180/5000\n",
      " - 10s - loss: 0.5487 - acc: 0.7898 - precision_m: 0.8359 - recall_m: 0.7431 - val_loss: 0.5444 - val_acc: 0.8073 - val_precision_m: 0.8142 - val_recall_m: 0.7963\n",
      "Epoch 181/5000\n",
      " - 10s - loss: 0.5493 - acc: 0.7841 - precision_m: 0.8329 - recall_m: 0.7413 - val_loss: 0.5395 - val_acc: 0.8306 - val_precision_m: 0.9131 - val_recall_m: 0.7362\n",
      "Epoch 182/5000\n",
      " - 10s - loss: 0.5430 - acc: 0.8021 - precision_m: 0.8437 - recall_m: 0.7548 - val_loss: 0.5398 - val_acc: 0.8300 - val_precision_m: 0.8862 - val_recall_m: 0.7531\n",
      "Epoch 183/5000\n",
      " - 10s - loss: 0.5421 - acc: 0.7949 - precision_m: 0.8425 - recall_m: 0.7475 - val_loss: 0.5403 - val_acc: 0.8178 - val_precision_m: 0.8415 - val_recall_m: 0.7804\n",
      "Epoch 184/5000\n",
      " - 10s - loss: 0.5425 - acc: 0.7909 - precision_m: 0.8397 - recall_m: 0.7471 - val_loss: 0.5373 - val_acc: 0.8300 - val_precision_m: 0.9192 - val_recall_m: 0.7303\n",
      "Epoch 185/5000\n",
      " - 10s - loss: 0.5386 - acc: 0.8032 - precision_m: 0.8514 - recall_m: 0.7526 - val_loss: 0.5356 - val_acc: 0.8311 - val_precision_m: 0.9061 - val_recall_m: 0.7344\n",
      "Epoch 186/5000\n",
      " - 10s - loss: 0.5381 - acc: 0.8046 - precision_m: 0.8493 - recall_m: 0.7561 - val_loss: 0.5294 - val_acc: 0.8378 - val_precision_m: 0.8874 - val_recall_m: 0.7795\n",
      "Epoch 187/5000\n",
      " - 10s - loss: 0.5344 - acc: 0.8063 - precision_m: 0.8570 - recall_m: 0.7529 - val_loss: 0.5321 - val_acc: 0.8073 - val_precision_m: 0.7899 - val_recall_m: 0.8479\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 188/5000\n",
      " - 10s - loss: 0.5333 - acc: 0.8028 - precision_m: 0.8448 - recall_m: 0.7591 - val_loss: 0.5290 - val_acc: 0.8322 - val_precision_m: 0.8507 - val_recall_m: 0.8106\n",
      "Epoch 189/5000\n",
      " - 10s - loss: 0.5333 - acc: 0.8075 - precision_m: 0.8539 - recall_m: 0.7550 - val_loss: 0.5387 - val_acc: 0.7370 - val_precision_m: 0.6906 - val_recall_m: 0.8908\n",
      "Epoch 190/5000\n",
      " - 10s - loss: 0.5338 - acc: 0.8016 - precision_m: 0.8588 - recall_m: 0.7475 - val_loss: 0.5263 - val_acc: 0.8378 - val_precision_m: 0.9288 - val_recall_m: 0.7426\n",
      "Epoch 191/5000\n",
      " - 10s - loss: 0.5291 - acc: 0.8097 - precision_m: 0.8556 - recall_m: 0.7647 - val_loss: 0.5397 - val_acc: 0.7652 - val_precision_m: 0.9745 - val_recall_m: 0.5568\n",
      "Epoch 192/5000\n",
      " - 10s - loss: 0.5261 - acc: 0.8135 - precision_m: 0.8610 - recall_m: 0.7618 - val_loss: 0.5199 - val_acc: 0.8466 - val_precision_m: 0.9107 - val_recall_m: 0.7723\n",
      "Epoch 193/5000\n",
      " - 10s - loss: 0.5225 - acc: 0.8163 - precision_m: 0.8642 - recall_m: 0.7606 - val_loss: 0.5273 - val_acc: 0.7636 - val_precision_m: 0.7211 - val_recall_m: 0.8669\n",
      "Epoch 194/5000\n",
      " - 10s - loss: 0.5226 - acc: 0.8118 - precision_m: 0.8575 - recall_m: 0.7623 - val_loss: 0.5180 - val_acc: 0.8383 - val_precision_m: 0.8650 - val_recall_m: 0.8049\n",
      "Epoch 195/5000\n",
      " - 10s - loss: 0.5270 - acc: 0.7960 - precision_m: 0.8476 - recall_m: 0.7556 - val_loss: 0.5462 - val_acc: 0.7359 - val_precision_m: 0.9754 - val_recall_m: 0.4998\n",
      "Epoch 196/5000\n",
      " - 10s - loss: 0.5209 - acc: 0.8097 - precision_m: 0.8600 - recall_m: 0.7574 - val_loss: 0.5164 - val_acc: 0.8256 - val_precision_m: 0.8325 - val_recall_m: 0.8109\n",
      "Epoch 197/5000\n",
      " - 10s - loss: 0.5180 - acc: 0.8186 - precision_m: 0.8649 - recall_m: 0.7663 - val_loss: 0.5117 - val_acc: 0.8461 - val_precision_m: 0.9272 - val_recall_m: 0.7618\n",
      "Epoch 198/5000\n",
      " - 10s - loss: 0.5196 - acc: 0.8040 - precision_m: 0.8583 - recall_m: 0.7559 - val_loss: 0.5154 - val_acc: 0.8189 - val_precision_m: 0.9374 - val_recall_m: 0.6898\n",
      "Epoch 199/5000\n",
      " - 10s - loss: 0.5187 - acc: 0.8005 - precision_m: 0.8512 - recall_m: 0.7592 - val_loss: 0.5128 - val_acc: 0.8328 - val_precision_m: 0.9268 - val_recall_m: 0.7271\n",
      "Epoch 200/5000\n",
      " - 10s - loss: 0.5124 - acc: 0.8258 - precision_m: 0.8760 - recall_m: 0.7700 - val_loss: 0.5078 - val_acc: 0.8455 - val_precision_m: 0.9139 - val_recall_m: 0.7732\n",
      "Epoch 201/5000\n",
      " - 10s - loss: 0.5150 - acc: 0.8113 - precision_m: 0.8623 - recall_m: 0.7602 - val_loss: 0.5046 - val_acc: 0.8527 - val_precision_m: 0.9075 - val_recall_m: 0.7961\n",
      "Epoch 202/5000\n",
      " - 10s - loss: 0.5088 - acc: 0.8235 - precision_m: 0.8726 - recall_m: 0.7706 - val_loss: 0.5040 - val_acc: 0.8483 - val_precision_m: 0.9015 - val_recall_m: 0.7932\n",
      "Epoch 203/5000\n",
      " - 10s - loss: 0.5098 - acc: 0.8216 - precision_m: 0.8735 - recall_m: 0.7672 - val_loss: 0.5154 - val_acc: 0.7791 - val_precision_m: 0.7414 - val_recall_m: 0.8773\n",
      "Epoch 204/5000\n",
      " - 10s - loss: 0.5077 - acc: 0.8227 - precision_m: 0.8689 - recall_m: 0.7767 - val_loss: 0.5049 - val_acc: 0.8367 - val_precision_m: 0.9638 - val_recall_m: 0.7024\n",
      "Epoch 205/5000\n",
      " - 10s - loss: 0.5041 - acc: 0.8263 - precision_m: 0.8721 - recall_m: 0.7755 - val_loss: 0.5038 - val_acc: 0.8162 - val_precision_m: 0.8024 - val_recall_m: 0.8469\n",
      "Epoch 206/5000\n",
      " - 10s - loss: 0.5045 - acc: 0.8183 - precision_m: 0.8665 - recall_m: 0.7657 - val_loss: 0.5126 - val_acc: 0.7984 - val_precision_m: 0.9749 - val_recall_m: 0.6135\n",
      "Epoch 207/5000\n",
      " - 10s - loss: 0.5018 - acc: 0.8225 - precision_m: 0.8750 - recall_m: 0.7665 - val_loss: 0.4999 - val_acc: 0.8350 - val_precision_m: 0.8418 - val_recall_m: 0.8298\n",
      "Epoch 208/5000\n",
      " - 10s - loss: 0.5010 - acc: 0.8250 - precision_m: 0.8781 - recall_m: 0.7662 - val_loss: 0.4985 - val_acc: 0.8411 - val_precision_m: 0.8785 - val_recall_m: 0.7844\n",
      "Epoch 209/5000\n",
      " - 10s - loss: 0.5008 - acc: 0.8243 - precision_m: 0.8752 - recall_m: 0.7673 - val_loss: 0.4943 - val_acc: 0.8494 - val_precision_m: 0.8949 - val_recall_m: 0.7835\n",
      "Epoch 210/5000\n",
      " - 10s - loss: 0.4963 - acc: 0.8254 - precision_m: 0.8722 - recall_m: 0.7746 - val_loss: 0.4939 - val_acc: 0.8378 - val_precision_m: 0.9369 - val_recall_m: 0.7277\n",
      "Epoch 211/5000\n",
      " - 10s - loss: 0.5001 - acc: 0.8122 - precision_m: 0.8637 - recall_m: 0.7665 - val_loss: 0.4873 - val_acc: 0.8555 - val_precision_m: 0.9294 - val_recall_m: 0.7699\n",
      "Epoch 212/5000\n",
      " - 10s - loss: 0.4946 - acc: 0.8261 - precision_m: 0.8762 - recall_m: 0.7705 - val_loss: 0.4898 - val_acc: 0.8555 - val_precision_m: 0.9430 - val_recall_m: 0.7585\n",
      "Epoch 213/5000\n",
      " - 10s - loss: 0.4960 - acc: 0.8199 - precision_m: 0.8728 - recall_m: 0.7697 - val_loss: 0.4869 - val_acc: 0.8605 - val_precision_m: 0.9466 - val_recall_m: 0.7637\n",
      "Epoch 214/5000\n",
      " - 10s - loss: 0.4919 - acc: 0.8314 - precision_m: 0.8798 - recall_m: 0.7769 - val_loss: 0.4865 - val_acc: 0.8566 - val_precision_m: 0.9078 - val_recall_m: 0.7844\n",
      "Epoch 215/5000\n",
      " - 10s - loss: 0.4896 - acc: 0.8307 - precision_m: 0.8786 - recall_m: 0.7776 - val_loss: 0.4871 - val_acc: 0.8444 - val_precision_m: 0.8675 - val_recall_m: 0.8156\n",
      "Epoch 216/5000\n",
      " - 10s - loss: 0.4867 - acc: 0.8376 - precision_m: 0.8872 - recall_m: 0.7799 - val_loss: 0.4808 - val_acc: 0.8616 - val_precision_m: 0.9211 - val_recall_m: 0.7918\n",
      "Epoch 217/5000\n",
      " - 10s - loss: 0.4841 - acc: 0.8396 - precision_m: 0.8890 - recall_m: 0.7830 - val_loss: 0.4810 - val_acc: 0.8522 - val_precision_m: 0.8826 - val_recall_m: 0.8126\n",
      "Epoch 218/5000\n",
      " - 10s - loss: 0.4851 - acc: 0.8356 - precision_m: 0.8858 - recall_m: 0.7785 - val_loss: 0.4904 - val_acc: 0.8068 - val_precision_m: 0.7798 - val_recall_m: 0.8659\n",
      "Epoch 219/5000\n",
      " - 10s - loss: 0.4819 - acc: 0.8341 - precision_m: 0.8859 - recall_m: 0.7775 - val_loss: 0.4992 - val_acc: 0.7984 - val_precision_m: 0.9807 - val_recall_m: 0.6158\n",
      "Epoch 220/5000\n",
      " - 10s - loss: 0.4820 - acc: 0.8331 - precision_m: 0.8855 - recall_m: 0.7762 - val_loss: 0.4781 - val_acc: 0.8477 - val_precision_m: 0.8765 - val_recall_m: 0.8211\n",
      "Epoch 221/5000\n",
      " - 10s - loss: 0.4799 - acc: 0.8400 - precision_m: 0.8899 - recall_m: 0.7836 - val_loss: 0.4860 - val_acc: 0.8178 - val_precision_m: 0.9581 - val_recall_m: 0.6718\n",
      "Epoch 222/5000\n",
      " - 10s - loss: 0.4775 - acc: 0.8415 - precision_m: 0.8899 - recall_m: 0.7821 - val_loss: 0.4699 - val_acc: 0.8571 - val_precision_m: 0.9305 - val_recall_m: 0.7733\n",
      "Epoch 223/5000\n",
      " - 10s - loss: 0.4765 - acc: 0.8371 - precision_m: 0.8903 - recall_m: 0.7807 - val_loss: 0.4738 - val_acc: 0.8516 - val_precision_m: 0.9545 - val_recall_m: 0.7393\n",
      "Epoch 224/5000\n",
      " - 10s - loss: 0.4776 - acc: 0.8223 - precision_m: 0.8763 - recall_m: 0.7691 - val_loss: 0.4684 - val_acc: 0.8616 - val_precision_m: 0.9300 - val_recall_m: 0.7825\n",
      "Epoch 225/5000\n",
      " - 10s - loss: 0.4740 - acc: 0.8386 - precision_m: 0.8882 - recall_m: 0.7812 - val_loss: 0.4675 - val_acc: 0.8544 - val_precision_m: 0.8851 - val_recall_m: 0.8049\n",
      "Epoch 226/5000\n",
      " - 10s - loss: 0.4719 - acc: 0.8385 - precision_m: 0.8901 - recall_m: 0.7817 - val_loss: 0.4692 - val_acc: 0.8511 - val_precision_m: 0.8585 - val_recall_m: 0.8407\n",
      "Epoch 227/5000\n",
      " - 10s - loss: 0.4728 - acc: 0.8381 - precision_m: 0.8889 - recall_m: 0.7833 - val_loss: 0.4756 - val_acc: 0.8289 - val_precision_m: 0.9616 - val_recall_m: 0.6898\n",
      "Epoch 228/5000\n",
      " - 10s - loss: 0.4691 - acc: 0.8414 - precision_m: 0.8927 - recall_m: 0.7810 - val_loss: 0.4644 - val_acc: 0.8538 - val_precision_m: 0.9336 - val_recall_m: 0.7701\n",
      "Epoch 229/5000\n",
      " - 10s - loss: 0.4677 - acc: 0.8410 - precision_m: 0.8919 - recall_m: 0.7837 - val_loss: 0.4695 - val_acc: 0.8261 - val_precision_m: 0.8065 - val_recall_m: 0.8638\n",
      "Epoch 230/5000\n",
      " - 10s - loss: 0.4680 - acc: 0.8376 - precision_m: 0.8870 - recall_m: 0.7823 - val_loss: 0.4685 - val_acc: 0.8461 - val_precision_m: 0.9762 - val_recall_m: 0.7107\n",
      "Epoch 231/5000\n",
      " - 10s - loss: 0.4657 - acc: 0.8430 - precision_m: 0.8912 - recall_m: 0.7876 - val_loss: 0.4565 - val_acc: 0.8638 - val_precision_m: 0.9145 - val_recall_m: 0.8103\n",
      "Epoch 232/5000\n",
      " - 10s - loss: 0.4643 - acc: 0.8427 - precision_m: 0.8888 - recall_m: 0.7902 - val_loss: 0.4590 - val_acc: 0.8522 - val_precision_m: 0.8719 - val_recall_m: 0.8253\n",
      "Epoch 233/5000\n",
      " - 10s - loss: 0.4662 - acc: 0.8352 - precision_m: 0.8876 - recall_m: 0.7809 - val_loss: 0.4606 - val_acc: 0.8494 - val_precision_m: 0.8856 - val_recall_m: 0.8204\n",
      "Epoch 234/5000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " - 10s - loss: 0.4584 - acc: 0.8464 - precision_m: 0.8957 - recall_m: 0.7897 - val_loss: 0.4546 - val_acc: 0.8677 - val_precision_m: 0.9260 - val_recall_m: 0.8065\n",
      "Epoch 235/5000\n",
      " - 10s - loss: 0.4605 - acc: 0.8390 - precision_m: 0.8859 - recall_m: 0.7873 - val_loss: 0.4572 - val_acc: 0.8483 - val_precision_m: 0.8845 - val_recall_m: 0.8121\n",
      "Epoch 236/5000\n",
      " - 10s - loss: 0.4623 - acc: 0.8364 - precision_m: 0.8872 - recall_m: 0.7822 - val_loss: 0.4525 - val_acc: 0.8654 - val_precision_m: 0.9261 - val_recall_m: 0.8006\n",
      "Epoch 237/5000\n",
      " - 10s - loss: 0.4596 - acc: 0.8389 - precision_m: 0.8925 - recall_m: 0.7794 - val_loss: 0.4624 - val_acc: 0.8300 - val_precision_m: 0.8134 - val_recall_m: 0.8603\n",
      "Epoch 238/5000\n",
      " - 10s - loss: 0.4561 - acc: 0.8432 - precision_m: 0.8925 - recall_m: 0.7868 - val_loss: 0.4637 - val_acc: 0.8212 - val_precision_m: 0.8005 - val_recall_m: 0.8703\n",
      "Epoch 239/5000\n",
      " - 10s - loss: 0.4565 - acc: 0.8426 - precision_m: 0.8948 - recall_m: 0.7850 - val_loss: 0.4479 - val_acc: 0.8627 - val_precision_m: 0.9479 - val_recall_m: 0.7742\n",
      "Epoch 240/5000\n",
      " - 10s - loss: 0.4536 - acc: 0.8461 - precision_m: 0.8966 - recall_m: 0.7907 - val_loss: 0.4638 - val_acc: 0.8267 - val_precision_m: 0.9765 - val_recall_m: 0.6735\n",
      "Epoch 241/5000\n",
      " - 10s - loss: 0.4503 - acc: 0.8511 - precision_m: 0.9049 - recall_m: 0.7913 - val_loss: 0.4474 - val_acc: 0.8583 - val_precision_m: 0.9366 - val_recall_m: 0.7690\n",
      "Epoch 242/5000\n",
      " - 10s - loss: 0.4489 - acc: 0.8493 - precision_m: 0.9025 - recall_m: 0.7905 - val_loss: 0.4503 - val_acc: 0.8433 - val_precision_m: 0.9329 - val_recall_m: 0.7424\n",
      "Epoch 243/5000\n",
      " - 10s - loss: 0.4466 - acc: 0.8536 - precision_m: 0.9040 - recall_m: 0.7937 - val_loss: 0.4429 - val_acc: 0.8677 - val_precision_m: 0.9253 - val_recall_m: 0.8063\n",
      "Epoch 244/5000\n",
      " - 10s - loss: 0.4465 - acc: 0.8461 - precision_m: 0.8953 - recall_m: 0.7912 - val_loss: 0.4443 - val_acc: 0.8583 - val_precision_m: 0.9367 - val_recall_m: 0.7688\n",
      "Epoch 245/5000\n",
      " - 10s - loss: 0.4453 - acc: 0.8494 - precision_m: 0.8990 - recall_m: 0.7942 - val_loss: 0.4494 - val_acc: 0.8394 - val_precision_m: 0.9721 - val_recall_m: 0.7009\n",
      "Epoch 246/5000\n",
      " - 10s - loss: 0.4471 - acc: 0.8429 - precision_m: 0.8951 - recall_m: 0.7877 - val_loss: 0.4462 - val_acc: 0.8450 - val_precision_m: 0.9536 - val_recall_m: 0.7266\n",
      "Epoch 247/5000\n",
      " - 10s - loss: 0.4443 - acc: 0.8490 - precision_m: 0.9029 - recall_m: 0.7879 - val_loss: 0.4413 - val_acc: 0.8533 - val_precision_m: 0.9457 - val_recall_m: 0.7497\n",
      "Epoch 248/5000\n",
      " - 10s - loss: 0.4454 - acc: 0.8454 - precision_m: 0.8973 - recall_m: 0.7903 - val_loss: 0.4370 - val_acc: 0.8688 - val_precision_m: 0.9464 - val_recall_m: 0.7875\n",
      "Epoch 249/5000\n",
      " - 10s - loss: 0.4436 - acc: 0.8427 - precision_m: 0.8977 - recall_m: 0.7859 - val_loss: 0.4378 - val_acc: 0.8643 - val_precision_m: 0.9020 - val_recall_m: 0.8257\n",
      "Epoch 250/5000\n",
      " - 10s - loss: 0.4382 - acc: 0.8559 - precision_m: 0.9078 - recall_m: 0.7965 - val_loss: 0.4388 - val_acc: 0.8544 - val_precision_m: 0.9657 - val_recall_m: 0.7347\n",
      "Epoch 251/5000\n",
      " - 10s - loss: 0.4373 - acc: 0.8554 - precision_m: 0.9080 - recall_m: 0.7941 - val_loss: 0.4333 - val_acc: 0.8688 - val_precision_m: 0.9332 - val_recall_m: 0.7996\n",
      "Epoch 252/5000\n",
      " - 10s - loss: 0.4373 - acc: 0.8515 - precision_m: 0.9037 - recall_m: 0.7950 - val_loss: 0.4482 - val_acc: 0.8372 - val_precision_m: 0.9566 - val_recall_m: 0.7105\n",
      "Epoch 253/5000\n",
      " - 10s - loss: 0.4371 - acc: 0.8464 - precision_m: 0.8992 - recall_m: 0.7900 - val_loss: 0.4521 - val_acc: 0.8239 - val_precision_m: 0.9819 - val_recall_m: 0.6648\n",
      "Epoch 254/5000\n",
      " - 10s - loss: 0.4405 - acc: 0.8415 - precision_m: 0.8939 - recall_m: 0.7893 - val_loss: 0.4538 - val_acc: 0.7890 - val_precision_m: 0.7466 - val_recall_m: 0.8937\n",
      "Epoch 255/5000\n",
      " - 10s - loss: 0.4361 - acc: 0.8483 - precision_m: 0.9017 - recall_m: 0.7898 - val_loss: 0.4290 - val_acc: 0.8649 - val_precision_m: 0.8915 - val_recall_m: 0.8390\n",
      "Epoch 256/5000\n",
      " - 10s - loss: 0.4311 - acc: 0.8517 - precision_m: 0.9014 - recall_m: 0.7956 - val_loss: 0.4739 - val_acc: 0.7946 - val_precision_m: 0.9915 - val_recall_m: 0.6017\n",
      "Epoch 257/5000\n",
      " - 10s - loss: 0.4344 - acc: 0.8512 - precision_m: 0.9024 - recall_m: 0.7949 - val_loss: 0.4234 - val_acc: 0.8704 - val_precision_m: 0.9559 - val_recall_m: 0.7825\n",
      "Epoch 258/5000\n",
      " - 10s - loss: 0.4287 - acc: 0.8554 - precision_m: 0.9098 - recall_m: 0.7921 - val_loss: 0.4241 - val_acc: 0.8654 - val_precision_m: 0.9062 - val_recall_m: 0.8223\n",
      "Epoch 259/5000\n",
      " - 10s - loss: 0.4279 - acc: 0.8587 - precision_m: 0.9102 - recall_m: 0.8014 - val_loss: 0.4304 - val_acc: 0.8594 - val_precision_m: 0.9706 - val_recall_m: 0.7474\n",
      "Epoch 260/5000\n",
      " - 10s - loss: 0.4291 - acc: 0.8501 - precision_m: 0.9024 - recall_m: 0.7922 - val_loss: 0.4337 - val_acc: 0.8466 - val_precision_m: 0.9781 - val_recall_m: 0.7101\n",
      "Epoch 261/5000\n",
      " - 10s - loss: 0.4278 - acc: 0.8530 - precision_m: 0.9054 - recall_m: 0.7936 - val_loss: 0.4281 - val_acc: 0.8549 - val_precision_m: 0.9672 - val_recall_m: 0.7352\n",
      "Epoch 262/5000\n",
      " - 10s - loss: 0.4248 - acc: 0.8604 - precision_m: 0.9123 - recall_m: 0.8014 - val_loss: 0.4186 - val_acc: 0.8749 - val_precision_m: 0.9367 - val_recall_m: 0.8092\n",
      "Epoch 263/5000\n",
      " - 10s - loss: 0.4249 - acc: 0.8568 - precision_m: 0.9074 - recall_m: 0.7977 - val_loss: 0.4323 - val_acc: 0.8494 - val_precision_m: 0.9794 - val_recall_m: 0.7148\n",
      "Epoch 264/5000\n",
      " - 10s - loss: 0.4211 - acc: 0.8593 - precision_m: 0.9099 - recall_m: 0.8029 - val_loss: 0.4274 - val_acc: 0.8583 - val_precision_m: 0.9492 - val_recall_m: 0.7569\n",
      "Epoch 265/5000\n",
      " - 10s - loss: 0.4233 - acc: 0.8538 - precision_m: 0.9026 - recall_m: 0.7980 - val_loss: 0.4195 - val_acc: 0.8671 - val_precision_m: 0.9490 - val_recall_m: 0.7811\n",
      "Epoch 266/5000\n",
      " - 10s - loss: 0.4185 - acc: 0.8613 - precision_m: 0.9124 - recall_m: 0.8018 - val_loss: 0.4129 - val_acc: 0.8760 - val_precision_m: 0.9436 - val_recall_m: 0.8035\n",
      "Epoch 267/5000\n",
      " - 10s - loss: 0.4202 - acc: 0.8590 - precision_m: 0.9115 - recall_m: 0.8005 - val_loss: 0.4232 - val_acc: 0.8488 - val_precision_m: 0.8581 - val_recall_m: 0.8565\n",
      "Epoch 268/5000\n",
      " - 10s - loss: 0.4180 - acc: 0.8585 - precision_m: 0.9081 - recall_m: 0.8003 - val_loss: 0.4123 - val_acc: 0.8743 - val_precision_m: 0.9452 - val_recall_m: 0.7995\n",
      "Epoch 269/5000\n",
      " - 10s - loss: 0.4186 - acc: 0.8611 - precision_m: 0.9125 - recall_m: 0.8023 - val_loss: 0.4119 - val_acc: 0.8749 - val_precision_m: 0.9245 - val_recall_m: 0.8222\n",
      "Epoch 270/5000\n",
      " - 10s - loss: 0.4146 - acc: 0.8623 - precision_m: 0.9162 - recall_m: 0.8010 - val_loss: 0.4097 - val_acc: 0.8704 - val_precision_m: 0.9214 - val_recall_m: 0.8166\n",
      "Epoch 271/5000\n",
      " - 10s - loss: 0.4170 - acc: 0.8534 - precision_m: 0.9050 - recall_m: 0.7980 - val_loss: 0.4334 - val_acc: 0.8372 - val_precision_m: 0.9727 - val_recall_m: 0.6970\n",
      "Epoch 272/5000\n",
      " - 10s - loss: 0.4149 - acc: 0.8579 - precision_m: 0.9077 - recall_m: 0.8026 - val_loss: 0.4268 - val_acc: 0.8405 - val_precision_m: 0.9705 - val_recall_m: 0.7038\n",
      "Epoch 273/5000\n",
      " - 10s - loss: 0.4145 - acc: 0.8573 - precision_m: 0.9086 - recall_m: 0.8034 - val_loss: 0.4196 - val_acc: 0.8605 - val_precision_m: 0.9700 - val_recall_m: 0.7436\n",
      "Epoch 274/5000\n",
      " - 10s - loss: 0.4099 - acc: 0.8640 - precision_m: 0.9141 - recall_m: 0.8065 - val_loss: 0.4169 - val_acc: 0.8522 - val_precision_m: 0.9574 - val_recall_m: 0.7380\n",
      "Epoch 275/5000\n",
      " - 10s - loss: 0.4086 - acc: 0.8631 - precision_m: 0.9149 - recall_m: 0.8061 - val_loss: 0.4119 - val_acc: 0.8610 - val_precision_m: 0.8890 - val_recall_m: 0.8338\n",
      "Epoch 276/5000\n",
      " - 10s - loss: 0.4087 - acc: 0.8611 - precision_m: 0.9123 - recall_m: 0.8051 - val_loss: 0.4178 - val_acc: 0.8499 - val_precision_m: 0.9665 - val_recall_m: 0.7263\n",
      "Epoch 277/5000\n",
      " - 10s - loss: 0.4086 - acc: 0.8608 - precision_m: 0.9128 - recall_m: 0.8030 - val_loss: 0.4048 - val_acc: 0.8754 - val_precision_m: 0.9581 - val_recall_m: 0.7895\n",
      "Epoch 278/5000\n",
      " - 10s - loss: 0.4038 - acc: 0.8669 - precision_m: 0.9185 - recall_m: 0.8067 - val_loss: 0.4081 - val_acc: 0.8583 - val_precision_m: 0.9506 - val_recall_m: 0.7560\n",
      "Epoch 279/5000\n",
      " - 10s - loss: 0.4118 - acc: 0.8524 - precision_m: 0.9021 - recall_m: 0.8000 - val_loss: 0.4036 - val_acc: 0.8726 - val_precision_m: 0.9603 - val_recall_m: 0.7823\n",
      "Epoch 280/5000\n",
      " - 10s - loss: 0.4068 - acc: 0.8615 - precision_m: 0.9145 - recall_m: 0.8010 - val_loss: 0.4198 - val_acc: 0.8344 - val_precision_m: 0.8174 - val_recall_m: 0.8877\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 281/5000\n",
      " - 10s - loss: 0.4033 - acc: 0.8643 - precision_m: 0.9175 - recall_m: 0.8061 - val_loss: 0.4063 - val_acc: 0.8632 - val_precision_m: 0.9039 - val_recall_m: 0.8202\n",
      "Epoch 282/5000\n",
      " - 10s - loss: 0.4035 - acc: 0.8624 - precision_m: 0.9138 - recall_m: 0.8075 - val_loss: 0.3968 - val_acc: 0.8749 - val_precision_m: 0.9177 - val_recall_m: 0.8297\n",
      "Epoch 283/5000\n",
      " - 10s - loss: 0.4041 - acc: 0.8612 - precision_m: 0.9109 - recall_m: 0.8070 - val_loss: 0.3958 - val_acc: 0.8754 - val_precision_m: 0.9499 - val_recall_m: 0.7891\n",
      "Epoch 284/5000\n",
      " - 10s - loss: 0.4025 - acc: 0.8600 - precision_m: 0.9140 - recall_m: 0.8018 - val_loss: 0.3957 - val_acc: 0.8743 - val_precision_m: 0.9193 - val_recall_m: 0.8268\n",
      "Epoch 285/5000\n",
      " - 10s - loss: 0.4048 - acc: 0.8583 - precision_m: 0.9090 - recall_m: 0.8052 - val_loss: 0.4090 - val_acc: 0.8511 - val_precision_m: 0.9744 - val_recall_m: 0.7215\n",
      "Epoch 286/5000\n",
      " - 10s - loss: 0.4000 - acc: 0.8635 - precision_m: 0.9169 - recall_m: 0.8057 - val_loss: 0.3982 - val_acc: 0.8693 - val_precision_m: 0.9580 - val_recall_m: 0.7787\n",
      "Epoch 287/5000\n",
      " - 10s - loss: 0.3977 - acc: 0.8635 - precision_m: 0.9176 - recall_m: 0.8044 - val_loss: 0.3932 - val_acc: 0.8710 - val_precision_m: 0.9127 - val_recall_m: 0.8277\n",
      "Epoch 288/5000\n",
      " - 10s - loss: 0.3961 - acc: 0.8663 - precision_m: 0.9186 - recall_m: 0.8076 - val_loss: 0.3921 - val_acc: 0.8754 - val_precision_m: 0.9250 - val_recall_m: 0.8225\n",
      "Epoch 289/5000\n",
      " - 10s - loss: 0.3955 - acc: 0.8674 - precision_m: 0.9177 - recall_m: 0.8110 - val_loss: 0.3915 - val_acc: 0.8715 - val_precision_m: 0.9619 - val_recall_m: 0.7791\n",
      "Epoch 290/5000\n",
      " - 10s - loss: 0.3932 - acc: 0.8672 - precision_m: 0.9178 - recall_m: 0.8072 - val_loss: 0.3974 - val_acc: 0.8605 - val_precision_m: 0.9622 - val_recall_m: 0.7495\n",
      "Epoch 291/5000\n",
      " - 10s - loss: 0.3980 - acc: 0.8597 - precision_m: 0.9115 - recall_m: 0.8066 - val_loss: 0.4222 - val_acc: 0.8300 - val_precision_m: 0.9914 - val_recall_m: 0.6695\n",
      "Epoch 292/5000\n",
      " - 10s - loss: 0.3914 - acc: 0.8726 - precision_m: 0.9260 - recall_m: 0.8129 - val_loss: 0.4065 - val_acc: 0.8455 - val_precision_m: 0.8340 - val_recall_m: 0.8850\n",
      "Epoch 293/5000\n",
      " - 10s - loss: 0.3922 - acc: 0.8689 - precision_m: 0.9191 - recall_m: 0.8129 - val_loss: 0.3899 - val_acc: 0.8771 - val_precision_m: 0.9606 - val_recall_m: 0.7908\n",
      "Epoch 294/5000\n",
      " - 10s - loss: 0.3906 - acc: 0.8690 - precision_m: 0.9207 - recall_m: 0.8121 - val_loss: 0.3973 - val_acc: 0.8660 - val_precision_m: 0.8662 - val_recall_m: 0.8758\n",
      "Epoch 295/5000\n",
      " - 10s - loss: 0.3971 - acc: 0.8581 - precision_m: 0.9089 - recall_m: 0.8059 - val_loss: 0.3868 - val_acc: 0.8848 - val_precision_m: 0.9657 - val_recall_m: 0.7938\n",
      "Epoch 296/5000\n",
      " - 10s - loss: 0.3903 - acc: 0.8675 - precision_m: 0.9223 - recall_m: 0.8076 - val_loss: 0.3900 - val_acc: 0.8699 - val_precision_m: 0.9693 - val_recall_m: 0.7702\n",
      "Epoch 297/5000\n",
      " - 10s - loss: 0.3870 - acc: 0.8682 - precision_m: 0.9180 - recall_m: 0.8127 - val_loss: 0.3916 - val_acc: 0.8654 - val_precision_m: 0.9754 - val_recall_m: 0.7483\n",
      "Epoch 298/5000\n",
      " - 10s - loss: 0.3886 - acc: 0.8671 - precision_m: 0.9185 - recall_m: 0.8114 - val_loss: 0.3798 - val_acc: 0.8815 - val_precision_m: 0.9398 - val_recall_m: 0.8195\n",
      "Epoch 299/5000\n",
      " - 10s - loss: 0.3839 - acc: 0.8731 - precision_m: 0.9263 - recall_m: 0.8131 - val_loss: 0.3818 - val_acc: 0.8726 - val_precision_m: 0.9130 - val_recall_m: 0.8312\n",
      "Epoch 300/5000\n",
      " - 10s - loss: 0.3832 - acc: 0.8713 - precision_m: 0.9206 - recall_m: 0.8164 - val_loss: 0.3783 - val_acc: 0.8859 - val_precision_m: 0.9410 - val_recall_m: 0.8264\n",
      "Epoch 301/5000\n",
      " - 10s - loss: 0.3822 - acc: 0.8733 - precision_m: 0.9253 - recall_m: 0.8149 - val_loss: 0.4048 - val_acc: 0.8444 - val_precision_m: 0.9847 - val_recall_m: 0.7015\n",
      "Epoch 302/5000\n",
      " - 10s - loss: 0.3857 - acc: 0.8666 - precision_m: 0.9167 - recall_m: 0.8109 - val_loss: 0.3866 - val_acc: 0.8704 - val_precision_m: 0.9507 - val_recall_m: 0.7937\n",
      "Epoch 303/5000\n",
      " - 10s - loss: 0.3901 - acc: 0.8607 - precision_m: 0.9125 - recall_m: 0.8080 - val_loss: 0.3869 - val_acc: 0.8688 - val_precision_m: 0.8717 - val_recall_m: 0.8725\n",
      "Epoch 304/5000\n",
      " - 10s - loss: 0.3786 - acc: 0.8744 - precision_m: 0.9245 - recall_m: 0.8188 - val_loss: 0.3967 - val_acc: 0.8527 - val_precision_m: 0.9819 - val_recall_m: 0.7187\n",
      "Epoch 305/5000\n",
      " - 10s - loss: 0.3820 - acc: 0.8693 - precision_m: 0.9208 - recall_m: 0.8135 - val_loss: 0.3930 - val_acc: 0.8555 - val_precision_m: 0.8460 - val_recall_m: 0.8874\n",
      "Epoch 306/5000\n",
      " - 10s - loss: 0.3794 - acc: 0.8697 - precision_m: 0.9194 - recall_m: 0.8153 - val_loss: 0.4020 - val_acc: 0.8439 - val_precision_m: 0.9872 - val_recall_m: 0.6982\n",
      "Epoch 307/5000\n",
      " - 10s - loss: 0.3819 - acc: 0.8669 - precision_m: 0.9202 - recall_m: 0.8100 - val_loss: 0.3743 - val_acc: 0.8882 - val_precision_m: 0.9275 - val_recall_m: 0.8458\n",
      "Epoch 308/5000\n",
      " - 10s - loss: 0.3802 - acc: 0.8707 - precision_m: 0.9210 - recall_m: 0.8166 - val_loss: 0.3730 - val_acc: 0.8793 - val_precision_m: 0.9392 - val_recall_m: 0.8151\n",
      "Epoch 309/5000\n",
      " - 10s - loss: 0.3785 - acc: 0.8686 - precision_m: 0.9207 - recall_m: 0.8118 - val_loss: 0.3749 - val_acc: 0.8793 - val_precision_m: 0.9477 - val_recall_m: 0.8077\n",
      "Epoch 310/5000\n",
      " - 10s - loss: 0.3735 - acc: 0.8751 - precision_m: 0.9257 - recall_m: 0.8168 - val_loss: 0.3883 - val_acc: 0.8588 - val_precision_m: 0.9817 - val_recall_m: 0.7311\n",
      "Epoch 311/5000\n",
      " - 10s - loss: 0.3741 - acc: 0.8737 - precision_m: 0.9252 - recall_m: 0.8159 - val_loss: 0.3816 - val_acc: 0.8699 - val_precision_m: 0.8757 - val_recall_m: 0.8775\n",
      "Epoch 312/5000\n",
      " - 10s - loss: 0.3732 - acc: 0.8732 - precision_m: 0.9233 - recall_m: 0.8176 - val_loss: 0.3712 - val_acc: 0.8821 - val_precision_m: 0.9619 - val_recall_m: 0.7918\n",
      "Epoch 313/5000\n",
      " - 10s - loss: 0.3769 - acc: 0.8691 - precision_m: 0.9216 - recall_m: 0.8125 - val_loss: 0.3690 - val_acc: 0.8821 - val_precision_m: 0.9292 - val_recall_m: 0.8318\n",
      "Epoch 314/5000\n",
      " - 10s - loss: 0.3737 - acc: 0.8736 - precision_m: 0.9233 - recall_m: 0.8181 - val_loss: 0.3765 - val_acc: 0.8749 - val_precision_m: 0.9627 - val_recall_m: 0.7772\n",
      "Epoch 315/5000\n",
      " - 10s - loss: 0.3708 - acc: 0.8800 - precision_m: 0.9314 - recall_m: 0.8233 - val_loss: 0.3760 - val_acc: 0.8743 - val_precision_m: 0.8821 - val_recall_m: 0.8703\n",
      "Epoch 316/5000\n",
      " - 10s - loss: 0.3705 - acc: 0.8721 - precision_m: 0.9190 - recall_m: 0.8197 - val_loss: 0.3772 - val_acc: 0.8693 - val_precision_m: 0.9752 - val_recall_m: 0.7636\n",
      "Epoch 317/5000\n",
      " - 10s - loss: 0.3701 - acc: 0.8749 - precision_m: 0.9242 - recall_m: 0.8190 - val_loss: 0.3820 - val_acc: 0.8610 - val_precision_m: 0.9815 - val_recall_m: 0.7350\n",
      "Epoch 318/5000\n",
      " - 10s - loss: 0.3693 - acc: 0.8740 - precision_m: 0.9274 - recall_m: 0.8163 - val_loss: 0.3664 - val_acc: 0.8821 - val_precision_m: 0.9185 - val_recall_m: 0.8430\n",
      "Epoch 319/5000\n",
      " - 10s - loss: 0.3720 - acc: 0.8717 - precision_m: 0.9220 - recall_m: 0.8183 - val_loss: 0.3894 - val_acc: 0.8439 - val_precision_m: 0.8181 - val_recall_m: 0.8953\n",
      "Epoch 320/5000\n",
      " - 10s - loss: 0.3746 - acc: 0.8682 - precision_m: 0.9179 - recall_m: 0.8155 - val_loss: 0.3654 - val_acc: 0.8865 - val_precision_m: 0.9520 - val_recall_m: 0.8175\n",
      "Epoch 321/5000\n",
      " - 10s - loss: 0.3649 - acc: 0.8775 - precision_m: 0.9250 - recall_m: 0.8226 - val_loss: 0.3617 - val_acc: 0.8882 - val_precision_m: 0.9574 - val_recall_m: 0.8155\n",
      "Epoch 322/5000\n",
      " - 10s - loss: 0.3663 - acc: 0.8751 - precision_m: 0.9256 - recall_m: 0.8191 - val_loss: 0.3637 - val_acc: 0.8848 - val_precision_m: 0.9089 - val_recall_m: 0.8669\n",
      "Epoch 323/5000\n",
      " - 10s - loss: 0.3666 - acc: 0.8733 - precision_m: 0.9234 - recall_m: 0.8193 - val_loss: 0.3777 - val_acc: 0.8621 - val_precision_m: 0.9816 - val_recall_m: 0.7445\n",
      "Epoch 324/5000\n",
      " - 10s - loss: 0.3635 - acc: 0.8754 - precision_m: 0.9246 - recall_m: 0.8214 - val_loss: 0.3757 - val_acc: 0.8616 - val_precision_m: 0.9736 - val_recall_m: 0.7428\n",
      "Epoch 325/5000\n",
      " - 10s - loss: 0.3603 - acc: 0.8786 - precision_m: 0.9307 - recall_m: 0.8199 - val_loss: 0.3601 - val_acc: 0.8843 - val_precision_m: 0.9197 - val_recall_m: 0.8536\n",
      "Epoch 326/5000\n",
      " - 10s - loss: 0.3686 - acc: 0.8698 - precision_m: 0.9234 - recall_m: 0.8142 - val_loss: 0.3705 - val_acc: 0.8754 - val_precision_m: 0.8807 - val_recall_m: 0.8825\n",
      "Epoch 327/5000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " - 10s - loss: 0.3608 - acc: 0.8781 - precision_m: 0.9260 - recall_m: 0.8251 - val_loss: 0.3689 - val_acc: 0.8688 - val_precision_m: 0.9728 - val_recall_m: 0.7636\n",
      "Epoch 328/5000\n",
      " - 10s - loss: 0.3585 - acc: 0.8834 - precision_m: 0.9326 - recall_m: 0.8275 - val_loss: 0.3588 - val_acc: 0.8870 - val_precision_m: 0.9480 - val_recall_m: 0.8283\n",
      "Epoch 329/5000\n",
      " - 10s - loss: 0.3579 - acc: 0.8804 - precision_m: 0.9296 - recall_m: 0.8245 - val_loss: 0.3614 - val_acc: 0.8826 - val_precision_m: 0.9612 - val_recall_m: 0.8005\n",
      "Epoch 330/5000\n",
      " - 10s - loss: 0.3578 - acc: 0.8781 - precision_m: 0.9285 - recall_m: 0.8220 - val_loss: 0.3691 - val_acc: 0.8649 - val_precision_m: 0.9631 - val_recall_m: 0.7580\n",
      "Epoch 331/5000\n",
      " - 10s - loss: 0.3612 - acc: 0.8726 - precision_m: 0.9225 - recall_m: 0.8189 - val_loss: 0.3565 - val_acc: 0.8854 - val_precision_m: 0.9148 - val_recall_m: 0.8610\n",
      "Epoch 332/5000\n",
      " - 10s - loss: 0.3569 - acc: 0.8785 - precision_m: 0.9297 - recall_m: 0.8225 - val_loss: 0.3537 - val_acc: 0.8843 - val_precision_m: 0.9287 - val_recall_m: 0.8359\n",
      "Epoch 333/5000\n",
      " - 10s - loss: 0.3663 - acc: 0.8677 - precision_m: 0.9170 - recall_m: 0.8180 - val_loss: 0.4111 - val_acc: 0.8317 - val_precision_m: 0.9869 - val_recall_m: 0.6761\n",
      "Epoch 334/5000\n",
      " - 10s - loss: 0.3570 - acc: 0.8778 - precision_m: 0.9256 - recall_m: 0.8241 - val_loss: 0.3607 - val_acc: 0.8815 - val_precision_m: 0.8994 - val_recall_m: 0.8635\n",
      "Epoch 335/5000\n",
      " - 10s - loss: 0.3530 - acc: 0.8845 - precision_m: 0.9336 - recall_m: 0.8295 - val_loss: 0.3574 - val_acc: 0.8810 - val_precision_m: 0.9701 - val_recall_m: 0.7900\n",
      "Epoch 336/5000\n",
      " - 10s - loss: 0.3582 - acc: 0.8728 - precision_m: 0.9217 - recall_m: 0.8195 - val_loss: 0.3846 - val_acc: 0.8378 - val_precision_m: 0.8029 - val_recall_m: 0.9081\n",
      "Epoch 337/5000\n",
      " - 10s - loss: 0.3556 - acc: 0.8810 - precision_m: 0.9286 - recall_m: 0.8288 - val_loss: 0.3565 - val_acc: 0.8771 - val_precision_m: 0.9259 - val_recall_m: 0.8246\n",
      "Epoch 338/5000\n",
      " - 10s - loss: 0.3530 - acc: 0.8780 - precision_m: 0.9273 - recall_m: 0.8244 - val_loss: 0.3477 - val_acc: 0.8937 - val_precision_m: 0.9679 - val_recall_m: 0.8164\n",
      "Epoch 339/5000\n",
      " - 10s - loss: 0.3511 - acc: 0.8805 - precision_m: 0.9312 - recall_m: 0.8244 - val_loss: 0.3522 - val_acc: 0.8870 - val_precision_m: 0.9209 - val_recall_m: 0.8501\n",
      "Epoch 340/5000\n",
      " - 10s - loss: 0.3536 - acc: 0.8778 - precision_m: 0.9296 - recall_m: 0.8230 - val_loss: 0.3487 - val_acc: 0.8876 - val_precision_m: 0.9428 - val_recall_m: 0.8348\n",
      "Epoch 341/5000\n",
      " - 10s - loss: 0.3488 - acc: 0.8848 - precision_m: 0.9378 - recall_m: 0.8255 - val_loss: 0.3530 - val_acc: 0.8810 - val_precision_m: 0.9676 - val_recall_m: 0.7915\n",
      "Epoch 342/5000\n",
      " - 10s - loss: 0.3492 - acc: 0.8801 - precision_m: 0.9316 - recall_m: 0.8226 - val_loss: 0.3911 - val_acc: 0.8433 - val_precision_m: 0.9900 - val_recall_m: 0.6950\n",
      "Epoch 343/5000\n",
      " - 10s - loss: 0.3477 - acc: 0.8827 - precision_m: 0.9314 - recall_m: 0.8269 - val_loss: 0.3539 - val_acc: 0.8854 - val_precision_m: 0.9256 - val_recall_m: 0.8327\n",
      "Epoch 344/5000\n",
      " - 10s - loss: 0.3488 - acc: 0.8797 - precision_m: 0.9287 - recall_m: 0.8247 - val_loss: 0.3463 - val_acc: 0.8909 - val_precision_m: 0.9522 - val_recall_m: 0.8255\n",
      "Epoch 345/5000\n",
      " - 10s - loss: 0.3549 - acc: 0.8743 - precision_m: 0.9261 - recall_m: 0.8193 - val_loss: 0.4315 - val_acc: 0.8079 - val_precision_m: 0.9938 - val_recall_m: 0.6260\n",
      "Epoch 346/5000\n",
      " - 10s - loss: 0.3517 - acc: 0.8776 - precision_m: 0.9296 - recall_m: 0.8224 - val_loss: 0.3531 - val_acc: 0.8760 - val_precision_m: 0.9011 - val_recall_m: 0.8513\n",
      "Epoch 347/5000\n",
      " - 10s - loss: 0.3448 - acc: 0.8839 - precision_m: 0.9337 - recall_m: 0.8290 - val_loss: 0.3424 - val_acc: 0.8893 - val_precision_m: 0.9517 - val_recall_m: 0.8301\n",
      "Epoch 348/5000\n",
      " - 10s - loss: 0.3465 - acc: 0.8804 - precision_m: 0.9300 - recall_m: 0.8271 - val_loss: 0.3468 - val_acc: 0.8848 - val_precision_m: 0.9262 - val_recall_m: 0.8401\n",
      "Epoch 349/5000\n",
      " - 10s - loss: 0.3430 - acc: 0.8842 - precision_m: 0.9337 - recall_m: 0.8289 - val_loss: 0.3431 - val_acc: 0.8898 - val_precision_m: 0.9300 - val_recall_m: 0.8459\n",
      "Epoch 350/5000\n",
      " - 10s - loss: 0.3445 - acc: 0.8828 - precision_m: 0.9334 - recall_m: 0.8278 - val_loss: 0.3548 - val_acc: 0.8726 - val_precision_m: 0.9806 - val_recall_m: 0.7652\n",
      "Epoch 351/5000\n",
      " - 10s - loss: 0.3444 - acc: 0.8821 - precision_m: 0.9329 - recall_m: 0.8266 - val_loss: 0.3425 - val_acc: 0.8904 - val_precision_m: 0.9630 - val_recall_m: 0.8143\n",
      "Epoch 352/5000\n",
      " - 10s - loss: 0.3460 - acc: 0.8777 - precision_m: 0.9283 - recall_m: 0.8246 - val_loss: 0.3515 - val_acc: 0.8765 - val_precision_m: 0.9759 - val_recall_m: 0.7690\n",
      "Epoch 353/5000\n",
      " - 10s - loss: 0.3406 - acc: 0.8832 - precision_m: 0.9347 - recall_m: 0.8267 - val_loss: 0.3477 - val_acc: 0.8804 - val_precision_m: 0.9068 - val_recall_m: 0.8532\n",
      "Epoch 354/5000\n",
      " - 10s - loss: 0.3442 - acc: 0.8808 - precision_m: 0.9315 - recall_m: 0.8269 - val_loss: 0.3454 - val_acc: 0.8859 - val_precision_m: 0.9229 - val_recall_m: 0.8461\n",
      "Epoch 355/5000\n",
      " - 10s - loss: 0.3450 - acc: 0.8787 - precision_m: 0.9276 - recall_m: 0.8285 - val_loss: 0.3437 - val_acc: 0.8765 - val_precision_m: 0.9687 - val_recall_m: 0.7827\n",
      "Epoch 356/5000\n",
      " - 10s - loss: 0.3388 - acc: 0.8842 - precision_m: 0.9333 - recall_m: 0.8300 - val_loss: 0.3433 - val_acc: 0.8837 - val_precision_m: 0.9703 - val_recall_m: 0.7946\n",
      "Epoch 357/5000\n",
      " - 10s - loss: 0.3360 - acc: 0.8840 - precision_m: 0.9328 - recall_m: 0.8301 - val_loss: 0.3751 - val_acc: 0.8522 - val_precision_m: 0.9823 - val_recall_m: 0.7181\n",
      "Epoch 358/5000\n",
      " - 10s - loss: 0.3404 - acc: 0.8803 - precision_m: 0.9302 - recall_m: 0.8267 - val_loss: 0.3462 - val_acc: 0.8815 - val_precision_m: 0.8961 - val_recall_m: 0.8761\n",
      "Epoch 359/5000\n",
      " - 10s - loss: 0.3381 - acc: 0.8841 - precision_m: 0.9337 - recall_m: 0.8302 - val_loss: 0.3595 - val_acc: 0.8649 - val_precision_m: 0.9827 - val_recall_m: 0.7413\n",
      "Epoch 360/5000\n",
      " - 10s - loss: 0.3430 - acc: 0.8783 - precision_m: 0.9307 - recall_m: 0.8238 - val_loss: 0.3411 - val_acc: 0.8870 - val_precision_m: 0.9683 - val_recall_m: 0.7953\n",
      "Epoch 361/5000\n",
      " - 10s - loss: 0.3337 - acc: 0.8885 - precision_m: 0.9372 - recall_m: 0.8340 - val_loss: 0.3337 - val_acc: 0.8937 - val_precision_m: 0.9348 - val_recall_m: 0.8564\n",
      "Epoch 362/5000\n",
      " - 10s - loss: 0.3323 - acc: 0.8919 - precision_m: 0.9416 - recall_m: 0.8367 - val_loss: 0.3353 - val_acc: 0.8882 - val_precision_m: 0.9177 - val_recall_m: 0.8574\n",
      "Epoch 363/5000\n",
      " - 10s - loss: 0.3297 - acc: 0.8911 - precision_m: 0.9410 - recall_m: 0.8364 - val_loss: 0.3325 - val_acc: 0.8948 - val_precision_m: 0.9477 - val_recall_m: 0.8370\n",
      "Epoch 364/5000\n",
      " - 10s - loss: 0.3329 - acc: 0.8871 - precision_m: 0.9356 - recall_m: 0.8330 - val_loss: 0.3314 - val_acc: 0.8909 - val_precision_m: 0.9302 - val_recall_m: 0.8554\n",
      "Epoch 365/5000\n",
      " - 10s - loss: 0.3339 - acc: 0.8850 - precision_m: 0.9346 - recall_m: 0.8320 - val_loss: 0.3304 - val_acc: 0.8937 - val_precision_m: 0.9501 - val_recall_m: 0.8326\n",
      "Epoch 366/5000\n",
      " - 10s - loss: 0.3411 - acc: 0.8770 - precision_m: 0.9271 - recall_m: 0.8256 - val_loss: 0.3322 - val_acc: 0.8904 - val_precision_m: 0.9527 - val_recall_m: 0.8307\n",
      "Epoch 367/5000\n",
      " - 10s - loss: 0.3286 - acc: 0.8904 - precision_m: 0.9412 - recall_m: 0.8348 - val_loss: 0.3296 - val_acc: 0.8893 - val_precision_m: 0.9353 - val_recall_m: 0.8388\n",
      "Epoch 368/5000\n",
      " - 10s - loss: 0.3295 - acc: 0.8874 - precision_m: 0.9393 - recall_m: 0.8296 - val_loss: 0.3279 - val_acc: 0.8987 - val_precision_m: 0.9483 - val_recall_m: 0.8526\n",
      "Epoch 369/5000\n",
      " - 10s - loss: 0.3275 - acc: 0.8896 - precision_m: 0.9417 - recall_m: 0.8329 - val_loss: 0.3351 - val_acc: 0.8826 - val_precision_m: 0.9577 - val_recall_m: 0.8035\n",
      "Epoch 370/5000\n",
      " - 10s - loss: 0.3291 - acc: 0.8898 - precision_m: 0.9406 - recall_m: 0.8337 - val_loss: 0.3263 - val_acc: 0.8942 - val_precision_m: 0.9629 - val_recall_m: 0.8216\n",
      "Epoch 371/5000\n",
      " - 10s - loss: 0.3268 - acc: 0.8891 - precision_m: 0.9399 - recall_m: 0.8327 - val_loss: 0.3414 - val_acc: 0.8843 - val_precision_m: 0.8922 - val_recall_m: 0.8796\n",
      "Epoch 372/5000\n",
      " - 10s - loss: 0.3282 - acc: 0.8867 - precision_m: 0.9347 - recall_m: 0.8343 - val_loss: 0.3280 - val_acc: 0.8937 - val_precision_m: 0.9337 - val_recall_m: 0.8507\n",
      "Epoch 373/5000\n",
      " - 10s - loss: 0.3281 - acc: 0.8894 - precision_m: 0.9384 - recall_m: 0.8360 - val_loss: 0.3431 - val_acc: 0.8826 - val_precision_m: 0.8831 - val_recall_m: 0.8949\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 374/5000\n",
      " - 10s - loss: 0.3284 - acc: 0.8883 - precision_m: 0.9370 - recall_m: 0.8348 - val_loss: 0.3297 - val_acc: 0.8870 - val_precision_m: 0.9673 - val_recall_m: 0.8042\n",
      "Epoch 375/5000\n",
      " - 10s - loss: 0.3283 - acc: 0.8852 - precision_m: 0.9347 - recall_m: 0.8312 - val_loss: 0.3239 - val_acc: 0.9009 - val_precision_m: 0.9630 - val_recall_m: 0.8351\n",
      "Epoch 376/5000\n",
      " - 10s - loss: 0.3258 - acc: 0.8882 - precision_m: 0.9368 - recall_m: 0.8353 - val_loss: 0.3305 - val_acc: 0.8848 - val_precision_m: 0.9681 - val_recall_m: 0.8066\n",
      "Epoch 377/5000\n",
      " - 10s - loss: 0.3287 - acc: 0.8843 - precision_m: 0.9326 - recall_m: 0.8328 - val_loss: 0.3237 - val_acc: 0.8915 - val_precision_m: 0.9224 - val_recall_m: 0.8641\n",
      "Epoch 378/5000\n",
      " - 10s - loss: 0.3237 - acc: 0.8899 - precision_m: 0.9367 - recall_m: 0.8376 - val_loss: 0.3250 - val_acc: 0.8904 - val_precision_m: 0.9300 - val_recall_m: 0.8474\n",
      "Epoch 379/5000\n",
      " - 10s - loss: 0.3267 - acc: 0.8882 - precision_m: 0.9347 - recall_m: 0.8373 - val_loss: 0.3395 - val_acc: 0.8815 - val_precision_m: 0.8825 - val_recall_m: 0.8940\n",
      "Epoch 380/5000\n",
      " - 10s - loss: 0.3243 - acc: 0.8876 - precision_m: 0.9357 - recall_m: 0.8349 - val_loss: 0.3226 - val_acc: 0.8915 - val_precision_m: 0.9597 - val_recall_m: 0.8195\n",
      "Epoch 381/5000\n",
      " - 10s - loss: 0.3254 - acc: 0.8893 - precision_m: 0.9370 - recall_m: 0.8389 - val_loss: 0.3371 - val_acc: 0.8776 - val_precision_m: 0.8801 - val_recall_m: 0.8881\n",
      "Epoch 382/5000\n",
      " - 10s - loss: 0.3240 - acc: 0.8839 - precision_m: 0.9312 - recall_m: 0.8328 - val_loss: 0.3202 - val_acc: 0.8926 - val_precision_m: 0.9525 - val_recall_m: 0.8283\n",
      "Epoch 383/5000\n",
      " - 10s - loss: 0.3223 - acc: 0.8911 - precision_m: 0.9397 - recall_m: 0.8378 - val_loss: 0.3426 - val_acc: 0.8715 - val_precision_m: 0.9742 - val_recall_m: 0.7613\n",
      "Epoch 384/5000\n",
      " - 10s - loss: 0.3183 - acc: 0.8915 - precision_m: 0.9394 - recall_m: 0.8383 - val_loss: 0.3200 - val_acc: 0.8931 - val_precision_m: 0.9491 - val_recall_m: 0.8401\n",
      "Epoch 385/5000\n",
      " - 10s - loss: 0.3179 - acc: 0.8907 - precision_m: 0.9392 - recall_m: 0.8370 - val_loss: 0.3275 - val_acc: 0.8898 - val_precision_m: 0.9105 - val_recall_m: 0.8677\n",
      "Epoch 386/5000\n",
      " - 10s - loss: 0.3225 - acc: 0.8899 - precision_m: 0.9368 - recall_m: 0.8390 - val_loss: 0.3379 - val_acc: 0.8848 - val_precision_m: 0.8847 - val_recall_m: 0.8981\n",
      "Epoch 387/5000\n",
      " - 10s - loss: 0.3268 - acc: 0.8844 - precision_m: 0.9306 - recall_m: 0.8357 - val_loss: 0.3181 - val_acc: 0.8965 - val_precision_m: 0.9592 - val_recall_m: 0.8371\n",
      "Epoch 388/5000\n",
      " - 10s - loss: 0.3202 - acc: 0.8861 - precision_m: 0.9346 - recall_m: 0.8346 - val_loss: 0.3170 - val_acc: 0.8981 - val_precision_m: 0.9682 - val_recall_m: 0.8244\n",
      "Epoch 389/5000\n",
      " - 10s - loss: 0.3189 - acc: 0.8912 - precision_m: 0.9401 - recall_m: 0.8382 - val_loss: 0.3415 - val_acc: 0.8754 - val_precision_m: 0.8674 - val_recall_m: 0.9009\n",
      "Epoch 390/5000\n",
      " - 10s - loss: 0.3172 - acc: 0.8888 - precision_m: 0.9368 - recall_m: 0.8366 - val_loss: 0.3277 - val_acc: 0.8826 - val_precision_m: 0.9796 - val_recall_m: 0.7917\n",
      "Epoch 391/5000\n",
      " - 10s - loss: 0.3128 - acc: 0.8946 - precision_m: 0.9422 - recall_m: 0.8425 - val_loss: 0.3192 - val_acc: 0.8992 - val_precision_m: 0.9562 - val_recall_m: 0.8446\n",
      "Epoch 392/5000\n",
      " - 10s - loss: 0.3184 - acc: 0.8896 - precision_m: 0.9362 - recall_m: 0.8393 - val_loss: 0.3173 - val_acc: 0.8981 - val_precision_m: 0.9339 - val_recall_m: 0.8584\n",
      "Epoch 393/5000\n",
      " - 10s - loss: 0.3150 - acc: 0.8913 - precision_m: 0.9383 - recall_m: 0.8400 - val_loss: 0.3140 - val_acc: 0.9020 - val_precision_m: 0.9345 - val_recall_m: 0.8648\n",
      "Epoch 394/5000\n",
      " - 10s - loss: 0.3099 - acc: 0.8966 - precision_m: 0.9453 - recall_m: 0.8429 - val_loss: 0.3115 - val_acc: 0.8981 - val_precision_m: 0.9485 - val_recall_m: 0.8500\n",
      "Epoch 395/5000\n",
      " - 10s - loss: 0.3096 - acc: 0.8955 - precision_m: 0.9443 - recall_m: 0.8418 - val_loss: 0.3208 - val_acc: 0.8915 - val_precision_m: 0.9683 - val_recall_m: 0.8114\n",
      "Epoch 396/5000\n",
      " - 10s - loss: 0.3105 - acc: 0.8942 - precision_m: 0.9420 - recall_m: 0.8412 - val_loss: 0.3091 - val_acc: 0.9037 - val_precision_m: 0.9441 - val_recall_m: 0.8651\n",
      "Epoch 397/5000\n",
      " - 10s - loss: 0.3169 - acc: 0.8893 - precision_m: 0.9367 - recall_m: 0.8384 - val_loss: 0.3159 - val_acc: 0.8976 - val_precision_m: 0.9291 - val_recall_m: 0.8695\n",
      "Epoch 398/5000\n",
      " - 10s - loss: 0.3160 - acc: 0.8907 - precision_m: 0.9381 - recall_m: 0.8397 - val_loss: 0.3130 - val_acc: 0.8976 - val_precision_m: 0.9555 - val_recall_m: 0.8418\n",
      "Epoch 399/5000\n",
      " - 10s - loss: 0.3131 - acc: 0.8946 - precision_m: 0.9413 - recall_m: 0.8446 - val_loss: 0.3325 - val_acc: 0.8749 - val_precision_m: 0.9620 - val_recall_m: 0.7916\n",
      "Epoch 400/5000\n",
      " - 10s - loss: 0.3123 - acc: 0.8915 - precision_m: 0.9391 - recall_m: 0.8396 - val_loss: 0.3289 - val_acc: 0.8765 - val_precision_m: 0.9797 - val_recall_m: 0.7736\n",
      "Epoch 401/5000\n",
      " - 10s - loss: 0.3135 - acc: 0.8926 - precision_m: 0.9389 - recall_m: 0.8427 - val_loss: 0.3197 - val_acc: 0.8931 - val_precision_m: 0.9037 - val_recall_m: 0.8899\n",
      "Epoch 402/5000\n",
      " - 10s - loss: 0.3095 - acc: 0.8948 - precision_m: 0.9426 - recall_m: 0.8433 - val_loss: 0.3134 - val_acc: 0.8909 - val_precision_m: 0.9674 - val_recall_m: 0.8116\n",
      "Epoch 403/5000\n",
      " - 10s - loss: 0.3062 - acc: 0.8973 - precision_m: 0.9433 - recall_m: 0.8459 - val_loss: 0.3113 - val_acc: 0.8970 - val_precision_m: 0.9636 - val_recall_m: 0.8265\n",
      "Epoch 404/5000\n",
      " - 10s - loss: 0.3068 - acc: 0.8965 - precision_m: 0.9459 - recall_m: 0.8432 - val_loss: 0.3234 - val_acc: 0.8798 - val_precision_m: 0.9796 - val_recall_m: 0.7793\n",
      "Epoch 405/5000\n",
      " - 10s - loss: 0.3136 - acc: 0.8896 - precision_m: 0.9391 - recall_m: 0.8387 - val_loss: 0.3099 - val_acc: 0.9003 - val_precision_m: 0.9378 - val_recall_m: 0.8661\n",
      "Epoch 406/5000\n",
      " - 10s - loss: 0.3075 - acc: 0.8947 - precision_m: 0.9425 - recall_m: 0.8424 - val_loss: 0.3100 - val_acc: 0.8953 - val_precision_m: 0.9164 - val_recall_m: 0.8797\n",
      "Epoch 407/5000\n",
      " - 10s - loss: 0.3083 - acc: 0.8939 - precision_m: 0.9417 - recall_m: 0.8431 - val_loss: 0.3189 - val_acc: 0.8893 - val_precision_m: 0.9718 - val_recall_m: 0.8119\n",
      "Epoch 408/5000\n",
      " - 10s - loss: 0.3043 - acc: 0.8967 - precision_m: 0.9424 - recall_m: 0.8468 - val_loss: 0.3066 - val_acc: 0.8998 - val_precision_m: 0.9587 - val_recall_m: 0.8431\n",
      "Epoch 409/5000\n",
      " - 10s - loss: 0.3064 - acc: 0.8944 - precision_m: 0.9410 - recall_m: 0.8448 - val_loss: 0.3210 - val_acc: 0.8848 - val_precision_m: 0.9764 - val_recall_m: 0.7920\n",
      "Epoch 410/5000\n",
      " - 10s - loss: 0.3032 - acc: 0.8967 - precision_m: 0.9425 - recall_m: 0.8462 - val_loss: 0.3236 - val_acc: 0.8832 - val_precision_m: 0.9811 - val_recall_m: 0.7922\n",
      "Epoch 411/5000\n",
      " - 10s - loss: 0.3048 - acc: 0.8959 - precision_m: 0.9435 - recall_m: 0.8438 - val_loss: 0.3009 - val_acc: 0.9070 - val_precision_m: 0.9533 - val_recall_m: 0.8564\n",
      "Epoch 412/5000\n",
      " - 10s - loss: 0.3085 - acc: 0.8954 - precision_m: 0.9424 - recall_m: 0.8452 - val_loss: 0.2992 - val_acc: 0.9009 - val_precision_m: 0.9566 - val_recall_m: 0.8406\n",
      "Epoch 413/5000\n",
      " - 10s - loss: 0.3023 - acc: 0.8960 - precision_m: 0.9421 - recall_m: 0.8460 - val_loss: 0.3164 - val_acc: 0.8832 - val_precision_m: 0.9760 - val_recall_m: 0.7889\n",
      "Epoch 414/5000\n",
      " - 10s - loss: 0.3049 - acc: 0.8938 - precision_m: 0.9406 - recall_m: 0.8433 - val_loss: 0.3020 - val_acc: 0.9009 - val_precision_m: 0.9311 - val_recall_m: 0.8739\n",
      "Epoch 415/5000\n",
      " - 10s - loss: 0.2986 - acc: 0.8991 - precision_m: 0.9440 - recall_m: 0.8496 - val_loss: 0.3250 - val_acc: 0.8760 - val_precision_m: 0.9793 - val_recall_m: 0.7651\n",
      "Epoch 416/5000\n",
      " - 10s - loss: 0.3142 - acc: 0.8840 - precision_m: 0.9308 - recall_m: 0.8375 - val_loss: 0.3169 - val_acc: 0.8870 - val_precision_m: 0.9786 - val_recall_m: 0.7938\n",
      "Epoch 417/5000\n",
      " - 10s - loss: 0.3063 - acc: 0.8935 - precision_m: 0.9412 - recall_m: 0.8438 - val_loss: 0.3023 - val_acc: 0.9020 - val_precision_m: 0.9456 - val_recall_m: 0.8614\n",
      "Epoch 418/5000\n",
      " - 10s - loss: 0.2999 - acc: 0.8964 - precision_m: 0.9447 - recall_m: 0.8442 - val_loss: 0.3003 - val_acc: 0.9092 - val_precision_m: 0.9573 - val_recall_m: 0.8635\n",
      "Epoch 419/5000\n",
      " - 10s - loss: 0.3073 - acc: 0.8903 - precision_m: 0.9362 - recall_m: 0.8428 - val_loss: 0.3120 - val_acc: 0.8865 - val_precision_m: 0.9660 - val_recall_m: 0.8119\n",
      "Epoch 420/5000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " - 10s - loss: 0.3007 - acc: 0.8984 - precision_m: 0.9427 - recall_m: 0.8510 - val_loss: 0.2990 - val_acc: 0.9037 - val_precision_m: 0.9426 - val_recall_m: 0.8671\n",
      "Epoch 421/5000\n",
      " - 10s - loss: 0.2973 - acc: 0.8985 - precision_m: 0.9441 - recall_m: 0.8489 - val_loss: 0.3181 - val_acc: 0.8826 - val_precision_m: 0.9782 - val_recall_m: 0.7860\n",
      "Epoch 422/5000\n",
      " - 10s - loss: 0.3016 - acc: 0.8958 - precision_m: 0.9433 - recall_m: 0.8443 - val_loss: 0.3041 - val_acc: 0.8953 - val_precision_m: 0.9654 - val_recall_m: 0.8216\n",
      "Epoch 423/5000\n",
      " - 10s - loss: 0.2975 - acc: 0.8967 - precision_m: 0.9428 - recall_m: 0.8464 - val_loss: 0.3058 - val_acc: 0.8948 - val_precision_m: 0.9709 - val_recall_m: 0.8156\n",
      "Epoch 424/5000\n",
      " - 10s - loss: 0.2977 - acc: 0.8980 - precision_m: 0.9439 - recall_m: 0.8480 - val_loss: 0.2970 - val_acc: 0.9042 - val_precision_m: 0.9579 - val_recall_m: 0.8532\n",
      "Epoch 425/5000\n",
      " - 10s - loss: 0.2993 - acc: 0.8942 - precision_m: 0.9386 - recall_m: 0.8462 - val_loss: 0.2955 - val_acc: 0.9020 - val_precision_m: 0.9619 - val_recall_m: 0.8379\n",
      "Epoch 426/5000\n",
      " - 10s - loss: 0.2975 - acc: 0.8997 - precision_m: 0.9440 - recall_m: 0.8518 - val_loss: 0.3200 - val_acc: 0.8793 - val_precision_m: 0.9835 - val_recall_m: 0.7758\n",
      "Epoch 427/5000\n",
      " - 10s - loss: 0.2955 - acc: 0.8977 - precision_m: 0.9442 - recall_m: 0.8472 - val_loss: 0.3118 - val_acc: 0.8887 - val_precision_m: 0.9754 - val_recall_m: 0.7999\n",
      "Epoch 428/5000\n",
      " - 10s - loss: 0.2922 - acc: 0.8983 - precision_m: 0.9451 - recall_m: 0.8474 - val_loss: 0.2916 - val_acc: 0.9070 - val_precision_m: 0.9587 - val_recall_m: 0.8502\n",
      "Epoch 429/5000\n",
      " - 10s - loss: 0.2925 - acc: 0.9016 - precision_m: 0.9471 - recall_m: 0.8515 - val_loss: 0.2957 - val_acc: 0.9031 - val_precision_m: 0.9498 - val_recall_m: 0.8587\n",
      "Epoch 430/5000\n",
      " - 10s - loss: 0.2912 - acc: 0.9010 - precision_m: 0.9461 - recall_m: 0.8510 - val_loss: 0.3053 - val_acc: 0.8942 - val_precision_m: 0.9072 - val_recall_m: 0.8882\n",
      "Epoch 431/5000\n",
      " - 10s - loss: 0.2935 - acc: 0.9007 - precision_m: 0.9466 - recall_m: 0.8510 - val_loss: 0.2974 - val_acc: 0.8959 - val_precision_m: 0.9735 - val_recall_m: 0.8153\n",
      "Epoch 432/5000\n",
      " - 10s - loss: 0.2911 - acc: 0.8991 - precision_m: 0.9443 - recall_m: 0.8497 - val_loss: 0.3037 - val_acc: 0.8882 - val_precision_m: 0.9778 - val_recall_m: 0.8043\n",
      "Epoch 433/5000\n",
      " - 10s - loss: 0.2915 - acc: 0.9006 - precision_m: 0.9480 - recall_m: 0.8490 - val_loss: 0.3020 - val_acc: 0.8959 - val_precision_m: 0.9722 - val_recall_m: 0.8240\n",
      "Epoch 434/5000\n",
      " - 10s - loss: 0.2904 - acc: 0.8995 - precision_m: 0.9452 - recall_m: 0.8502 - val_loss: 0.2939 - val_acc: 0.9037 - val_precision_m: 0.9491 - val_recall_m: 0.8602\n",
      "Epoch 435/5000\n",
      " - 10s - loss: 0.2948 - acc: 0.8967 - precision_m: 0.9426 - recall_m: 0.8473 - val_loss: 0.3069 - val_acc: 0.8859 - val_precision_m: 0.9777 - val_recall_m: 0.7931\n",
      "Epoch 436/5000\n",
      " - 10s - loss: 0.2939 - acc: 0.8996 - precision_m: 0.9439 - recall_m: 0.8516 - val_loss: 0.3209 - val_acc: 0.8765 - val_precision_m: 0.9809 - val_recall_m: 0.7655\n",
      "Epoch 437/5000\n",
      " - 10s - loss: 0.2906 - acc: 0.9010 - precision_m: 0.9463 - recall_m: 0.8516 - val_loss: 0.3049 - val_acc: 0.8865 - val_precision_m: 0.9775 - val_recall_m: 0.7934\n",
      "Epoch 438/5000\n",
      " - 10s - loss: 0.2909 - acc: 0.8993 - precision_m: 0.9448 - recall_m: 0.8496 - val_loss: 0.2870 - val_acc: 0.9053 - val_precision_m: 0.9596 - val_recall_m: 0.8530\n",
      "Epoch 439/5000\n",
      " - 10s - loss: 0.2894 - acc: 0.8992 - precision_m: 0.9426 - recall_m: 0.8525 - val_loss: 0.2979 - val_acc: 0.9009 - val_precision_m: 0.9149 - val_recall_m: 0.8931\n",
      "Epoch 440/5000\n",
      " - 10s - loss: 0.2920 - acc: 0.9005 - precision_m: 0.9460 - recall_m: 0.8511 - val_loss: 0.2956 - val_acc: 0.8959 - val_precision_m: 0.9745 - val_recall_m: 0.8147\n",
      "Epoch 441/5000\n",
      " - 10s - loss: 0.2849 - acc: 0.9021 - precision_m: 0.9455 - recall_m: 0.8538 - val_loss: 0.2857 - val_acc: 0.9131 - val_precision_m: 0.9691 - val_recall_m: 0.8592\n",
      "Epoch 442/5000\n",
      " - 10s - loss: 0.2890 - acc: 0.9018 - precision_m: 0.9456 - recall_m: 0.8544 - val_loss: 0.3098 - val_acc: 0.8743 - val_precision_m: 0.9806 - val_recall_m: 0.7610\n",
      "Epoch 443/5000\n",
      " - 10s - loss: 0.2868 - acc: 0.9037 - precision_m: 0.9495 - recall_m: 0.8544 - val_loss: 0.3283 - val_acc: 0.8660 - val_precision_m: 0.9853 - val_recall_m: 0.7413\n",
      "Epoch 444/5000\n",
      " - 10s - loss: 0.2901 - acc: 0.8999 - precision_m: 0.9437 - recall_m: 0.8525 - val_loss: 0.2873 - val_acc: 0.9053 - val_precision_m: 0.9487 - val_recall_m: 0.8642\n",
      "Epoch 445/5000\n",
      " - 10s - loss: 0.2878 - acc: 0.9027 - precision_m: 0.9468 - recall_m: 0.8551 - val_loss: 0.2884 - val_acc: 0.9070 - val_precision_m: 0.9556 - val_recall_m: 0.8602\n",
      "Epoch 446/5000\n",
      " - 10s - loss: 0.2868 - acc: 0.9012 - precision_m: 0.9436 - recall_m: 0.8569 - val_loss: 0.2867 - val_acc: 0.9147 - val_precision_m: 0.9486 - val_recall_m: 0.8831\n",
      "Epoch 447/5000\n",
      " - 10s - loss: 0.2859 - acc: 0.9009 - precision_m: 0.9457 - recall_m: 0.8529 - val_loss: 0.2898 - val_acc: 0.9064 - val_precision_m: 0.9467 - val_recall_m: 0.8686\n",
      "Epoch 448/5000\n",
      " - 10s - loss: 0.2846 - acc: 0.9028 - precision_m: 0.9472 - recall_m: 0.8543 - val_loss: 0.2913 - val_acc: 0.9075 - val_precision_m: 0.9296 - val_recall_m: 0.8894\n",
      "Epoch 449/5000\n",
      " - 10s - loss: 0.2807 - acc: 0.9050 - precision_m: 0.9512 - recall_m: 0.8538 - val_loss: 0.3124 - val_acc: 0.8821 - val_precision_m: 0.9797 - val_recall_m: 0.7845\n",
      "Epoch 450/5000\n",
      " - 10s - loss: 0.2849 - acc: 0.9016 - precision_m: 0.9449 - recall_m: 0.8551 - val_loss: 0.2925 - val_acc: 0.9009 - val_precision_m: 0.9724 - val_recall_m: 0.8329\n",
      "Epoch 451/5000\n",
      " - 10s - loss: 0.2861 - acc: 0.9018 - precision_m: 0.9464 - recall_m: 0.8538 - val_loss: 0.2868 - val_acc: 0.9075 - val_precision_m: 0.9439 - val_recall_m: 0.8664\n",
      "Epoch 452/5000\n",
      " - 10s - loss: 0.2804 - acc: 0.9040 - precision_m: 0.9498 - recall_m: 0.8537 - val_loss: 0.2879 - val_acc: 0.9014 - val_precision_m: 0.9627 - val_recall_m: 0.8429\n",
      "Epoch 453/5000\n",
      " - 10s - loss: 0.2873 - acc: 0.8997 - precision_m: 0.9446 - recall_m: 0.8530 - val_loss: 0.2833 - val_acc: 0.9092 - val_precision_m: 0.9507 - val_recall_m: 0.8699\n",
      "Epoch 454/5000\n",
      " - 10s - loss: 0.2772 - acc: 0.9077 - precision_m: 0.9523 - recall_m: 0.8591 - val_loss: 0.2894 - val_acc: 0.8937 - val_precision_m: 0.9686 - val_recall_m: 0.8227\n",
      "Epoch 455/5000\n",
      " - 10s - loss: 0.2794 - acc: 0.9048 - precision_m: 0.9472 - recall_m: 0.8589 - val_loss: 0.2877 - val_acc: 0.9014 - val_precision_m: 0.9474 - val_recall_m: 0.8581\n",
      "Epoch 456/5000\n",
      " - 10s - loss: 0.2803 - acc: 0.9023 - precision_m: 0.9481 - recall_m: 0.8530 - val_loss: 0.3008 - val_acc: 0.8898 - val_precision_m: 0.9777 - val_recall_m: 0.8070\n",
      "Epoch 457/5000\n",
      " - 10s - loss: 0.2862 - acc: 0.8994 - precision_m: 0.9420 - recall_m: 0.8547 - val_loss: 0.2809 - val_acc: 0.9053 - val_precision_m: 0.9555 - val_recall_m: 0.8504\n",
      "Epoch 458/5000\n",
      " - 10s - loss: 0.2748 - acc: 0.9093 - precision_m: 0.9532 - recall_m: 0.8608 - val_loss: 0.2817 - val_acc: 0.9081 - val_precision_m: 0.9596 - val_recall_m: 0.8580\n",
      "Epoch 459/5000\n",
      " - 10s - loss: 0.2846 - acc: 0.8988 - precision_m: 0.9393 - recall_m: 0.8558 - val_loss: 0.3080 - val_acc: 0.8821 - val_precision_m: 0.9833 - val_recall_m: 0.7879\n",
      "Epoch 460/5000\n",
      " - 10s - loss: 0.2786 - acc: 0.9039 - precision_m: 0.9472 - recall_m: 0.8564 - val_loss: 0.2826 - val_acc: 0.9092 - val_precision_m: 0.9426 - val_recall_m: 0.8778\n",
      "Epoch 461/5000\n",
      " - 10s - loss: 0.2754 - acc: 0.9043 - precision_m: 0.9483 - recall_m: 0.8557 - val_loss: 0.2778 - val_acc: 0.9131 - val_precision_m: 0.9543 - val_recall_m: 0.8738\n",
      "Epoch 462/5000\n",
      " - 10s - loss: 0.2772 - acc: 0.9042 - precision_m: 0.9464 - recall_m: 0.8586 - val_loss: 0.2899 - val_acc: 0.9009 - val_precision_m: 0.9770 - val_recall_m: 0.8288\n",
      "Epoch 463/5000\n",
      " - 10s - loss: 0.2795 - acc: 0.9026 - precision_m: 0.9462 - recall_m: 0.8569 - val_loss: 0.2882 - val_acc: 0.8992 - val_precision_m: 0.9757 - val_recall_m: 0.8266\n",
      "Epoch 464/5000\n",
      " - 10s - loss: 0.2785 - acc: 0.9049 - precision_m: 0.9482 - recall_m: 0.8587 - val_loss: 0.2882 - val_acc: 0.8992 - val_precision_m: 0.9747 - val_recall_m: 0.8207\n",
      "Epoch 465/5000\n",
      " - 10s - loss: 0.2763 - acc: 0.9064 - precision_m: 0.9494 - recall_m: 0.8601 - val_loss: 0.3139 - val_acc: 0.8782 - val_precision_m: 0.9832 - val_recall_m: 0.7664\n",
      "Epoch 466/5000\n",
      " - 10s - loss: 0.2742 - acc: 0.9080 - precision_m: 0.9511 - recall_m: 0.8609 - val_loss: 0.2892 - val_acc: 0.8920 - val_precision_m: 0.9744 - val_recall_m: 0.8069\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 467/5000\n",
      " - 10s - loss: 0.2725 - acc: 0.9072 - precision_m: 0.9534 - recall_m: 0.8568 - val_loss: 0.2796 - val_acc: 0.9059 - val_precision_m: 0.9719 - val_recall_m: 0.8354\n",
      "Epoch 468/5000\n",
      " - 10s - loss: 0.2752 - acc: 0.9048 - precision_m: 0.9447 - recall_m: 0.8620 - val_loss: 0.2798 - val_acc: 0.9053 - val_precision_m: 0.9275 - val_recall_m: 0.8870\n",
      "Epoch 469/5000\n",
      " - 10s - loss: 0.2766 - acc: 0.9045 - precision_m: 0.9475 - recall_m: 0.8589 - val_loss: 0.3093 - val_acc: 0.8765 - val_precision_m: 0.8623 - val_recall_m: 0.9125\n",
      "Epoch 470/5000\n",
      " - 10s - loss: 0.2760 - acc: 0.9058 - precision_m: 0.9464 - recall_m: 0.8618 - val_loss: 0.2774 - val_acc: 0.9059 - val_precision_m: 0.9724 - val_recall_m: 0.8348\n",
      "Epoch 471/5000\n",
      " - 10s - loss: 0.2762 - acc: 0.9050 - precision_m: 0.9463 - recall_m: 0.8614 - val_loss: 0.2813 - val_acc: 0.8976 - val_precision_m: 0.9725 - val_recall_m: 0.8194\n",
      "Epoch 472/5000\n",
      " - 10s - loss: 0.2791 - acc: 0.9020 - precision_m: 0.9447 - recall_m: 0.8579 - val_loss: 0.2897 - val_acc: 0.8898 - val_precision_m: 0.9801 - val_recall_m: 0.8053\n",
      "Epoch 473/5000\n",
      " - 10s - loss: 0.2768 - acc: 0.9028 - precision_m: 0.9451 - recall_m: 0.8567 - val_loss: 0.2910 - val_acc: 0.9037 - val_precision_m: 0.9160 - val_recall_m: 0.8902\n",
      "Epoch 474/5000\n",
      " - 10s - loss: 0.2693 - acc: 0.9101 - precision_m: 0.9527 - recall_m: 0.8638 - val_loss: 0.2684 - val_acc: 0.9120 - val_precision_m: 0.9645 - val_recall_m: 0.8612\n",
      "Epoch 475/5000\n",
      " - 10s - loss: 0.2730 - acc: 0.9052 - precision_m: 0.9463 - recall_m: 0.8607 - val_loss: 0.2822 - val_acc: 0.9003 - val_precision_m: 0.9199 - val_recall_m: 0.8852\n",
      "Epoch 476/5000\n",
      " - 10s - loss: 0.2667 - acc: 0.9085 - precision_m: 0.9499 - recall_m: 0.8628 - val_loss: 0.2786 - val_acc: 0.8981 - val_precision_m: 0.9664 - val_recall_m: 0.8326\n",
      "Epoch 477/5000\n",
      " - 10s - loss: 0.2687 - acc: 0.9066 - precision_m: 0.9504 - recall_m: 0.8595 - val_loss: 0.2734 - val_acc: 0.9081 - val_precision_m: 0.9623 - val_recall_m: 0.8559\n",
      "Epoch 478/5000\n",
      " - 10s - loss: 0.2700 - acc: 0.9088 - precision_m: 0.9518 - recall_m: 0.8633 - val_loss: 0.2687 - val_acc: 0.9142 - val_precision_m: 0.9656 - val_recall_m: 0.8640\n",
      "Epoch 479/5000\n",
      " - 10s - loss: 0.2721 - acc: 0.9057 - precision_m: 0.9489 - recall_m: 0.8600 - val_loss: 0.2890 - val_acc: 0.9064 - val_precision_m: 0.9079 - val_recall_m: 0.9126\n",
      "Epoch 480/5000\n",
      " - 10s - loss: 0.2717 - acc: 0.9056 - precision_m: 0.9511 - recall_m: 0.8578 - val_loss: 0.2733 - val_acc: 0.9103 - val_precision_m: 0.9380 - val_recall_m: 0.8848\n",
      "Epoch 481/5000\n",
      " - 10s - loss: 0.2690 - acc: 0.9067 - precision_m: 0.9464 - recall_m: 0.8643 - val_loss: 0.3162 - val_acc: 0.8754 - val_precision_m: 0.9807 - val_recall_m: 0.7709\n",
      "Epoch 482/5000\n",
      " - 10s - loss: 0.2712 - acc: 0.9059 - precision_m: 0.9484 - recall_m: 0.8605 - val_loss: 0.2720 - val_acc: 0.9114 - val_precision_m: 0.9412 - val_recall_m: 0.8835\n",
      "Epoch 483/5000\n",
      " - 10s - loss: 0.2675 - acc: 0.9088 - precision_m: 0.9503 - recall_m: 0.8640 - val_loss: 0.2985 - val_acc: 0.8876 - val_precision_m: 0.9825 - val_recall_m: 0.7994\n",
      "Epoch 484/5000\n",
      " - 10s - loss: 0.2672 - acc: 0.9069 - precision_m: 0.9487 - recall_m: 0.8614 - val_loss: 0.2717 - val_acc: 0.9164 - val_precision_m: 0.9452 - val_recall_m: 0.8906\n",
      "Epoch 485/5000\n",
      " - 10s - loss: 0.2676 - acc: 0.9071 - precision_m: 0.9495 - recall_m: 0.8613 - val_loss: 0.2797 - val_acc: 0.9014 - val_precision_m: 0.9738 - val_recall_m: 0.8327\n",
      "Epoch 486/5000\n",
      " - 10s - loss: 0.2724 - acc: 0.9031 - precision_m: 0.9449 - recall_m: 0.8595 - val_loss: 0.2659 - val_acc: 0.9147 - val_precision_m: 0.9535 - val_recall_m: 0.8701\n",
      "Epoch 487/5000\n",
      " - 10s - loss: 0.2654 - acc: 0.9077 - precision_m: 0.9500 - recall_m: 0.8621 - val_loss: 0.2652 - val_acc: 0.9158 - val_precision_m: 0.9658 - val_recall_m: 0.8674\n",
      "Epoch 488/5000\n",
      " - 10s - loss: 0.2658 - acc: 0.9096 - precision_m: 0.9495 - recall_m: 0.8664 - val_loss: 0.2667 - val_acc: 0.9103 - val_precision_m: 0.9659 - val_recall_m: 0.8504\n",
      "Epoch 489/5000\n",
      " - 10s - loss: 0.2665 - acc: 0.9066 - precision_m: 0.9492 - recall_m: 0.8622 - val_loss: 0.2709 - val_acc: 0.9075 - val_precision_m: 0.9668 - val_recall_m: 0.8515\n",
      "Epoch 490/5000\n",
      " - 10s - loss: 0.2706 - acc: 0.9056 - precision_m: 0.9480 - recall_m: 0.8614 - val_loss: 0.2684 - val_acc: 0.9075 - val_precision_m: 0.9558 - val_recall_m: 0.8614\n",
      "Epoch 491/5000\n",
      " - 10s - loss: 0.2667 - acc: 0.9057 - precision_m: 0.9492 - recall_m: 0.8604 - val_loss: 0.2650 - val_acc: 0.9131 - val_precision_m: 0.9443 - val_recall_m: 0.8843\n",
      "Epoch 492/5000\n",
      " - 10s - loss: 0.2631 - acc: 0.9109 - precision_m: 0.9527 - recall_m: 0.8659 - val_loss: 0.2724 - val_acc: 0.9120 - val_precision_m: 0.9229 - val_recall_m: 0.9058\n",
      "Epoch 493/5000\n",
      " - 10s - loss: 0.2666 - acc: 0.9076 - precision_m: 0.9499 - recall_m: 0.8624 - val_loss: 0.2685 - val_acc: 0.9059 - val_precision_m: 0.9717 - val_recall_m: 0.8356\n",
      "Epoch 494/5000\n",
      " - 10s - loss: 0.2624 - acc: 0.9095 - precision_m: 0.9526 - recall_m: 0.8638 - val_loss: 0.2658 - val_acc: 0.9131 - val_precision_m: 0.9447 - val_recall_m: 0.8841\n",
      "Epoch 495/5000\n",
      " - 10s - loss: 0.2612 - acc: 0.9105 - precision_m: 0.9522 - recall_m: 0.8663 - val_loss: 0.2663 - val_acc: 0.9164 - val_precision_m: 0.9620 - val_recall_m: 0.8729\n",
      "Epoch 496/5000\n",
      " - 10s - loss: 0.2577 - acc: 0.9157 - precision_m: 0.9564 - recall_m: 0.8716 - val_loss: 0.2800 - val_acc: 0.8998 - val_precision_m: 0.9793 - val_recall_m: 0.8246\n",
      "Epoch 497/5000\n",
      " - 10s - loss: 0.2594 - acc: 0.9119 - precision_m: 0.9533 - recall_m: 0.8670 - val_loss: 0.2639 - val_acc: 0.9114 - val_precision_m: 0.9624 - val_recall_m: 0.8628\n",
      "Epoch 498/5000\n",
      " - 10s - loss: 0.2653 - acc: 0.9077 - precision_m: 0.9491 - recall_m: 0.8643 - val_loss: 0.2641 - val_acc: 0.9136 - val_precision_m: 0.9673 - val_recall_m: 0.8615\n",
      "Epoch 499/5000\n",
      " - 10s - loss: 0.2581 - acc: 0.9119 - precision_m: 0.9518 - recall_m: 0.8687 - val_loss: 0.2700 - val_acc: 0.9081 - val_precision_m: 0.9750 - val_recall_m: 0.8368\n",
      "Epoch 500/5000\n",
      " - 10s - loss: 0.2636 - acc: 0.9109 - precision_m: 0.9528 - recall_m: 0.8673 - val_loss: 0.2738 - val_acc: 0.9131 - val_precision_m: 0.9287 - val_recall_m: 0.9017\n",
      "Epoch 501/5000\n",
      " - 10s - loss: 0.2594 - acc: 0.9123 - precision_m: 0.9542 - recall_m: 0.8673 - val_loss: 0.2621 - val_acc: 0.9192 - val_precision_m: 0.9562 - val_recall_m: 0.8841\n",
      "Epoch 502/5000\n",
      " - 10s - loss: 0.2568 - acc: 0.9122 - precision_m: 0.9529 - recall_m: 0.8682 - val_loss: 0.2809 - val_acc: 0.8976 - val_precision_m: 0.9001 - val_recall_m: 0.9048\n",
      "Epoch 503/5000\n",
      " - 10s - loss: 0.2647 - acc: 0.9075 - precision_m: 0.9484 - recall_m: 0.8647 - val_loss: 0.2862 - val_acc: 0.9003 - val_precision_m: 0.8984 - val_recall_m: 0.9203\n",
      "Epoch 504/5000\n",
      " - 10s - loss: 0.2600 - acc: 0.9107 - precision_m: 0.9528 - recall_m: 0.8664 - val_loss: 0.2560 - val_acc: 0.9219 - val_precision_m: 0.9652 - val_recall_m: 0.8798\n",
      "Epoch 505/5000\n",
      " - 10s - loss: 0.2552 - acc: 0.9152 - precision_m: 0.9558 - recall_m: 0.8715 - val_loss: 0.2723 - val_acc: 0.9025 - val_precision_m: 0.9817 - val_recall_m: 0.8206\n",
      "Epoch 506/5000\n",
      " - 10s - loss: 0.2594 - acc: 0.9116 - precision_m: 0.9532 - recall_m: 0.8674 - val_loss: 0.2610 - val_acc: 0.9181 - val_precision_m: 0.9410 - val_recall_m: 0.8976\n",
      "Epoch 507/5000\n",
      " - 10s - loss: 0.2584 - acc: 0.9114 - precision_m: 0.9511 - recall_m: 0.8688 - val_loss: 0.2660 - val_acc: 0.9136 - val_precision_m: 0.9348 - val_recall_m: 0.8894\n",
      "Epoch 508/5000\n",
      " - 10s - loss: 0.2567 - acc: 0.9133 - precision_m: 0.9528 - recall_m: 0.8704 - val_loss: 0.2632 - val_acc: 0.9125 - val_precision_m: 0.9723 - val_recall_m: 0.8558\n",
      "Epoch 509/5000\n",
      " - 10s - loss: 0.2552 - acc: 0.9116 - precision_m: 0.9511 - recall_m: 0.8694 - val_loss: 0.2581 - val_acc: 0.9164 - val_precision_m: 0.9584 - val_recall_m: 0.8757\n",
      "Epoch 510/5000\n",
      " - 10s - loss: 0.2550 - acc: 0.9141 - precision_m: 0.9542 - recall_m: 0.8715 - val_loss: 0.2620 - val_acc: 0.9175 - val_precision_m: 0.9346 - val_recall_m: 0.9039\n",
      "Epoch 511/5000\n",
      " - 10s - loss: 0.2572 - acc: 0.9126 - precision_m: 0.9519 - recall_m: 0.8699 - val_loss: 0.2567 - val_acc: 0.9175 - val_precision_m: 0.9662 - val_recall_m: 0.8710\n",
      "Epoch 512/5000\n",
      " - 10s - loss: 0.2559 - acc: 0.9124 - precision_m: 0.9520 - recall_m: 0.8699 - val_loss: 0.2691 - val_acc: 0.9037 - val_precision_m: 0.9760 - val_recall_m: 0.8346\n",
      "Epoch 513/5000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " - 10s - loss: 0.2527 - acc: 0.9130 - precision_m: 0.9548 - recall_m: 0.8681 - val_loss: 0.2568 - val_acc: 0.9120 - val_precision_m: 0.9387 - val_recall_m: 0.8884\n",
      "Epoch 514/5000\n",
      " - 10s - loss: 0.2585 - acc: 0.9092 - precision_m: 0.9502 - recall_m: 0.8650 - val_loss: 0.2568 - val_acc: 0.9158 - val_precision_m: 0.9504 - val_recall_m: 0.8825\n",
      "Epoch 515/5000\n",
      " - 10s - loss: 0.2561 - acc: 0.9117 - precision_m: 0.9501 - recall_m: 0.8706 - val_loss: 0.2593 - val_acc: 0.9175 - val_precision_m: 0.9413 - val_recall_m: 0.8966\n",
      "Epoch 516/5000\n",
      " - 10s - loss: 0.2499 - acc: 0.9168 - precision_m: 0.9563 - recall_m: 0.8744 - val_loss: 0.2590 - val_acc: 0.9158 - val_precision_m: 0.9407 - val_recall_m: 0.8933\n",
      "Epoch 517/5000\n",
      " - 10s - loss: 0.2545 - acc: 0.9143 - precision_m: 0.9530 - recall_m: 0.8724 - val_loss: 0.2712 - val_acc: 0.9031 - val_precision_m: 0.9067 - val_recall_m: 0.9081\n",
      "Epoch 518/5000\n",
      " - 10s - loss: 0.2606 - acc: 0.9069 - precision_m: 0.9473 - recall_m: 0.8659 - val_loss: 0.2715 - val_acc: 0.9053 - val_precision_m: 0.9738 - val_recall_m: 0.8397\n",
      "Epoch 519/5000\n",
      " - 10s - loss: 0.2546 - acc: 0.9124 - precision_m: 0.9525 - recall_m: 0.8711 - val_loss: 0.2500 - val_acc: 0.9208 - val_precision_m: 0.9589 - val_recall_m: 0.8839\n",
      "Epoch 520/5000\n",
      " - 10s - loss: 0.2517 - acc: 0.9146 - precision_m: 0.9537 - recall_m: 0.8726 - val_loss: 0.2548 - val_acc: 0.9169 - val_precision_m: 0.9552 - val_recall_m: 0.8808\n",
      "Epoch 521/5000\n",
      " - 10s - loss: 0.2519 - acc: 0.9127 - precision_m: 0.9523 - recall_m: 0.8708 - val_loss: 0.2637 - val_acc: 0.9020 - val_precision_m: 0.9783 - val_recall_m: 0.8224\n",
      "Epoch 522/5000\n",
      " - 10s - loss: 0.2529 - acc: 0.9115 - precision_m: 0.9495 - recall_m: 0.8712 - val_loss: 0.2535 - val_acc: 0.9186 - val_precision_m: 0.9671 - val_recall_m: 0.8714\n",
      "Epoch 523/5000\n",
      " - 10s - loss: 0.2495 - acc: 0.9127 - precision_m: 0.9561 - recall_m: 0.8658 - val_loss: 0.2537 - val_acc: 0.9208 - val_precision_m: 0.9535 - val_recall_m: 0.8905\n",
      "Epoch 524/5000\n",
      " - 10s - loss: 0.2506 - acc: 0.9136 - precision_m: 0.9525 - recall_m: 0.8724 - val_loss: 0.2545 - val_acc: 0.9097 - val_precision_m: 0.9713 - val_recall_m: 0.8512\n",
      "Epoch 525/5000\n",
      " - 10s - loss: 0.2485 - acc: 0.9139 - precision_m: 0.9517 - recall_m: 0.8726 - val_loss: 0.2653 - val_acc: 0.9009 - val_precision_m: 0.9792 - val_recall_m: 0.8196\n",
      "Epoch 526/5000\n",
      " - 10s - loss: 0.2472 - acc: 0.9165 - precision_m: 0.9543 - recall_m: 0.8759 - val_loss: 0.2536 - val_acc: 0.9142 - val_precision_m: 0.9596 - val_recall_m: 0.8712\n",
      "Epoch 527/5000\n",
      " - 10s - loss: 0.2497 - acc: 0.9130 - precision_m: 0.9535 - recall_m: 0.8702 - val_loss: 0.2465 - val_acc: 0.9219 - val_precision_m: 0.9609 - val_recall_m: 0.8840\n",
      "Epoch 528/5000\n",
      " - 10s - loss: 0.2503 - acc: 0.9129 - precision_m: 0.9530 - recall_m: 0.8701 - val_loss: 0.2658 - val_acc: 0.9042 - val_precision_m: 0.9792 - val_recall_m: 0.8256\n",
      "Epoch 529/5000\n",
      " - 10s - loss: 0.2474 - acc: 0.9143 - precision_m: 0.9541 - recall_m: 0.8718 - val_loss: 0.2498 - val_acc: 0.9236 - val_precision_m: 0.9674 - val_recall_m: 0.8812\n",
      "Epoch 530/5000\n",
      " - 10s - loss: 0.2482 - acc: 0.9150 - precision_m: 0.9527 - recall_m: 0.8746 - val_loss: 0.2513 - val_acc: 0.9181 - val_precision_m: 0.9481 - val_recall_m: 0.8835\n",
      "Epoch 531/5000\n",
      " - 10s - loss: 0.2448 - acc: 0.9192 - precision_m: 0.9587 - recall_m: 0.8768 - val_loss: 0.3063 - val_acc: 0.8726 - val_precision_m: 0.9816 - val_recall_m: 0.7637\n",
      "Epoch 532/5000\n",
      " - 10s - loss: 0.2437 - acc: 0.9178 - precision_m: 0.9583 - recall_m: 0.8752 - val_loss: 0.2543 - val_acc: 0.9186 - val_precision_m: 0.9374 - val_recall_m: 0.9038\n",
      "Epoch 533/5000\n",
      " - 10s - loss: 0.2440 - acc: 0.9162 - precision_m: 0.9547 - recall_m: 0.8745 - val_loss: 0.2508 - val_acc: 0.9214 - val_precision_m: 0.9432 - val_recall_m: 0.9014\n",
      "Epoch 534/5000\n",
      " - 10s - loss: 0.2458 - acc: 0.9166 - precision_m: 0.9556 - recall_m: 0.8749 - val_loss: 0.2581 - val_acc: 0.9097 - val_precision_m: 0.9743 - val_recall_m: 0.8483\n",
      "Epoch 535/5000\n",
      " - 10s - loss: 0.2455 - acc: 0.9170 - precision_m: 0.9564 - recall_m: 0.8756 - val_loss: 0.2453 - val_acc: 0.9225 - val_precision_m: 0.9587 - val_recall_m: 0.8883\n",
      "Epoch 536/5000\n",
      " - 10s - loss: 0.2481 - acc: 0.9138 - precision_m: 0.9530 - recall_m: 0.8718 - val_loss: 0.2471 - val_acc: 0.9175 - val_precision_m: 0.9642 - val_recall_m: 0.8730\n",
      "Epoch 537/5000\n",
      " - 10s - loss: 0.2425 - acc: 0.9185 - precision_m: 0.9572 - recall_m: 0.8770 - val_loss: 0.2493 - val_acc: 0.9169 - val_precision_m: 0.9548 - val_recall_m: 0.8816\n",
      "Epoch 538/5000\n",
      " - 10s - loss: 0.2496 - acc: 0.9128 - precision_m: 0.9526 - recall_m: 0.8701 - val_loss: 0.2464 - val_acc: 0.9169 - val_precision_m: 0.9755 - val_recall_m: 0.8532\n",
      "Epoch 539/5000\n",
      " - 10s - loss: 0.2431 - acc: 0.9170 - precision_m: 0.9558 - recall_m: 0.8752 - val_loss: 0.2464 - val_acc: 0.9203 - val_precision_m: 0.9670 - val_recall_m: 0.8754\n",
      "Epoch 540/5000\n",
      " - 10s - loss: 0.2427 - acc: 0.9166 - precision_m: 0.9558 - recall_m: 0.8752 - val_loss: 0.2528 - val_acc: 0.9175 - val_precision_m: 0.9358 - val_recall_m: 0.9031\n",
      "Epoch 541/5000\n",
      " - 10s - loss: 0.2460 - acc: 0.9147 - precision_m: 0.9532 - recall_m: 0.8738 - val_loss: 0.2529 - val_acc: 0.9158 - val_precision_m: 0.9366 - val_recall_m: 0.8985\n",
      "Epoch 542/5000\n",
      " - 10s - loss: 0.2477 - acc: 0.9134 - precision_m: 0.9516 - recall_m: 0.8728 - val_loss: 0.2691 - val_acc: 0.8981 - val_precision_m: 0.9792 - val_recall_m: 0.8216\n",
      "Epoch 543/5000\n",
      " - 10s - loss: 0.2416 - acc: 0.9173 - precision_m: 0.9575 - recall_m: 0.8744 - val_loss: 0.2445 - val_acc: 0.9225 - val_precision_m: 0.9595 - val_recall_m: 0.8872\n",
      "Epoch 544/5000\n",
      " - 10s - loss: 0.2432 - acc: 0.9168 - precision_m: 0.9540 - recall_m: 0.8779 - val_loss: 0.2520 - val_acc: 0.9103 - val_precision_m: 0.9741 - val_recall_m: 0.8492\n",
      "Epoch 545/5000\n",
      " - 10s - loss: 0.2450 - acc: 0.9154 - precision_m: 0.9556 - recall_m: 0.8733 - val_loss: 0.2587 - val_acc: 0.9064 - val_precision_m: 0.9772 - val_recall_m: 0.8392\n",
      "Epoch 546/5000\n",
      " - 10s - loss: 0.2398 - acc: 0.9164 - precision_m: 0.9533 - recall_m: 0.8776 - val_loss: 0.2387 - val_acc: 0.9225 - val_precision_m: 0.9675 - val_recall_m: 0.8793\n",
      "Epoch 547/5000\n",
      " - 10s - loss: 0.2407 - acc: 0.9174 - precision_m: 0.9552 - recall_m: 0.8776 - val_loss: 0.2569 - val_acc: 0.9064 - val_precision_m: 0.9104 - val_recall_m: 0.9093\n",
      "Epoch 548/5000\n",
      " - 10s - loss: 0.2404 - acc: 0.9184 - precision_m: 0.9553 - recall_m: 0.8784 - val_loss: 0.2688 - val_acc: 0.8959 - val_precision_m: 0.9815 - val_recall_m: 0.8078\n",
      "Epoch 549/5000\n",
      " - 10s - loss: 0.2486 - acc: 0.9120 - precision_m: 0.9485 - recall_m: 0.8759 - val_loss: 0.2449 - val_acc: 0.9219 - val_precision_m: 0.9505 - val_recall_m: 0.8956\n",
      "Epoch 550/5000\n",
      " - 10s - loss: 0.2387 - acc: 0.9185 - precision_m: 0.9583 - recall_m: 0.8764 - val_loss: 0.2467 - val_acc: 0.9192 - val_precision_m: 0.9359 - val_recall_m: 0.9056\n",
      "Epoch 551/5000\n",
      " - 10s - loss: 0.2389 - acc: 0.9175 - precision_m: 0.9559 - recall_m: 0.8767 - val_loss: 0.2640 - val_acc: 0.9092 - val_precision_m: 0.9104 - val_recall_m: 0.9158\n",
      "Epoch 552/5000\n",
      " - 10s - loss: 0.2382 - acc: 0.9184 - precision_m: 0.9587 - recall_m: 0.8755 - val_loss: 0.2429 - val_acc: 0.9208 - val_precision_m: 0.9757 - val_recall_m: 0.8677\n",
      "Epoch 553/5000\n",
      " - 10s - loss: 0.2361 - acc: 0.9203 - precision_m: 0.9595 - recall_m: 0.8779 - val_loss: 0.2448 - val_acc: 0.9175 - val_precision_m: 0.9658 - val_recall_m: 0.8711\n",
      "Epoch 554/5000\n",
      " - 10s - loss: 0.2452 - acc: 0.9149 - precision_m: 0.9539 - recall_m: 0.8741 - val_loss: 0.2576 - val_acc: 0.9197 - val_precision_m: 0.9274 - val_recall_m: 0.9170\n",
      "Epoch 555/5000\n",
      " - 10s - loss: 0.2409 - acc: 0.9159 - precision_m: 0.9523 - recall_m: 0.8766 - val_loss: 0.2473 - val_acc: 0.9136 - val_precision_m: 0.9776 - val_recall_m: 0.8524\n",
      "Epoch 556/5000\n",
      " - 10s - loss: 0.2364 - acc: 0.9192 - precision_m: 0.9575 - recall_m: 0.8783 - val_loss: 0.2374 - val_acc: 0.9291 - val_precision_m: 0.9648 - val_recall_m: 0.8946\n",
      "Epoch 557/5000\n",
      " - 10s - loss: 0.2457 - acc: 0.9143 - precision_m: 0.9513 - recall_m: 0.8756 - val_loss: 0.2499 - val_acc: 0.9164 - val_precision_m: 0.9350 - val_recall_m: 0.9006\n",
      "Epoch 558/5000\n",
      " - 10s - loss: 0.2382 - acc: 0.9189 - precision_m: 0.9562 - recall_m: 0.8796 - val_loss: 0.2399 - val_acc: 0.9241 - val_precision_m: 0.9595 - val_recall_m: 0.8903\n",
      "Epoch 559/5000\n",
      " - 10s - loss: 0.2370 - acc: 0.9192 - precision_m: 0.9554 - recall_m: 0.8809 - val_loss: 0.2428 - val_acc: 0.9192 - val_precision_m: 0.9748 - val_recall_m: 0.8660\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 560/5000\n",
      " - 10s - loss: 0.2331 - acc: 0.9188 - precision_m: 0.9562 - recall_m: 0.8777 - val_loss: 0.2554 - val_acc: 0.9037 - val_precision_m: 0.9806 - val_recall_m: 0.8311\n",
      "Epoch 561/5000\n",
      " - 10s - loss: 0.2414 - acc: 0.9165 - precision_m: 0.9539 - recall_m: 0.8767 - val_loss: 0.2491 - val_acc: 0.9097 - val_precision_m: 0.9773 - val_recall_m: 0.8455\n",
      "Epoch 562/5000\n",
      " - 10s - loss: 0.2347 - acc: 0.9194 - precision_m: 0.9573 - recall_m: 0.8799 - val_loss: 0.2647 - val_acc: 0.9003 - val_precision_m: 0.9794 - val_recall_m: 0.8185\n",
      "Epoch 563/5000\n",
      " - 10s - loss: 0.2342 - acc: 0.9209 - precision_m: 0.9591 - recall_m: 0.8801 - val_loss: 0.2646 - val_acc: 0.8931 - val_precision_m: 0.9814 - val_recall_m: 0.8102\n",
      "Epoch 564/5000\n",
      " - 10s - loss: 0.2331 - acc: 0.9220 - precision_m: 0.9578 - recall_m: 0.8842 - val_loss: 0.2502 - val_acc: 0.9048 - val_precision_m: 0.9760 - val_recall_m: 0.8368\n",
      "Epoch 565/5000\n",
      " - 10s - loss: 0.2324 - acc: 0.9225 - precision_m: 0.9598 - recall_m: 0.8816 - val_loss: 0.2422 - val_acc: 0.9147 - val_precision_m: 0.9702 - val_recall_m: 0.8542\n",
      "Epoch 566/5000\n",
      " - 10s - loss: 0.2384 - acc: 0.9171 - precision_m: 0.9551 - recall_m: 0.8780 - val_loss: 0.2537 - val_acc: 0.9131 - val_precision_m: 0.9260 - val_recall_m: 0.9047\n",
      "Epoch 567/5000\n",
      " - 10s - loss: 0.2316 - acc: 0.9240 - precision_m: 0.9596 - recall_m: 0.8854 - val_loss: 0.2555 - val_acc: 0.9075 - val_precision_m: 0.9797 - val_recall_m: 0.8396\n",
      "Epoch 568/5000\n",
      " - 10s - loss: 0.2331 - acc: 0.9199 - precision_m: 0.9558 - recall_m: 0.8814 - val_loss: 0.2645 - val_acc: 0.9097 - val_precision_m: 0.9044 - val_recall_m: 0.9249\n",
      "Epoch 569/5000\n",
      " - 10s - loss: 0.2448 - acc: 0.9116 - precision_m: 0.9471 - recall_m: 0.8760 - val_loss: 0.2472 - val_acc: 0.9164 - val_precision_m: 0.9283 - val_recall_m: 0.9090\n",
      "Epoch 570/5000\n",
      " - 10s - loss: 0.2322 - acc: 0.9220 - precision_m: 0.9582 - recall_m: 0.8837 - val_loss: 0.2967 - val_acc: 0.8810 - val_precision_m: 0.9884 - val_recall_m: 0.7821\n",
      "Epoch 571/5000\n",
      " - 10s - loss: 0.2412 - acc: 0.9151 - precision_m: 0.9520 - recall_m: 0.8775 - val_loss: 0.2595 - val_acc: 0.9109 - val_precision_m: 0.9093 - val_recall_m: 0.9201\n",
      "Epoch 572/5000\n",
      " - 10s - loss: 0.2357 - acc: 0.9188 - precision_m: 0.9564 - recall_m: 0.8799 - val_loss: 0.2408 - val_acc: 0.9181 - val_precision_m: 0.9736 - val_recall_m: 0.8651\n",
      "Epoch 573/5000\n",
      " - 10s - loss: 0.2366 - acc: 0.9162 - precision_m: 0.9522 - recall_m: 0.8785 - val_loss: 0.2392 - val_acc: 0.9192 - val_precision_m: 0.9703 - val_recall_m: 0.8701\n",
      "Epoch 574/5000\n",
      " - 10s - loss: 0.2263 - acc: 0.9242 - precision_m: 0.9633 - recall_m: 0.8825 - val_loss: 0.2357 - val_acc: 0.9208 - val_precision_m: 0.9716 - val_recall_m: 0.8721\n",
      "Epoch 575/5000\n",
      " - 10s - loss: 0.2268 - acc: 0.9229 - precision_m: 0.9594 - recall_m: 0.8837 - val_loss: 0.2333 - val_acc: 0.9258 - val_precision_m: 0.9545 - val_recall_m: 0.8988\n",
      "Epoch 576/5000\n",
      " - 10s - loss: 0.2288 - acc: 0.9224 - precision_m: 0.9599 - recall_m: 0.8822 - val_loss: 0.2332 - val_acc: 0.9247 - val_precision_m: 0.9737 - val_recall_m: 0.8774\n",
      "Epoch 577/5000\n",
      " - 10s - loss: 0.2285 - acc: 0.9216 - precision_m: 0.9578 - recall_m: 0.8829 - val_loss: 0.2358 - val_acc: 0.9236 - val_precision_m: 0.9445 - val_recall_m: 0.9050\n",
      "Epoch 578/5000\n",
      " - 10s - loss: 0.2336 - acc: 0.9196 - precision_m: 0.9562 - recall_m: 0.8810 - val_loss: 0.2400 - val_acc: 0.9169 - val_precision_m: 0.9777 - val_recall_m: 0.8592\n",
      "Epoch 579/5000\n",
      " - 10s - loss: 0.2320 - acc: 0.9199 - precision_m: 0.9534 - recall_m: 0.8847 - val_loss: 0.2317 - val_acc: 0.9247 - val_precision_m: 0.9662 - val_recall_m: 0.8842\n",
      "Epoch 580/5000\n",
      " - 10s - loss: 0.2346 - acc: 0.9168 - precision_m: 0.9528 - recall_m: 0.8793 - val_loss: 0.2667 - val_acc: 0.8965 - val_precision_m: 0.9815 - val_recall_m: 0.8165\n",
      "Epoch 581/5000\n",
      " - 10s - loss: 0.2323 - acc: 0.9186 - precision_m: 0.9551 - recall_m: 0.8812 - val_loss: 0.2406 - val_acc: 0.9142 - val_precision_m: 0.9765 - val_recall_m: 0.8550\n",
      "Epoch 582/5000\n",
      " - 10s - loss: 0.2285 - acc: 0.9216 - precision_m: 0.9570 - recall_m: 0.8838 - val_loss: 0.2361 - val_acc: 0.9181 - val_precision_m: 0.9812 - val_recall_m: 0.8581\n",
      "Epoch 583/5000\n",
      " - 10s - loss: 0.2256 - acc: 0.9232 - precision_m: 0.9604 - recall_m: 0.8830 - val_loss: 0.2341 - val_acc: 0.9158 - val_precision_m: 0.9711 - val_recall_m: 0.8627\n",
      "Epoch 584/5000\n",
      " - 10s - loss: 0.2282 - acc: 0.9226 - precision_m: 0.9581 - recall_m: 0.8846 - val_loss: 0.2398 - val_acc: 0.9158 - val_precision_m: 0.9790 - val_recall_m: 0.8561\n",
      "Epoch 585/5000\n",
      " - 10s - loss: 0.2282 - acc: 0.9222 - precision_m: 0.9595 - recall_m: 0.8826 - val_loss: 0.2395 - val_acc: 0.9164 - val_precision_m: 0.9714 - val_recall_m: 0.8643\n",
      "Epoch 586/5000\n",
      " - 10s - loss: 0.2230 - acc: 0.9249 - precision_m: 0.9606 - recall_m: 0.8866 - val_loss: 0.2267 - val_acc: 0.9302 - val_precision_m: 0.9593 - val_recall_m: 0.9015\n",
      "Epoch 587/5000\n",
      " - 10s - loss: 0.2230 - acc: 0.9244 - precision_m: 0.9596 - recall_m: 0.8872 - val_loss: 0.2356 - val_acc: 0.9214 - val_precision_m: 0.9789 - val_recall_m: 0.8660\n",
      "Epoch 588/5000\n",
      " - 10s - loss: 0.2290 - acc: 0.9215 - precision_m: 0.9564 - recall_m: 0.8847 - val_loss: 0.2363 - val_acc: 0.9258 - val_precision_m: 0.9381 - val_recall_m: 0.9167\n",
      "Epoch 589/5000\n",
      " - 10s - loss: 0.2292 - acc: 0.9205 - precision_m: 0.9558 - recall_m: 0.8828 - val_loss: 0.2271 - val_acc: 0.9297 - val_precision_m: 0.9580 - val_recall_m: 0.9030\n",
      "Epoch 590/5000\n",
      " - 10s - loss: 0.2253 - acc: 0.9235 - precision_m: 0.9565 - recall_m: 0.8880 - val_loss: 0.2280 - val_acc: 0.9252 - val_precision_m: 0.9644 - val_recall_m: 0.8872\n",
      "Epoch 591/5000\n",
      " - 10s - loss: 0.2218 - acc: 0.9234 - precision_m: 0.9587 - recall_m: 0.8858 - val_loss: 0.2362 - val_acc: 0.9164 - val_precision_m: 0.9700 - val_recall_m: 0.8649\n",
      "Epoch 592/5000\n",
      " - 10s - loss: 0.2264 - acc: 0.9223 - precision_m: 0.9566 - recall_m: 0.8854 - val_loss: 0.2335 - val_acc: 0.9269 - val_precision_m: 0.9443 - val_recall_m: 0.9122\n",
      "Epoch 593/5000\n",
      " - 10s - loss: 0.2226 - acc: 0.9229 - precision_m: 0.9600 - recall_m: 0.8838 - val_loss: 0.2484 - val_acc: 0.9125 - val_precision_m: 0.9152 - val_recall_m: 0.9170\n",
      "Epoch 594/5000\n",
      " - 10s - loss: 0.2259 - acc: 0.9224 - precision_m: 0.9596 - recall_m: 0.8827 - val_loss: 0.2328 - val_acc: 0.9236 - val_precision_m: 0.9464 - val_recall_m: 0.9028\n",
      "Epoch 595/5000\n",
      " - 10s - loss: 0.2250 - acc: 0.9226 - precision_m: 0.9571 - recall_m: 0.8862 - val_loss: 0.2300 - val_acc: 0.9214 - val_precision_m: 0.9747 - val_recall_m: 0.8702\n",
      "Epoch 596/5000\n",
      " - 10s - loss: 0.2199 - acc: 0.9258 - precision_m: 0.9605 - recall_m: 0.8895 - val_loss: 0.2252 - val_acc: 0.9308 - val_precision_m: 0.9668 - val_recall_m: 0.8955\n",
      "Epoch 597/5000\n",
      " - 10s - loss: 0.2229 - acc: 0.9248 - precision_m: 0.9604 - recall_m: 0.8869 - val_loss: 0.2610 - val_acc: 0.9003 - val_precision_m: 0.9830 - val_recall_m: 0.8159\n",
      "Epoch 598/5000\n",
      " - 10s - loss: 0.2265 - acc: 0.9213 - precision_m: 0.9557 - recall_m: 0.8849 - val_loss: 0.2309 - val_acc: 0.9247 - val_precision_m: 0.9395 - val_recall_m: 0.9129\n",
      "Epoch 599/5000\n",
      " - 10s - loss: 0.2191 - acc: 0.9270 - precision_m: 0.9603 - recall_m: 0.8915 - val_loss: 0.2264 - val_acc: 0.9192 - val_precision_m: 0.9671 - val_recall_m: 0.8659\n",
      "Epoch 600/5000\n",
      " - 10s - loss: 0.2284 - acc: 0.9229 - precision_m: 0.9600 - recall_m: 0.8848 - val_loss: 0.2662 - val_acc: 0.8959 - val_precision_m: 0.9816 - val_recall_m: 0.8155\n",
      "Epoch 601/5000\n",
      " - 10s - loss: 0.2262 - acc: 0.9246 - precision_m: 0.9574 - recall_m: 0.8906 - val_loss: 0.2354 - val_acc: 0.9147 - val_precision_m: 0.9777 - val_recall_m: 0.8549\n",
      "Epoch 602/5000\n",
      " - 10s - loss: 0.2186 - acc: 0.9248 - precision_m: 0.9591 - recall_m: 0.8873 - val_loss: 0.2319 - val_acc: 0.9203 - val_precision_m: 0.9736 - val_recall_m: 0.8694\n",
      "Epoch 603/5000\n",
      " - 10s - loss: 0.2183 - acc: 0.9272 - precision_m: 0.9617 - recall_m: 0.8901 - val_loss: 0.2348 - val_acc: 0.9169 - val_precision_m: 0.9744 - val_recall_m: 0.8618\n",
      "Epoch 604/5000\n",
      " - 10s - loss: 0.2195 - acc: 0.9265 - precision_m: 0.9611 - recall_m: 0.8897 - val_loss: 0.2232 - val_acc: 0.9291 - val_precision_m: 0.9619 - val_recall_m: 0.9050\n",
      "Epoch 605/5000\n",
      " - 10s - loss: 0.2198 - acc: 0.9245 - precision_m: 0.9584 - recall_m: 0.8880 - val_loss: 0.2643 - val_acc: 0.8926 - val_precision_m: 0.9776 - val_recall_m: 0.8057\n",
      "Epoch 606/5000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " - 10s - loss: 0.2206 - acc: 0.9256 - precision_m: 0.9601 - recall_m: 0.8887 - val_loss: 0.2267 - val_acc: 0.9236 - val_precision_m: 0.9432 - val_recall_m: 0.9059\n",
      "Epoch 607/5000\n",
      " - 10s - loss: 0.2215 - acc: 0.9230 - precision_m: 0.9560 - recall_m: 0.8887 - val_loss: 0.2230 - val_acc: 0.9302 - val_precision_m: 0.9605 - val_recall_m: 0.9007\n",
      "Epoch 608/5000\n",
      " - 10s - loss: 0.2220 - acc: 0.9241 - precision_m: 0.9598 - recall_m: 0.8863 - val_loss: 0.2252 - val_acc: 0.9219 - val_precision_m: 0.9780 - val_recall_m: 0.8680\n",
      "Epoch 609/5000\n",
      " - 10s - loss: 0.2175 - acc: 0.9263 - precision_m: 0.9607 - recall_m: 0.8894 - val_loss: 0.2270 - val_acc: 0.9302 - val_precision_m: 0.9568 - val_recall_m: 0.9047\n",
      "Epoch 610/5000\n",
      " - 10s - loss: 0.2146 - acc: 0.9285 - precision_m: 0.9628 - recall_m: 0.8909 - val_loss: 0.2253 - val_acc: 0.9241 - val_precision_m: 0.9716 - val_recall_m: 0.8779\n",
      "Epoch 611/5000\n",
      " - 10s - loss: 0.2186 - acc: 0.9256 - precision_m: 0.9598 - recall_m: 0.8900 - val_loss: 0.2208 - val_acc: 0.9264 - val_precision_m: 0.9710 - val_recall_m: 0.8833\n",
      "Epoch 612/5000\n",
      " - 10s - loss: 0.2164 - acc: 0.9260 - precision_m: 0.9613 - recall_m: 0.8879 - val_loss: 0.2235 - val_acc: 0.9252 - val_precision_m: 0.9708 - val_recall_m: 0.8889\n",
      "Epoch 613/5000\n",
      " - 10s - loss: 0.2142 - acc: 0.9274 - precision_m: 0.9607 - recall_m: 0.8913 - val_loss: 0.2266 - val_acc: 0.9219 - val_precision_m: 0.9376 - val_recall_m: 0.9089\n",
      "Epoch 614/5000\n",
      " - 10s - loss: 0.2195 - acc: 0.9226 - precision_m: 0.9567 - recall_m: 0.8869 - val_loss: 0.2391 - val_acc: 0.9197 - val_precision_m: 0.9196 - val_recall_m: 0.9337\n",
      "Epoch 615/5000\n",
      " - 10s - loss: 0.2207 - acc: 0.9229 - precision_m: 0.9574 - recall_m: 0.8874 - val_loss: 0.2534 - val_acc: 0.8976 - val_precision_m: 0.9814 - val_recall_m: 0.8113\n",
      "Epoch 616/5000\n",
      " - 10s - loss: 0.2180 - acc: 0.9255 - precision_m: 0.9614 - recall_m: 0.8882 - val_loss: 0.2185 - val_acc: 0.9258 - val_precision_m: 0.9632 - val_recall_m: 0.8896\n",
      "Epoch 617/5000\n",
      " - 10s - loss: 0.2202 - acc: 0.9229 - precision_m: 0.9553 - recall_m: 0.8895 - val_loss: 0.2247 - val_acc: 0.9197 - val_precision_m: 0.9706 - val_recall_m: 0.8710\n",
      "Epoch 618/5000\n",
      " - 10s - loss: 0.2216 - acc: 0.9231 - precision_m: 0.9588 - recall_m: 0.8868 - val_loss: 0.2234 - val_acc: 0.9286 - val_precision_m: 0.9626 - val_recall_m: 0.9029\n",
      "Epoch 619/5000\n",
      " - 10s - loss: 0.2154 - acc: 0.9276 - precision_m: 0.9589 - recall_m: 0.8947 - val_loss: 0.2331 - val_acc: 0.9219 - val_precision_m: 0.9300 - val_recall_m: 0.9252\n",
      "Epoch 620/5000\n",
      " - 10s - loss: 0.2133 - acc: 0.9284 - precision_m: 0.9621 - recall_m: 0.8919 - val_loss: 0.2237 - val_acc: 0.9264 - val_precision_m: 0.9488 - val_recall_m: 0.9134\n",
      "Epoch 621/5000\n",
      " - 10s - loss: 0.2160 - acc: 0.9269 - precision_m: 0.9613 - recall_m: 0.8906 - val_loss: 0.2156 - val_acc: 0.9302 - val_precision_m: 0.9626 - val_recall_m: 0.8985\n",
      "Epoch 622/5000\n",
      " - 10s - loss: 0.2159 - acc: 0.9267 - precision_m: 0.9591 - recall_m: 0.8929 - val_loss: 0.2433 - val_acc: 0.9059 - val_precision_m: 0.9818 - val_recall_m: 0.8344\n",
      "Epoch 623/5000\n",
      " - 10s - loss: 0.2133 - acc: 0.9271 - precision_m: 0.9599 - recall_m: 0.8923 - val_loss: 0.2244 - val_acc: 0.9308 - val_precision_m: 0.9632 - val_recall_m: 0.9067\n",
      "Epoch 624/5000\n",
      " - 10s - loss: 0.2174 - acc: 0.9261 - precision_m: 0.9621 - recall_m: 0.8884 - val_loss: 0.2312 - val_acc: 0.9208 - val_precision_m: 0.9326 - val_recall_m: 0.9201\n",
      "Epoch 625/5000\n",
      " - 10s - loss: 0.2139 - acc: 0.9276 - precision_m: 0.9609 - recall_m: 0.8934 - val_loss: 0.2194 - val_acc: 0.9297 - val_precision_m: 0.9629 - val_recall_m: 0.9050\n",
      "Epoch 626/5000\n",
      " - 10s - loss: 0.2163 - acc: 0.9272 - precision_m: 0.9620 - recall_m: 0.8902 - val_loss: 0.2176 - val_acc: 0.9269 - val_precision_m: 0.9644 - val_recall_m: 0.8901\n",
      "Epoch 627/5000\n",
      " - 10s - loss: 0.2104 - acc: 0.9300 - precision_m: 0.9629 - recall_m: 0.8948 - val_loss: 0.2303 - val_acc: 0.9169 - val_precision_m: 0.9800 - val_recall_m: 0.8569\n",
      "Epoch 628/5000\n",
      " - 10s - loss: 0.2107 - acc: 0.9288 - precision_m: 0.9626 - recall_m: 0.8933 - val_loss: 0.2167 - val_acc: 0.9280 - val_precision_m: 0.9535 - val_recall_m: 0.9039\n",
      "Epoch 629/5000\n",
      " - 10s - loss: 0.2106 - acc: 0.9285 - precision_m: 0.9620 - recall_m: 0.8923 - val_loss: 0.2510 - val_acc: 0.9097 - val_precision_m: 0.8984 - val_recall_m: 0.9394\n",
      "Epoch 630/5000\n",
      " - 10s - loss: 0.2120 - acc: 0.9283 - precision_m: 0.9633 - recall_m: 0.8912 - val_loss: 0.2251 - val_acc: 0.9269 - val_precision_m: 0.9379 - val_recall_m: 0.9192\n",
      "Epoch 631/5000\n",
      " - 10s - loss: 0.2120 - acc: 0.9266 - precision_m: 0.9598 - recall_m: 0.8918 - val_loss: 0.2348 - val_acc: 0.9131 - val_precision_m: 0.9798 - val_recall_m: 0.8501\n",
      "Epoch 632/5000\n",
      " - 10s - loss: 0.2082 - acc: 0.9292 - precision_m: 0.9628 - recall_m: 0.8938 - val_loss: 0.2198 - val_acc: 0.9252 - val_precision_m: 0.9728 - val_recall_m: 0.8793\n",
      "Epoch 633/5000\n",
      " - 10s - loss: 0.2150 - acc: 0.9242 - precision_m: 0.9579 - recall_m: 0.8899 - val_loss: 0.2155 - val_acc: 0.9336 - val_precision_m: 0.9630 - val_recall_m: 0.9121\n",
      "Epoch 634/5000\n",
      " - 10s - loss: 0.2143 - acc: 0.9253 - precision_m: 0.9588 - recall_m: 0.8904 - val_loss: 0.2165 - val_acc: 0.9352 - val_precision_m: 0.9571 - val_recall_m: 0.9137\n",
      "Epoch 635/5000\n",
      " - 10s - loss: 0.2063 - acc: 0.9296 - precision_m: 0.9634 - recall_m: 0.8930 - val_loss: 0.2229 - val_acc: 0.9225 - val_precision_m: 0.9814 - val_recall_m: 0.8662\n",
      "Epoch 636/5000\n",
      " - 10s - loss: 0.2141 - acc: 0.9256 - precision_m: 0.9589 - recall_m: 0.8904 - val_loss: 0.2389 - val_acc: 0.9109 - val_precision_m: 0.9820 - val_recall_m: 0.8506\n",
      "Epoch 637/5000\n",
      " - 10s - loss: 0.2083 - acc: 0.9290 - precision_m: 0.9634 - recall_m: 0.8932 - val_loss: 0.2171 - val_acc: 0.9236 - val_precision_m: 0.9770 - val_recall_m: 0.8723\n",
      "Epoch 638/5000\n",
      " - 10s - loss: 0.2088 - acc: 0.9295 - precision_m: 0.9615 - recall_m: 0.8964 - val_loss: 0.2435 - val_acc: 0.9081 - val_precision_m: 0.9795 - val_recall_m: 0.8482\n",
      "Epoch 639/5000\n",
      " - 10s - loss: 0.2143 - acc: 0.9255 - precision_m: 0.9600 - recall_m: 0.8891 - val_loss: 0.2203 - val_acc: 0.9275 - val_precision_m: 0.9720 - val_recall_m: 0.8920\n",
      "Epoch 640/5000\n",
      " - 10s - loss: 0.2072 - acc: 0.9312 - precision_m: 0.9636 - recall_m: 0.8966 - val_loss: 0.2158 - val_acc: 0.9236 - val_precision_m: 0.9563 - val_recall_m: 0.8927\n",
      "Epoch 641/5000\n",
      " - 10s - loss: 0.2110 - acc: 0.9270 - precision_m: 0.9591 - recall_m: 0.8931 - val_loss: 0.2406 - val_acc: 0.9230 - val_precision_m: 0.9214 - val_recall_m: 0.9377\n",
      "Epoch 642/5000\n",
      " - 10s - loss: 0.2150 - acc: 0.9248 - precision_m: 0.9585 - recall_m: 0.8900 - val_loss: 0.2336 - val_acc: 0.9181 - val_precision_m: 0.9123 - val_recall_m: 0.9388\n",
      "Epoch 643/5000\n",
      " - 10s - loss: 0.2168 - acc: 0.9254 - precision_m: 0.9580 - recall_m: 0.8918 - val_loss: 0.2145 - val_acc: 0.9297 - val_precision_m: 0.9626 - val_recall_m: 0.8975\n",
      "Epoch 644/5000\n",
      " - 10s - loss: 0.2035 - acc: 0.9326 - precision_m: 0.9644 - recall_m: 0.8995 - val_loss: 0.2095 - val_acc: 0.9369 - val_precision_m: 0.9705 - val_recall_m: 0.9038\n",
      "Epoch 645/5000\n",
      " - 10s - loss: 0.2074 - acc: 0.9311 - precision_m: 0.9621 - recall_m: 0.8987 - val_loss: 0.2153 - val_acc: 0.9236 - val_precision_m: 0.9731 - val_recall_m: 0.8762\n",
      "Epoch 646/5000\n",
      " - 10s - loss: 0.2071 - acc: 0.9293 - precision_m: 0.9628 - recall_m: 0.8945 - val_loss: 0.2149 - val_acc: 0.9286 - val_precision_m: 0.9762 - val_recall_m: 0.8824\n",
      "Epoch 647/5000\n",
      " - 10s - loss: 0.2053 - acc: 0.9306 - precision_m: 0.9633 - recall_m: 0.8960 - val_loss: 0.2066 - val_acc: 0.9302 - val_precision_m: 0.9670 - val_recall_m: 0.8946\n",
      "Epoch 648/5000\n",
      " - 10s - loss: 0.2097 - acc: 0.9274 - precision_m: 0.9597 - recall_m: 0.8939 - val_loss: 0.2214 - val_acc: 0.9258 - val_precision_m: 0.9322 - val_recall_m: 0.9301\n",
      "Epoch 649/5000\n",
      " - 10s - loss: 0.2110 - acc: 0.9268 - precision_m: 0.9611 - recall_m: 0.8912 - val_loss: 0.2258 - val_acc: 0.9181 - val_precision_m: 0.9225 - val_recall_m: 0.9261\n",
      "Epoch 650/5000\n",
      " - 10s - loss: 0.2099 - acc: 0.9274 - precision_m: 0.9579 - recall_m: 0.8957 - val_loss: 0.2682 - val_acc: 0.8876 - val_precision_m: 0.9858 - val_recall_m: 0.7960\n",
      "Epoch 651/5000\n",
      " - 10s - loss: 0.2073 - acc: 0.9284 - precision_m: 0.9618 - recall_m: 0.8932 - val_loss: 0.2128 - val_acc: 0.9302 - val_precision_m: 0.9481 - val_recall_m: 0.9211\n",
      "Epoch 652/5000\n",
      " - 10s - loss: 0.2018 - acc: 0.9313 - precision_m: 0.9618 - recall_m: 0.8985 - val_loss: 0.2171 - val_acc: 0.9247 - val_precision_m: 0.9782 - val_recall_m: 0.8730\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 653/5000\n",
      " - 10s - loss: 0.2042 - acc: 0.9312 - precision_m: 0.9643 - recall_m: 0.8963 - val_loss: 0.2113 - val_acc: 0.9330 - val_precision_m: 0.9589 - val_recall_m: 0.9151\n",
      "Epoch 654/5000\n",
      " - 10s - loss: 0.2044 - acc: 0.9314 - precision_m: 0.9629 - recall_m: 0.8983 - val_loss: 0.2209 - val_acc: 0.9203 - val_precision_m: 0.9813 - val_recall_m: 0.8627\n",
      "Epoch 655/5000\n",
      " - 10s - loss: 0.2109 - acc: 0.9283 - precision_m: 0.9580 - recall_m: 0.8978 - val_loss: 0.2679 - val_acc: 0.8870 - val_precision_m: 0.9848 - val_recall_m: 0.7888\n",
      "Epoch 656/5000\n",
      " - 10s - loss: 0.2153 - acc: 0.9232 - precision_m: 0.9540 - recall_m: 0.8930 - val_loss: 0.2123 - val_acc: 0.9286 - val_precision_m: 0.9772 - val_recall_m: 0.8818\n",
      "Epoch 657/5000\n",
      " - 10s - loss: 0.2013 - acc: 0.9333 - precision_m: 0.9652 - recall_m: 0.9000 - val_loss: 0.2088 - val_acc: 0.9336 - val_precision_m: 0.9570 - val_recall_m: 0.9182\n",
      "Epoch 658/5000\n",
      " - 10s - loss: 0.2032 - acc: 0.9293 - precision_m: 0.9601 - recall_m: 0.8969 - val_loss: 0.2109 - val_acc: 0.9313 - val_precision_m: 0.9515 - val_recall_m: 0.9192\n",
      "Epoch 659/5000\n",
      " - 10s - loss: 0.2032 - acc: 0.9314 - precision_m: 0.9635 - recall_m: 0.8980 - val_loss: 0.2050 - val_acc: 0.9313 - val_precision_m: 0.9620 - val_recall_m: 0.9088\n",
      "Epoch 660/5000\n",
      " - 10s - loss: 0.2009 - acc: 0.9341 - precision_m: 0.9652 - recall_m: 0.9012 - val_loss: 0.2098 - val_acc: 0.9291 - val_precision_m: 0.9766 - val_recall_m: 0.8910\n",
      "Epoch 661/5000\n",
      " - 10s - loss: 0.1992 - acc: 0.9336 - precision_m: 0.9654 - recall_m: 0.9000 - val_loss: 0.2163 - val_acc: 0.9230 - val_precision_m: 0.9783 - val_recall_m: 0.8702\n",
      "Epoch 662/5000\n",
      " - 10s - loss: 0.2032 - acc: 0.9312 - precision_m: 0.9622 - recall_m: 0.8987 - val_loss: 0.2045 - val_acc: 0.9363 - val_precision_m: 0.9693 - val_recall_m: 0.9108\n",
      "Epoch 663/5000\n",
      " - 10s - loss: 0.2004 - acc: 0.9340 - precision_m: 0.9654 - recall_m: 0.9004 - val_loss: 0.2030 - val_acc: 0.9330 - val_precision_m: 0.9699 - val_recall_m: 0.9040\n",
      "Epoch 664/5000\n",
      " - 10s - loss: 0.2022 - acc: 0.9314 - precision_m: 0.9654 - recall_m: 0.8965 - val_loss: 0.2198 - val_acc: 0.9258 - val_precision_m: 0.9337 - val_recall_m: 0.9285\n",
      "Epoch 665/5000\n",
      " - 10s - loss: 0.2011 - acc: 0.9313 - precision_m: 0.9632 - recall_m: 0.8978 - val_loss: 0.2135 - val_acc: 0.9247 - val_precision_m: 0.9782 - val_recall_m: 0.8732\n",
      "Epoch 666/5000\n",
      " - 10s - loss: 0.1972 - acc: 0.9341 - precision_m: 0.9654 - recall_m: 0.9013 - val_loss: 0.2263 - val_acc: 0.9153 - val_precision_m: 0.9226 - val_recall_m: 0.9205\n",
      "Epoch 667/5000\n",
      " - 10s - loss: 0.1993 - acc: 0.9332 - precision_m: 0.9642 - recall_m: 0.9000 - val_loss: 0.2121 - val_acc: 0.9258 - val_precision_m: 0.9390 - val_recall_m: 0.9225\n",
      "Epoch 668/5000\n",
      " - 10s - loss: 0.2012 - acc: 0.9330 - precision_m: 0.9626 - recall_m: 0.9020 - val_loss: 0.2068 - val_acc: 0.9330 - val_precision_m: 0.9583 - val_recall_m: 0.9159\n",
      "Epoch 669/5000\n",
      " - 10s - loss: 0.2043 - acc: 0.9306 - precision_m: 0.9616 - recall_m: 0.8987 - val_loss: 0.2205 - val_acc: 0.9158 - val_precision_m: 0.9799 - val_recall_m: 0.8552\n",
      "Epoch 670/5000\n",
      " - 10s - loss: 0.2023 - acc: 0.9323 - precision_m: 0.9639 - recall_m: 0.8992 - val_loss: 0.2015 - val_acc: 0.9336 - val_precision_m: 0.9740 - val_recall_m: 0.9011\n",
      "Epoch 671/5000\n",
      " - 10s - loss: 0.1979 - acc: 0.9350 - precision_m: 0.9662 - recall_m: 0.9023 - val_loss: 0.2093 - val_acc: 0.9291 - val_precision_m: 0.9771 - val_recall_m: 0.8899\n",
      "Epoch 672/5000\n",
      " - 10s - loss: 0.2002 - acc: 0.9316 - precision_m: 0.9618 - recall_m: 0.8996 - val_loss: 0.2041 - val_acc: 0.9336 - val_precision_m: 0.9663 - val_recall_m: 0.9092\n",
      "Epoch 673/5000\n",
      " - 10s - loss: 0.1971 - acc: 0.9348 - precision_m: 0.9670 - recall_m: 0.9014 - val_loss: 0.2076 - val_acc: 0.9330 - val_precision_m: 0.9786 - val_recall_m: 0.8960\n",
      "Epoch 674/5000\n",
      " - 10s - loss: 0.1987 - acc: 0.9322 - precision_m: 0.9619 - recall_m: 0.9018 - val_loss: 0.2053 - val_acc: 0.9313 - val_precision_m: 0.9671 - val_recall_m: 0.9039\n",
      "Epoch 675/5000\n",
      " - 10s - loss: 0.1997 - acc: 0.9323 - precision_m: 0.9629 - recall_m: 0.8999 - val_loss: 0.2015 - val_acc: 0.9358 - val_precision_m: 0.9700 - val_recall_m: 0.9089\n",
      "Epoch 676/5000\n",
      " - 10s - loss: 0.2003 - acc: 0.9329 - precision_m: 0.9635 - recall_m: 0.9009 - val_loss: 0.2024 - val_acc: 0.9391 - val_precision_m: 0.9597 - val_recall_m: 0.9262\n",
      "Epoch 677/5000\n",
      " - 10s - loss: 0.1968 - acc: 0.9351 - precision_m: 0.9641 - recall_m: 0.9043 - val_loss: 0.2093 - val_acc: 0.9291 - val_precision_m: 0.9422 - val_recall_m: 0.9253\n",
      "Epoch 678/5000\n",
      " - 10s - loss: 0.1987 - acc: 0.9320 - precision_m: 0.9651 - recall_m: 0.8972 - val_loss: 0.2084 - val_acc: 0.9313 - val_precision_m: 0.9433 - val_recall_m: 0.9287\n",
      "Epoch 679/5000\n",
      " - 10s - loss: 0.1980 - acc: 0.9318 - precision_m: 0.9594 - recall_m: 0.9026 - val_loss: 0.2082 - val_acc: 0.9241 - val_precision_m: 0.9772 - val_recall_m: 0.8806\n",
      "Epoch 680/5000\n",
      " - 10s - loss: 0.2041 - acc: 0.9276 - precision_m: 0.9582 - recall_m: 0.8960 - val_loss: 0.2076 - val_acc: 0.9280 - val_precision_m: 0.9490 - val_recall_m: 0.9093\n",
      "Epoch 681/5000\n",
      " - 10s - loss: 0.1987 - acc: 0.9325 - precision_m: 0.9641 - recall_m: 0.8991 - val_loss: 0.2075 - val_acc: 0.9358 - val_precision_m: 0.9622 - val_recall_m: 0.9169\n",
      "Epoch 682/5000\n",
      " - 10s - loss: 0.1936 - acc: 0.9330 - precision_m: 0.9621 - recall_m: 0.9015 - val_loss: 0.2060 - val_acc: 0.9302 - val_precision_m: 0.9786 - val_recall_m: 0.8834\n",
      "Epoch 683/5000\n",
      " - 10s - loss: 0.1963 - acc: 0.9335 - precision_m: 0.9622 - recall_m: 0.9041 - val_loss: 0.2138 - val_acc: 0.9258 - val_precision_m: 0.9319 - val_recall_m: 0.9302\n",
      "Epoch 684/5000\n",
      " - 10s - loss: 0.1974 - acc: 0.9324 - precision_m: 0.9611 - recall_m: 0.9029 - val_loss: 0.2563 - val_acc: 0.8942 - val_precision_m: 0.9873 - val_recall_m: 0.8072\n",
      "Epoch 685/5000\n",
      " - 10s - loss: 0.2007 - acc: 0.9321 - precision_m: 0.9620 - recall_m: 0.9015 - val_loss: 0.2006 - val_acc: 0.9330 - val_precision_m: 0.9660 - val_recall_m: 0.9081\n",
      "Epoch 686/5000\n",
      " - 10s - loss: 0.1993 - acc: 0.9339 - precision_m: 0.9631 - recall_m: 0.9027 - val_loss: 0.2108 - val_acc: 0.9297 - val_precision_m: 0.9365 - val_recall_m: 0.9334\n",
      "Epoch 687/5000\n",
      " - 10s - loss: 0.2023 - acc: 0.9310 - precision_m: 0.9603 - recall_m: 0.9010 - val_loss: 0.2188 - val_acc: 0.9086 - val_precision_m: 0.9795 - val_recall_m: 0.8414\n",
      "Epoch 688/5000\n",
      " - 10s - loss: 0.1994 - acc: 0.9304 - precision_m: 0.9618 - recall_m: 0.8980 - val_loss: 0.1974 - val_acc: 0.9313 - val_precision_m: 0.9732 - val_recall_m: 0.8979\n",
      "Epoch 689/5000\n",
      " - 10s - loss: 0.1902 - acc: 0.9387 - precision_m: 0.9693 - recall_m: 0.9063 - val_loss: 0.2027 - val_acc: 0.9308 - val_precision_m: 0.9784 - val_recall_m: 0.8919\n",
      "Epoch 690/5000\n",
      " - 10s - loss: 0.1973 - acc: 0.9332 - precision_m: 0.9639 - recall_m: 0.9012 - val_loss: 0.2024 - val_acc: 0.9336 - val_precision_m: 0.9731 - val_recall_m: 0.9019\n",
      "Epoch 691/5000\n",
      " - 10s - loss: 0.2009 - acc: 0.9305 - precision_m: 0.9603 - recall_m: 0.9002 - val_loss: 0.1981 - val_acc: 0.9341 - val_precision_m: 0.9671 - val_recall_m: 0.9089\n",
      "Epoch 692/5000\n",
      " - 10s - loss: 0.1919 - acc: 0.9362 - precision_m: 0.9649 - recall_m: 0.9056 - val_loss: 0.1973 - val_acc: 0.9336 - val_precision_m: 0.9743 - val_recall_m: 0.9011\n",
      "Epoch 693/5000\n",
      " - 10s - loss: 0.1937 - acc: 0.9364 - precision_m: 0.9668 - recall_m: 0.9042 - val_loss: 0.2009 - val_acc: 0.9313 - val_precision_m: 0.9599 - val_recall_m: 0.9034\n",
      "Epoch 694/5000\n",
      " - 10s - loss: 0.1932 - acc: 0.9347 - precision_m: 0.9644 - recall_m: 0.9036 - val_loss: 0.2028 - val_acc: 0.9286 - val_precision_m: 0.9716 - val_recall_m: 0.8936\n",
      "Epoch 695/5000\n",
      " - 10s - loss: 0.1930 - acc: 0.9344 - precision_m: 0.9626 - recall_m: 0.9044 - val_loss: 0.1995 - val_acc: 0.9408 - val_precision_m: 0.9708 - val_recall_m: 0.9183\n",
      "Epoch 696/5000\n",
      " - 10s - loss: 0.1915 - acc: 0.9347 - precision_m: 0.9642 - recall_m: 0.9038 - val_loss: 0.2144 - val_acc: 0.9181 - val_precision_m: 0.9788 - val_recall_m: 0.8674\n",
      "Epoch 697/5000\n",
      " - 10s - loss: 0.1889 - acc: 0.9388 - precision_m: 0.9662 - recall_m: 0.9092 - val_loss: 0.1971 - val_acc: 0.9347 - val_precision_m: 0.9765 - val_recall_m: 0.9009\n",
      "Epoch 698/5000\n",
      " - 10s - loss: 0.1917 - acc: 0.9356 - precision_m: 0.9632 - recall_m: 0.9069 - val_loss: 0.1995 - val_acc: 0.9358 - val_precision_m: 0.9584 - val_recall_m: 0.9210\n",
      "Epoch 699/5000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " - 10s - loss: 0.1899 - acc: 0.9370 - precision_m: 0.9658 - recall_m: 0.9060 - val_loss: 0.1909 - val_acc: 0.9380 - val_precision_m: 0.9632 - val_recall_m: 0.9130\n",
      "Epoch 700/5000\n",
      " - 10s - loss: 0.1901 - acc: 0.9356 - precision_m: 0.9651 - recall_m: 0.9040 - val_loss: 0.2153 - val_acc: 0.9186 - val_precision_m: 0.9802 - val_recall_m: 0.8672\n",
      "Epoch 701/5000\n",
      " - 10s - loss: 0.1916 - acc: 0.9349 - precision_m: 0.9636 - recall_m: 0.9043 - val_loss: 0.1977 - val_acc: 0.9363 - val_precision_m: 0.9692 - val_recall_m: 0.9109\n",
      "Epoch 702/5000\n",
      " - 10s - loss: 0.1984 - acc: 0.9321 - precision_m: 0.9624 - recall_m: 0.9006 - val_loss: 0.2064 - val_acc: 0.9236 - val_precision_m: 0.9794 - val_recall_m: 0.8775\n",
      "Epoch 703/5000\n",
      " - 10s - loss: 0.2009 - acc: 0.9311 - precision_m: 0.9603 - recall_m: 0.9020 - val_loss: 0.2005 - val_acc: 0.9341 - val_precision_m: 0.9495 - val_recall_m: 0.9270\n",
      "Epoch 704/5000\n",
      " - 10s - loss: 0.1866 - acc: 0.9378 - precision_m: 0.9671 - recall_m: 0.9071 - val_loss: 0.1975 - val_acc: 0.9374 - val_precision_m: 0.9664 - val_recall_m: 0.9158\n",
      "Epoch 705/5000\n",
      " - 10s - loss: 0.1904 - acc: 0.9359 - precision_m: 0.9645 - recall_m: 0.9060 - val_loss: 0.1956 - val_acc: 0.9319 - val_precision_m: 0.9762 - val_recall_m: 0.8962\n",
      "Epoch 706/5000\n",
      " - 10s - loss: 0.1962 - acc: 0.9324 - precision_m: 0.9624 - recall_m: 0.9022 - val_loss: 0.1989 - val_acc: 0.9341 - val_precision_m: 0.9701 - val_recall_m: 0.9061\n",
      "Epoch 707/5000\n",
      " - 10s - loss: 0.1889 - acc: 0.9357 - precision_m: 0.9643 - recall_m: 0.9059 - val_loss: 0.1938 - val_acc: 0.9352 - val_precision_m: 0.9703 - val_recall_m: 0.9005\n",
      "Epoch 708/5000\n",
      " - 10s - loss: 0.1963 - acc: 0.9312 - precision_m: 0.9616 - recall_m: 0.9004 - val_loss: 0.2752 - val_acc: 0.8870 - val_precision_m: 0.9884 - val_recall_m: 0.7931\n",
      "Epoch 709/5000\n",
      " - 10s - loss: 0.1934 - acc: 0.9338 - precision_m: 0.9619 - recall_m: 0.9048 - val_loss: 0.2087 - val_acc: 0.9203 - val_precision_m: 0.9780 - val_recall_m: 0.8729\n",
      "Epoch 710/5000\n",
      " - 10s - loss: 0.1834 - acc: 0.9385 - precision_m: 0.9683 - recall_m: 0.9072 - val_loss: 0.1924 - val_acc: 0.9363 - val_precision_m: 0.9745 - val_recall_m: 0.9057\n",
      "Epoch 711/5000\n",
      " - 10s - loss: 0.1844 - acc: 0.9386 - precision_m: 0.9669 - recall_m: 0.9086 - val_loss: 0.2015 - val_acc: 0.9275 - val_precision_m: 0.9795 - val_recall_m: 0.8852\n",
      "Epoch 712/5000\n",
      " - 10s - loss: 0.1874 - acc: 0.9370 - precision_m: 0.9659 - recall_m: 0.9069 - val_loss: 0.1997 - val_acc: 0.9264 - val_precision_m: 0.9795 - val_recall_m: 0.8758\n",
      "Epoch 713/5000\n",
      " - 10s - loss: 0.1928 - acc: 0.9340 - precision_m: 0.9601 - recall_m: 0.9077 - val_loss: 0.1966 - val_acc: 0.9341 - val_precision_m: 0.9765 - val_recall_m: 0.9003\n",
      "Epoch 714/5000\n",
      " - 10s - loss: 0.1845 - acc: 0.9367 - precision_m: 0.9656 - recall_m: 0.9058 - val_loss: 0.1989 - val_acc: 0.9269 - val_precision_m: 0.9794 - val_recall_m: 0.8839\n",
      "Epoch 715/5000\n",
      " - 10s - loss: 0.1901 - acc: 0.9354 - precision_m: 0.9639 - recall_m: 0.9060 - val_loss: 0.1937 - val_acc: 0.9363 - val_precision_m: 0.9776 - val_recall_m: 0.9028\n",
      "Epoch 716/5000\n",
      " - 10s - loss: 0.1896 - acc: 0.9360 - precision_m: 0.9655 - recall_m: 0.9063 - val_loss: 0.1913 - val_acc: 0.9396 - val_precision_m: 0.9735 - val_recall_m: 0.9132\n",
      "Epoch 717/5000\n",
      " - 10s - loss: 0.1888 - acc: 0.9367 - precision_m: 0.9654 - recall_m: 0.9064 - val_loss: 0.2067 - val_acc: 0.9214 - val_precision_m: 0.9823 - val_recall_m: 0.8629\n",
      "Epoch 718/5000\n",
      " - 10s - loss: 0.1907 - acc: 0.9336 - precision_m: 0.9625 - recall_m: 0.9038 - val_loss: 0.1871 - val_acc: 0.9402 - val_precision_m: 0.9704 - val_recall_m: 0.9172\n",
      "Epoch 719/5000\n",
      " - 10s - loss: 0.1850 - acc: 0.9384 - precision_m: 0.9655 - recall_m: 0.9100 - val_loss: 0.1892 - val_acc: 0.9391 - val_precision_m: 0.9650 - val_recall_m: 0.9200\n",
      "Epoch 720/5000\n",
      " - 10s - loss: 0.1861 - acc: 0.9376 - precision_m: 0.9659 - recall_m: 0.9082 - val_loss: 0.1933 - val_acc: 0.9352 - val_precision_m: 0.9743 - val_recall_m: 0.9040\n",
      "Epoch 721/5000\n",
      " - 10s - loss: 0.1841 - acc: 0.9387 - precision_m: 0.9677 - recall_m: 0.9080 - val_loss: 0.2202 - val_acc: 0.9125 - val_precision_m: 0.9835 - val_recall_m: 0.8534\n",
      "Epoch 722/5000\n",
      " - 10s - loss: 0.1852 - acc: 0.9385 - precision_m: 0.9657 - recall_m: 0.9111 - val_loss: 0.2329 - val_acc: 0.9070 - val_precision_m: 0.9818 - val_recall_m: 0.8439\n",
      "Epoch 723/5000\n",
      " - 10s - loss: 0.1829 - acc: 0.9394 - precision_m: 0.9679 - recall_m: 0.9091 - val_loss: 0.2013 - val_acc: 0.9230 - val_precision_m: 0.9814 - val_recall_m: 0.8671\n",
      "Epoch 724/5000\n",
      " - 10s - loss: 0.1824 - acc: 0.9398 - precision_m: 0.9668 - recall_m: 0.9111 - val_loss: 0.1900 - val_acc: 0.9352 - val_precision_m: 0.9505 - val_recall_m: 0.9282\n",
      "Epoch 725/5000\n",
      " - 10s - loss: 0.1849 - acc: 0.9368 - precision_m: 0.9643 - recall_m: 0.9081 - val_loss: 0.2051 - val_acc: 0.9214 - val_precision_m: 0.9778 - val_recall_m: 0.8743\n",
      "Epoch 726/5000\n",
      " - 10s - loss: 0.1829 - acc: 0.9386 - precision_m: 0.9662 - recall_m: 0.9093 - val_loss: 0.2000 - val_acc: 0.9336 - val_precision_m: 0.9572 - val_recall_m: 0.9182\n",
      "Epoch 727/5000\n",
      " - 10s - loss: 0.1844 - acc: 0.9393 - precision_m: 0.9666 - recall_m: 0.9108 - val_loss: 0.1954 - val_acc: 0.9291 - val_precision_m: 0.9730 - val_recall_m: 0.8935\n",
      "Epoch 728/5000\n",
      " - 10s - loss: 0.1795 - acc: 0.9397 - precision_m: 0.9680 - recall_m: 0.9094 - val_loss: 0.1919 - val_acc: 0.9352 - val_precision_m: 0.9543 - val_recall_m: 0.9244\n",
      "Epoch 729/5000\n",
      " - 10s - loss: 0.1857 - acc: 0.9380 - precision_m: 0.9638 - recall_m: 0.9112 - val_loss: 0.1932 - val_acc: 0.9385 - val_precision_m: 0.9567 - val_recall_m: 0.9285\n",
      "Epoch 730/5000\n",
      " - 10s - loss: 0.1908 - acc: 0.9346 - precision_m: 0.9637 - recall_m: 0.9057 - val_loss: 0.2326 - val_acc: 0.9186 - val_precision_m: 0.9003 - val_recall_m: 0.9553\n",
      "Epoch 731/5000\n",
      " - 10s - loss: 0.2061 - acc: 0.9255 - precision_m: 0.9538 - recall_m: 0.8993 - val_loss: 0.1944 - val_acc: 0.9258 - val_precision_m: 0.9803 - val_recall_m: 0.8805\n",
      "Epoch 732/5000\n",
      " - 10s - loss: 0.1791 - acc: 0.9415 - precision_m: 0.9684 - recall_m: 0.9128 - val_loss: 0.1914 - val_acc: 0.9380 - val_precision_m: 0.9490 - val_recall_m: 0.9355\n",
      "Epoch 733/5000\n",
      " - 10s - loss: 0.1775 - acc: 0.9413 - precision_m: 0.9687 - recall_m: 0.9119 - val_loss: 0.1891 - val_acc: 0.9363 - val_precision_m: 0.9725 - val_recall_m: 0.9078\n",
      "Epoch 734/5000\n",
      " - 10s - loss: 0.1828 - acc: 0.9375 - precision_m: 0.9642 - recall_m: 0.9093 - val_loss: 0.1901 - val_acc: 0.9324 - val_precision_m: 0.9786 - val_recall_m: 0.8946\n",
      "Epoch 735/5000\n",
      " - 10s - loss: 0.1769 - acc: 0.9425 - precision_m: 0.9693 - recall_m: 0.9143 - val_loss: 0.1923 - val_acc: 0.9385 - val_precision_m: 0.9538 - val_recall_m: 0.9313\n",
      "Epoch 736/5000\n",
      " - 10s - loss: 0.1797 - acc: 0.9389 - precision_m: 0.9656 - recall_m: 0.9107 - val_loss: 0.1882 - val_acc: 0.9369 - val_precision_m: 0.9673 - val_recall_m: 0.9144\n",
      "Epoch 737/5000\n",
      " - 10s - loss: 0.1798 - acc: 0.9403 - precision_m: 0.9685 - recall_m: 0.9104 - val_loss: 0.1899 - val_acc: 0.9391 - val_precision_m: 0.9577 - val_recall_m: 0.9282\n",
      "Epoch 738/5000\n",
      " - 10s - loss: 0.1798 - acc: 0.9408 - precision_m: 0.9679 - recall_m: 0.9126 - val_loss: 0.1888 - val_acc: 0.9385 - val_precision_m: 0.9529 - val_recall_m: 0.9326\n",
      "Epoch 739/5000\n",
      " - 10s - loss: 0.1804 - acc: 0.9407 - precision_m: 0.9681 - recall_m: 0.9118 - val_loss: 0.1913 - val_acc: 0.9336 - val_precision_m: 0.9767 - val_recall_m: 0.8987\n",
      "Epoch 740/5000\n",
      " - 10s - loss: 0.1788 - acc: 0.9407 - precision_m: 0.9682 - recall_m: 0.9115 - val_loss: 0.1898 - val_acc: 0.9330 - val_precision_m: 0.9752 - val_recall_m: 0.8917\n",
      "Epoch 741/5000\n",
      " - 10s - loss: 0.1754 - acc: 0.9422 - precision_m: 0.9697 - recall_m: 0.9128 - val_loss: 0.1921 - val_acc: 0.9308 - val_precision_m: 0.9796 - val_recall_m: 0.8833\n",
      "Epoch 742/5000\n",
      " - 10s - loss: 0.1841 - acc: 0.9380 - precision_m: 0.9656 - recall_m: 0.9095 - val_loss: 0.1943 - val_acc: 0.9280 - val_precision_m: 0.9771 - val_recall_m: 0.8882\n",
      "Epoch 743/5000\n",
      " - 10s - loss: 0.1824 - acc: 0.9378 - precision_m: 0.9647 - recall_m: 0.9102 - val_loss: 0.2034 - val_acc: 0.9302 - val_precision_m: 0.9312 - val_recall_m: 0.9405\n",
      "Epoch 744/5000\n",
      " - 10s - loss: 0.1807 - acc: 0.9378 - precision_m: 0.9651 - recall_m: 0.9090 - val_loss: 0.1941 - val_acc: 0.9258 - val_precision_m: 0.9784 - val_recall_m: 0.8829\n",
      "Epoch 745/5000\n",
      " - 10s - loss: 0.1800 - acc: 0.9410 - precision_m: 0.9680 - recall_m: 0.9131 - val_loss: 0.1954 - val_acc: 0.9302 - val_precision_m: 0.9796 - val_recall_m: 0.8897\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 746/5000\n",
      " - 10s - loss: 0.1873 - acc: 0.9357 - precision_m: 0.9640 - recall_m: 0.9073 - val_loss: 0.1905 - val_acc: 0.9347 - val_precision_m: 0.9722 - val_recall_m: 0.9046\n",
      "Epoch 747/5000\n",
      " - 10s - loss: 0.1772 - acc: 0.9404 - precision_m: 0.9684 - recall_m: 0.9109 - val_loss: 0.1789 - val_acc: 0.9391 - val_precision_m: 0.9704 - val_recall_m: 0.9151\n",
      "Epoch 748/5000\n",
      " - 10s - loss: 0.1783 - acc: 0.9393 - precision_m: 0.9661 - recall_m: 0.9112 - val_loss: 0.1812 - val_acc: 0.9408 - val_precision_m: 0.9716 - val_recall_m: 0.9170\n",
      "Epoch 749/5000\n",
      " - 10s - loss: 0.1753 - acc: 0.9422 - precision_m: 0.9687 - recall_m: 0.9142 - val_loss: 0.1914 - val_acc: 0.9330 - val_precision_m: 0.9818 - val_recall_m: 0.8929\n",
      "Epoch 750/5000\n",
      " - 10s - loss: 0.1776 - acc: 0.9417 - precision_m: 0.9683 - recall_m: 0.9136 - val_loss: 0.1864 - val_acc: 0.9352 - val_precision_m: 0.9797 - val_recall_m: 0.8987\n",
      "Epoch 751/5000\n",
      " - 10s - loss: 0.1774 - acc: 0.9413 - precision_m: 0.9673 - recall_m: 0.9139 - val_loss: 0.1857 - val_acc: 0.9430 - val_precision_m: 0.9800 - val_recall_m: 0.9133\n",
      "Epoch 752/5000\n",
      " - 10s - loss: 0.1799 - acc: 0.9379 - precision_m: 0.9651 - recall_m: 0.9102 - val_loss: 0.1831 - val_acc: 0.9396 - val_precision_m: 0.9696 - val_recall_m: 0.9171\n",
      "Epoch 753/5000\n",
      " - 10s - loss: 0.1741 - acc: 0.9414 - precision_m: 0.9694 - recall_m: 0.9124 - val_loss: 0.1810 - val_acc: 0.9380 - val_precision_m: 0.9777 - val_recall_m: 0.9056\n",
      "Epoch 754/5000\n",
      " - 10s - loss: 0.1762 - acc: 0.9402 - precision_m: 0.9664 - recall_m: 0.9127 - val_loss: 0.2050 - val_acc: 0.9192 - val_precision_m: 0.9791 - val_recall_m: 0.8700\n",
      "Epoch 755/5000\n",
      " - 10s - loss: 0.1747 - acc: 0.9417 - precision_m: 0.9686 - recall_m: 0.9136 - val_loss: 0.1838 - val_acc: 0.9446 - val_precision_m: 0.9678 - val_recall_m: 0.9280\n",
      "Epoch 756/5000\n",
      " - 10s - loss: 0.1833 - acc: 0.9365 - precision_m: 0.9641 - recall_m: 0.9085 - val_loss: 0.1931 - val_acc: 0.9341 - val_precision_m: 0.9394 - val_recall_m: 0.9388\n",
      "Epoch 757/5000\n",
      " - 10s - loss: 0.1741 - acc: 0.9419 - precision_m: 0.9675 - recall_m: 0.9152 - val_loss: 0.1800 - val_acc: 0.9430 - val_precision_m: 0.9718 - val_recall_m: 0.9211\n",
      "Epoch 758/5000\n",
      " - 10s - loss: 0.1762 - acc: 0.9417 - precision_m: 0.9676 - recall_m: 0.9149 - val_loss: 0.1801 - val_acc: 0.9408 - val_precision_m: 0.9717 - val_recall_m: 0.9172\n",
      "Epoch 759/5000\n",
      " - 10s - loss: 0.1735 - acc: 0.9426 - precision_m: 0.9688 - recall_m: 0.9152 - val_loss: 0.2069 - val_acc: 0.9169 - val_precision_m: 0.9845 - val_recall_m: 0.8604\n",
      "Epoch 760/5000\n",
      " - 10s - loss: 0.1747 - acc: 0.9407 - precision_m: 0.9674 - recall_m: 0.9125 - val_loss: 0.1851 - val_acc: 0.9369 - val_precision_m: 0.9786 - val_recall_m: 0.9033\n",
      "Epoch 761/5000\n",
      " - 10s - loss: 0.1775 - acc: 0.9406 - precision_m: 0.9675 - recall_m: 0.9127 - val_loss: 0.1830 - val_acc: 0.9374 - val_precision_m: 0.9775 - val_recall_m: 0.9050\n",
      "Epoch 762/5000\n",
      " - 10s - loss: 0.1751 - acc: 0.9411 - precision_m: 0.9669 - recall_m: 0.9142 - val_loss: 0.2249 - val_acc: 0.9031 - val_precision_m: 0.9830 - val_recall_m: 0.8356\n",
      "Epoch 763/5000\n",
      " - 10s - loss: 0.1784 - acc: 0.9388 - precision_m: 0.9649 - recall_m: 0.9106 - val_loss: 0.1789 - val_acc: 0.9424 - val_precision_m: 0.9749 - val_recall_m: 0.9170\n",
      "Epoch 764/5000\n",
      " - 10s - loss: 0.1743 - acc: 0.9424 - precision_m: 0.9684 - recall_m: 0.9155 - val_loss: 0.1891 - val_acc: 0.9369 - val_precision_m: 0.9777 - val_recall_m: 0.9039\n",
      "Epoch 765/5000\n",
      " - 10s - loss: 0.1744 - acc: 0.9410 - precision_m: 0.9674 - recall_m: 0.9147 - val_loss: 0.1904 - val_acc: 0.9396 - val_precision_m: 0.9467 - val_recall_m: 0.9415\n",
      "Epoch 766/5000\n",
      " - 10s - loss: 0.1758 - acc: 0.9409 - precision_m: 0.9674 - recall_m: 0.9132 - val_loss: 0.1824 - val_acc: 0.9435 - val_precision_m: 0.9669 - val_recall_m: 0.9273\n",
      "Epoch 767/5000\n",
      " - 10s - loss: 0.1765 - acc: 0.9401 - precision_m: 0.9673 - recall_m: 0.9114 - val_loss: 0.1897 - val_acc: 0.9358 - val_precision_m: 0.9808 - val_recall_m: 0.8919\n",
      "Epoch 768/5000\n",
      " - 10s - loss: 0.1765 - acc: 0.9396 - precision_m: 0.9661 - recall_m: 0.9118 - val_loss: 0.1877 - val_acc: 0.9319 - val_precision_m: 0.9637 - val_recall_m: 0.9079\n",
      "Epoch 769/5000\n",
      " - 10s - loss: 0.1735 - acc: 0.9420 - precision_m: 0.9667 - recall_m: 0.9154 - val_loss: 0.1784 - val_acc: 0.9430 - val_precision_m: 0.9759 - val_recall_m: 0.9170\n",
      "Epoch 770/5000\n",
      " - 10s - loss: 0.1806 - acc: 0.9377 - precision_m: 0.9646 - recall_m: 0.9100 - val_loss: 0.1825 - val_acc: 0.9380 - val_precision_m: 0.9767 - val_recall_m: 0.9071\n",
      "Epoch 771/5000\n",
      " - 10s - loss: 0.1780 - acc: 0.9388 - precision_m: 0.9645 - recall_m: 0.9122 - val_loss: 0.1904 - val_acc: 0.9236 - val_precision_m: 0.9771 - val_recall_m: 0.8791\n",
      "Epoch 772/5000\n",
      " - 10s - loss: 0.1679 - acc: 0.9447 - precision_m: 0.9708 - recall_m: 0.9172 - val_loss: 0.1768 - val_acc: 0.9419 - val_precision_m: 0.9598 - val_recall_m: 0.9315\n",
      "Epoch 773/5000\n",
      " - 10s - loss: 0.1702 - acc: 0.9437 - precision_m: 0.9712 - recall_m: 0.9149 - val_loss: 0.1832 - val_acc: 0.9341 - val_precision_m: 0.9788 - val_recall_m: 0.8979\n",
      "Epoch 774/5000\n",
      " - 10s - loss: 0.1712 - acc: 0.9418 - precision_m: 0.9671 - recall_m: 0.9151 - val_loss: 0.1842 - val_acc: 0.9385 - val_precision_m: 0.9569 - val_recall_m: 0.9282\n",
      "Epoch 775/5000\n",
      " - 10s - loss: 0.1794 - acc: 0.9408 - precision_m: 0.9662 - recall_m: 0.9152 - val_loss: 0.1796 - val_acc: 0.9430 - val_precision_m: 0.9607 - val_recall_m: 0.9323\n",
      "Epoch 776/5000\n",
      " - 10s - loss: 0.1678 - acc: 0.9431 - precision_m: 0.9681 - recall_m: 0.9170 - val_loss: 0.1924 - val_acc: 0.9380 - val_precision_m: 0.9399 - val_recall_m: 0.9455\n",
      "Epoch 777/5000\n",
      " - 10s - loss: 0.1684 - acc: 0.9431 - precision_m: 0.9695 - recall_m: 0.9162 - val_loss: 0.1776 - val_acc: 0.9413 - val_precision_m: 0.9618 - val_recall_m: 0.9282\n",
      "Epoch 778/5000\n",
      " - 10s - loss: 0.1698 - acc: 0.9432 - precision_m: 0.9681 - recall_m: 0.9170 - val_loss: 0.2211 - val_acc: 0.9064 - val_precision_m: 0.9865 - val_recall_m: 0.8393\n",
      "Epoch 779/5000\n",
      " - 10s - loss: 0.1736 - acc: 0.9426 - precision_m: 0.9683 - recall_m: 0.9162 - val_loss: 0.1876 - val_acc: 0.9363 - val_precision_m: 0.9355 - val_recall_m: 0.9468\n",
      "Epoch 780/5000\n",
      " - 10s - loss: 0.1690 - acc: 0.9434 - precision_m: 0.9695 - recall_m: 0.9165 - val_loss: 0.1821 - val_acc: 0.9358 - val_precision_m: 0.9799 - val_recall_m: 0.9002\n",
      "Epoch 781/5000\n",
      " - 10s - loss: 0.1665 - acc: 0.9455 - precision_m: 0.9700 - recall_m: 0.9200 - val_loss: 0.1777 - val_acc: 0.9435 - val_precision_m: 0.9669 - val_recall_m: 0.9272\n",
      "Epoch 782/5000\n",
      " - 10s - loss: 0.1724 - acc: 0.9412 - precision_m: 0.9665 - recall_m: 0.9149 - val_loss: 0.1740 - val_acc: 0.9413 - val_precision_m: 0.9767 - val_recall_m: 0.9126\n",
      "Epoch 783/5000\n",
      " - 10s - loss: 0.1710 - acc: 0.9432 - precision_m: 0.9671 - recall_m: 0.9173 - val_loss: 0.1839 - val_acc: 0.9330 - val_precision_m: 0.9796 - val_recall_m: 0.8952\n",
      "Epoch 784/5000\n",
      " - 10s - loss: 0.1701 - acc: 0.9423 - precision_m: 0.9669 - recall_m: 0.9160 - val_loss: 0.1861 - val_acc: 0.9291 - val_precision_m: 0.9805 - val_recall_m: 0.8864\n",
      "Epoch 785/5000\n",
      " - 10s - loss: 0.1696 - acc: 0.9433 - precision_m: 0.9693 - recall_m: 0.9162 - val_loss: 0.1729 - val_acc: 0.9419 - val_precision_m: 0.9768 - val_recall_m: 0.9142\n",
      "Epoch 786/5000\n",
      " - 10s - loss: 0.1655 - acc: 0.9456 - precision_m: 0.9714 - recall_m: 0.9181 - val_loss: 0.1832 - val_acc: 0.9369 - val_precision_m: 0.9371 - val_recall_m: 0.9466\n",
      "Epoch 787/5000\n",
      " - 10s - loss: 0.1654 - acc: 0.9449 - precision_m: 0.9701 - recall_m: 0.9187 - val_loss: 0.1746 - val_acc: 0.9396 - val_precision_m: 0.9605 - val_recall_m: 0.9263\n",
      "Epoch 788/5000\n",
      " - 10s - loss: 0.1675 - acc: 0.9453 - precision_m: 0.9690 - recall_m: 0.9202 - val_loss: 0.1893 - val_acc: 0.9385 - val_precision_m: 0.9530 - val_recall_m: 0.9325\n",
      "Epoch 789/5000\n",
      " - 10s - loss: 0.1794 - acc: 0.9384 - precision_m: 0.9653 - recall_m: 0.9108 - val_loss: 0.1808 - val_acc: 0.9446 - val_precision_m: 0.9562 - val_recall_m: 0.9403\n",
      "Epoch 790/5000\n",
      " - 10s - loss: 0.1681 - acc: 0.9426 - precision_m: 0.9681 - recall_m: 0.9159 - val_loss: 0.1738 - val_acc: 0.9413 - val_precision_m: 0.9685 - val_recall_m: 0.9212\n",
      "Epoch 791/5000\n",
      " - 10s - loss: 0.1734 - acc: 0.9412 - precision_m: 0.9659 - recall_m: 0.9162 - val_loss: 0.1930 - val_acc: 0.9280 - val_precision_m: 0.9814 - val_recall_m: 0.8838\n",
      "Epoch 792/5000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " - 10s - loss: 0.1646 - acc: 0.9453 - precision_m: 0.9702 - recall_m: 0.9187 - val_loss: 0.1900 - val_acc: 0.9269 - val_precision_m: 0.9826 - val_recall_m: 0.8808\n",
      "Epoch 793/5000\n",
      " - 10s - loss: 0.1725 - acc: 0.9410 - precision_m: 0.9674 - recall_m: 0.9134 - val_loss: 0.1744 - val_acc: 0.9452 - val_precision_m: 0.9642 - val_recall_m: 0.9334\n",
      "Epoch 794/5000\n",
      " - 10s - loss: 0.1680 - acc: 0.9442 - precision_m: 0.9682 - recall_m: 0.9190 - val_loss: 0.1715 - val_acc: 0.9396 - val_precision_m: 0.9757 - val_recall_m: 0.9110\n",
      "Epoch 795/5000\n",
      " - 10s - loss: 0.1678 - acc: 0.9433 - precision_m: 0.9676 - recall_m: 0.9177 - val_loss: 0.1786 - val_acc: 0.9452 - val_precision_m: 0.9550 - val_recall_m: 0.9434\n",
      "Epoch 796/5000\n",
      " - 10s - loss: 0.1617 - acc: 0.9458 - precision_m: 0.9698 - recall_m: 0.9207 - val_loss: 0.1739 - val_acc: 0.9396 - val_precision_m: 0.9790 - val_recall_m: 0.9079\n",
      "Epoch 797/5000\n",
      " - 10s - loss: 0.1643 - acc: 0.9450 - precision_m: 0.9704 - recall_m: 0.9184 - val_loss: 0.1790 - val_acc: 0.9291 - val_precision_m: 0.9773 - val_recall_m: 0.8897\n",
      "Epoch 798/5000\n",
      " - 10s - loss: 0.1636 - acc: 0.9461 - precision_m: 0.9716 - recall_m: 0.9191 - val_loss: 0.1856 - val_acc: 0.9297 - val_precision_m: 0.9763 - val_recall_m: 0.8916\n",
      "Epoch 799/5000\n",
      " - 10s - loss: 0.1640 - acc: 0.9453 - precision_m: 0.9700 - recall_m: 0.9190 - val_loss: 0.1717 - val_acc: 0.9452 - val_precision_m: 0.9689 - val_recall_m: 0.9284\n",
      "Epoch 800/5000\n",
      " - 10s - loss: 0.1632 - acc: 0.9463 - precision_m: 0.9709 - recall_m: 0.9208 - val_loss: 0.1806 - val_acc: 0.9413 - val_precision_m: 0.9521 - val_recall_m: 0.9384\n",
      "Epoch 801/5000\n",
      " - 10s - loss: 0.1660 - acc: 0.9445 - precision_m: 0.9694 - recall_m: 0.9187 - val_loss: 0.1681 - val_acc: 0.9424 - val_precision_m: 0.9726 - val_recall_m: 0.9192\n",
      "Epoch 802/5000\n",
      " - 10s - loss: 0.1646 - acc: 0.9443 - precision_m: 0.9687 - recall_m: 0.9185 - val_loss: 0.1800 - val_acc: 0.9391 - val_precision_m: 0.9516 - val_recall_m: 0.9345\n",
      "Epoch 803/5000\n",
      " - 10s - loss: 0.1632 - acc: 0.9452 - precision_m: 0.9708 - recall_m: 0.9184 - val_loss: 0.1721 - val_acc: 0.9396 - val_precision_m: 0.9623 - val_recall_m: 0.9245\n",
      "Epoch 804/5000\n",
      " - 10s - loss: 0.1654 - acc: 0.9444 - precision_m: 0.9710 - recall_m: 0.9169 - val_loss: 0.1930 - val_acc: 0.9352 - val_precision_m: 0.9315 - val_recall_m: 0.9499\n",
      "Epoch 805/5000\n",
      " - 10s - loss: 0.1758 - acc: 0.9391 - precision_m: 0.9627 - recall_m: 0.9165 - val_loss: 0.1739 - val_acc: 0.9446 - val_precision_m: 0.9739 - val_recall_m: 0.9223\n",
      "Epoch 806/5000\n",
      " - 10s - loss: 0.1675 - acc: 0.9428 - precision_m: 0.9671 - recall_m: 0.9179 - val_loss: 0.2058 - val_acc: 0.9380 - val_precision_m: 0.9256 - val_recall_m: 0.9631\n",
      "Epoch 807/5000\n",
      " - 10s - loss: 0.1650 - acc: 0.9443 - precision_m: 0.9679 - recall_m: 0.9195 - val_loss: 0.1697 - val_acc: 0.9463 - val_precision_m: 0.9782 - val_recall_m: 0.9214\n",
      "Epoch 808/5000\n",
      " - 10s - loss: 0.1670 - acc: 0.9428 - precision_m: 0.9690 - recall_m: 0.9156 - val_loss: 0.1696 - val_acc: 0.9457 - val_precision_m: 0.9657 - val_recall_m: 0.9321\n",
      "Epoch 809/5000\n",
      " - 10s - loss: 0.1648 - acc: 0.9444 - precision_m: 0.9696 - recall_m: 0.9182 - val_loss: 0.1780 - val_acc: 0.9402 - val_precision_m: 0.9800 - val_recall_m: 0.9081\n",
      "Epoch 810/5000\n",
      " - 10s - loss: 0.1645 - acc: 0.9463 - precision_m: 0.9718 - recall_m: 0.9199 - val_loss: 0.1807 - val_acc: 0.9347 - val_precision_m: 0.9832 - val_recall_m: 0.8946\n",
      "Epoch 811/5000\n",
      " - 10s - loss: 0.1667 - acc: 0.9428 - precision_m: 0.9674 - recall_m: 0.9170 - val_loss: 0.1702 - val_acc: 0.9391 - val_precision_m: 0.9745 - val_recall_m: 0.9111\n",
      "Epoch 812/5000\n",
      " - 10s - loss: 0.1613 - acc: 0.9480 - precision_m: 0.9718 - recall_m: 0.9228 - val_loss: 0.1748 - val_acc: 0.9352 - val_precision_m: 0.9807 - val_recall_m: 0.8974\n",
      "Epoch 813/5000\n",
      " - 10s - loss: 0.1648 - acc: 0.9449 - precision_m: 0.9674 - recall_m: 0.9216 - val_loss: 0.1646 - val_acc: 0.9441 - val_precision_m: 0.9668 - val_recall_m: 0.9285\n",
      "Epoch 814/5000\n",
      " - 10s - loss: 0.1574 - acc: 0.9500 - precision_m: 0.9730 - recall_m: 0.9262 - val_loss: 0.1654 - val_acc: 0.9513 - val_precision_m: 0.9703 - val_recall_m: 0.9391\n",
      "Epoch 815/5000\n",
      " - 10s - loss: 0.1635 - acc: 0.9451 - precision_m: 0.9697 - recall_m: 0.9193 - val_loss: 0.1846 - val_acc: 0.9435 - val_precision_m: 0.9465 - val_recall_m: 0.9499\n",
      "Epoch 816/5000\n",
      " - 10s - loss: 0.1638 - acc: 0.9440 - precision_m: 0.9680 - recall_m: 0.9190 - val_loss: 0.1676 - val_acc: 0.9385 - val_precision_m: 0.9766 - val_recall_m: 0.9078\n",
      "Epoch 817/5000\n",
      " - 10s - loss: 0.1634 - acc: 0.9444 - precision_m: 0.9674 - recall_m: 0.9209 - val_loss: 0.1696 - val_acc: 0.9441 - val_precision_m: 0.9758 - val_recall_m: 0.9187\n",
      "Epoch 818/5000\n",
      " - 10s - loss: 0.1613 - acc: 0.9455 - precision_m: 0.9700 - recall_m: 0.9201 - val_loss: 0.1725 - val_acc: 0.9363 - val_precision_m: 0.9809 - val_recall_m: 0.9000\n",
      "Epoch 819/5000\n",
      " - 10s - loss: 0.1626 - acc: 0.9465 - precision_m: 0.9697 - recall_m: 0.9224 - val_loss: 0.1734 - val_acc: 0.9408 - val_precision_m: 0.9675 - val_recall_m: 0.9215\n",
      "Epoch 820/5000\n",
      " - 10s - loss: 0.1711 - acc: 0.9405 - precision_m: 0.9653 - recall_m: 0.9146 - val_loss: 0.1665 - val_acc: 0.9463 - val_precision_m: 0.9627 - val_recall_m: 0.9362\n",
      "Epoch 821/5000\n",
      " - 10s - loss: 0.1616 - acc: 0.9445 - precision_m: 0.9677 - recall_m: 0.9206 - val_loss: 0.1811 - val_acc: 0.9413 - val_precision_m: 0.9427 - val_recall_m: 0.9491\n",
      "Epoch 822/5000\n",
      " - 10s - loss: 0.1614 - acc: 0.9481 - precision_m: 0.9723 - recall_m: 0.9227 - val_loss: 0.1708 - val_acc: 0.9374 - val_precision_m: 0.9733 - val_recall_m: 0.9096\n",
      "Epoch 823/5000\n",
      " - 10s - loss: 0.1594 - acc: 0.9471 - precision_m: 0.9710 - recall_m: 0.9217 - val_loss: 0.1671 - val_acc: 0.9402 - val_precision_m: 0.9757 - val_recall_m: 0.9122\n",
      "Epoch 824/5000\n",
      " - 10s - loss: 0.1639 - acc: 0.9461 - precision_m: 0.9694 - recall_m: 0.9222 - val_loss: 0.1683 - val_acc: 0.9413 - val_precision_m: 0.9606 - val_recall_m: 0.9294\n",
      "Epoch 825/5000\n",
      " - 10s - loss: 0.1625 - acc: 0.9457 - precision_m: 0.9700 - recall_m: 0.9205 - val_loss: 0.1672 - val_acc: 0.9385 - val_precision_m: 0.9722 - val_recall_m: 0.9119\n",
      "Epoch 826/5000\n",
      " - 10s - loss: 0.1565 - acc: 0.9477 - precision_m: 0.9719 - recall_m: 0.9218 - val_loss: 0.1635 - val_acc: 0.9463 - val_precision_m: 0.9720 - val_recall_m: 0.9273\n",
      "Epoch 827/5000\n",
      " - 10s - loss: 0.1583 - acc: 0.9471 - precision_m: 0.9706 - recall_m: 0.9227 - val_loss: 0.1819 - val_acc: 0.9280 - val_precision_m: 0.9828 - val_recall_m: 0.8828\n",
      "Epoch 828/5000\n",
      " - 10s - loss: 0.1584 - acc: 0.9463 - precision_m: 0.9696 - recall_m: 0.9214 - val_loss: 0.1778 - val_acc: 0.9308 - val_precision_m: 0.9795 - val_recall_m: 0.8909\n",
      "Epoch 829/5000\n",
      " - 10s - loss: 0.1628 - acc: 0.9447 - precision_m: 0.9682 - recall_m: 0.9208 - val_loss: 0.1677 - val_acc: 0.9474 - val_precision_m: 0.9586 - val_recall_m: 0.9435\n",
      "Epoch 830/5000\n",
      " - 10s - loss: 0.1576 - acc: 0.9468 - precision_m: 0.9697 - recall_m: 0.9221 - val_loss: 0.1631 - val_acc: 0.9419 - val_precision_m: 0.9790 - val_recall_m: 0.9121\n",
      "Epoch 831/5000\n",
      " - 10s - loss: 0.1600 - acc: 0.9466 - precision_m: 0.9701 - recall_m: 0.9225 - val_loss: 0.1636 - val_acc: 0.9480 - val_precision_m: 0.9721 - val_recall_m: 0.9304\n",
      "Epoch 832/5000\n",
      " - 10s - loss: 0.1558 - acc: 0.9482 - precision_m: 0.9712 - recall_m: 0.9241 - val_loss: 0.1654 - val_acc: 0.9468 - val_precision_m: 0.9721 - val_recall_m: 0.9283\n",
      "Epoch 833/5000\n",
      " - 10s - loss: 0.1586 - acc: 0.9479 - precision_m: 0.9712 - recall_m: 0.9236 - val_loss: 0.1636 - val_acc: 0.9474 - val_precision_m: 0.9591 - val_recall_m: 0.9424\n",
      "Epoch 834/5000\n",
      " - 10s - loss: 0.1560 - acc: 0.9483 - precision_m: 0.9708 - recall_m: 0.9247 - val_loss: 0.1650 - val_acc: 0.9435 - val_precision_m: 0.9780 - val_recall_m: 0.9167\n",
      "Epoch 835/5000\n",
      " - 10s - loss: 0.1602 - acc: 0.9466 - precision_m: 0.9685 - recall_m: 0.9242 - val_loss: 0.1856 - val_acc: 0.9258 - val_precision_m: 0.9805 - val_recall_m: 0.8808\n",
      "Epoch 836/5000\n",
      " - 10s - loss: 0.1648 - acc: 0.9433 - precision_m: 0.9669 - recall_m: 0.9196 - val_loss: 0.1751 - val_acc: 0.9457 - val_precision_m: 0.9444 - val_recall_m: 0.9563\n",
      "Epoch 837/5000\n",
      " - 10s - loss: 0.1567 - acc: 0.9465 - precision_m: 0.9709 - recall_m: 0.9205 - val_loss: 0.1694 - val_acc: 0.9480 - val_precision_m: 0.9592 - val_recall_m: 0.9441\n",
      "Epoch 838/5000\n",
      " - 10s - loss: 0.1661 - acc: 0.9426 - precision_m: 0.9648 - recall_m: 0.9201 - val_loss: 0.1635 - val_acc: 0.9485 - val_precision_m: 0.9770 - val_recall_m: 0.9265\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 839/5000\n",
      " - 10s - loss: 0.1563 - acc: 0.9486 - precision_m: 0.9718 - recall_m: 0.9250 - val_loss: 0.2275 - val_acc: 0.8970 - val_precision_m: 0.9852 - val_recall_m: 0.8224\n",
      "Epoch 840/5000\n",
      " - 10s - loss: 0.1544 - acc: 0.9485 - precision_m: 0.9709 - recall_m: 0.9247 - val_loss: 0.1756 - val_acc: 0.9313 - val_precision_m: 0.9817 - val_recall_m: 0.8897\n",
      "Epoch 841/5000\n",
      " - 10s - loss: 0.1513 - acc: 0.9507 - precision_m: 0.9730 - recall_m: 0.9272 - val_loss: 0.1600 - val_acc: 0.9480 - val_precision_m: 0.9812 - val_recall_m: 0.9210\n",
      "Epoch 842/5000\n",
      " - 10s - loss: 0.1571 - acc: 0.9482 - precision_m: 0.9707 - recall_m: 0.9249 - val_loss: 0.1835 - val_acc: 0.9225 - val_precision_m: 0.9835 - val_recall_m: 0.8717\n",
      "Epoch 843/5000\n",
      " - 10s - loss: 0.1618 - acc: 0.9431 - precision_m: 0.9673 - recall_m: 0.9184 - val_loss: 0.1626 - val_acc: 0.9468 - val_precision_m: 0.9561 - val_recall_m: 0.9448\n",
      "Epoch 844/5000\n",
      " - 10s - loss: 0.1545 - acc: 0.9488 - precision_m: 0.9717 - recall_m: 0.9249 - val_loss: 0.1604 - val_acc: 0.9518 - val_precision_m: 0.9693 - val_recall_m: 0.9408\n",
      "Epoch 845/5000\n",
      " - 10s - loss: 0.1519 - acc: 0.9486 - precision_m: 0.9706 - recall_m: 0.9261 - val_loss: 0.1649 - val_acc: 0.9402 - val_precision_m: 0.9809 - val_recall_m: 0.9073\n",
      "Epoch 846/5000\n",
      " - 10s - loss: 0.1522 - acc: 0.9501 - precision_m: 0.9717 - recall_m: 0.9275 - val_loss: 0.1848 - val_acc: 0.9269 - val_precision_m: 0.9838 - val_recall_m: 0.8800\n",
      "Epoch 847/5000\n",
      " - 10s - loss: 0.1632 - acc: 0.9442 - precision_m: 0.9686 - recall_m: 0.9199 - val_loss: 0.1654 - val_acc: 0.9502 - val_precision_m: 0.9753 - val_recall_m: 0.9315\n",
      "Epoch 848/5000\n",
      " - 10s - loss: 0.1515 - acc: 0.9520 - precision_m: 0.9728 - recall_m: 0.9307 - val_loss: 0.1637 - val_acc: 0.9463 - val_precision_m: 0.9612 - val_recall_m: 0.9391\n",
      "Epoch 849/5000\n",
      " - 10s - loss: 0.1539 - acc: 0.9487 - precision_m: 0.9707 - recall_m: 0.9256 - val_loss: 0.1846 - val_acc: 0.9269 - val_precision_m: 0.9826 - val_recall_m: 0.8807\n",
      "Epoch 850/5000\n",
      " - 10s - loss: 0.1533 - acc: 0.9492 - precision_m: 0.9710 - recall_m: 0.9266 - val_loss: 0.1847 - val_acc: 0.9264 - val_precision_m: 0.9839 - val_recall_m: 0.8783\n",
      "Epoch 851/5000\n",
      " - 10s - loss: 0.1552 - acc: 0.9485 - precision_m: 0.9707 - recall_m: 0.9251 - val_loss: 0.2025 - val_acc: 0.9374 - val_precision_m: 0.9174 - val_recall_m: 0.9727\n",
      "Epoch 852/5000\n",
      " - 10s - loss: 0.1559 - acc: 0.9488 - precision_m: 0.9707 - recall_m: 0.9254 - val_loss: 0.1637 - val_acc: 0.9419 - val_precision_m: 0.9597 - val_recall_m: 0.9314\n",
      "Epoch 853/5000\n",
      " - 10s - loss: 0.1496 - acc: 0.9497 - precision_m: 0.9725 - recall_m: 0.9253 - val_loss: 0.1612 - val_acc: 0.9474 - val_precision_m: 0.9719 - val_recall_m: 0.9291\n",
      "Epoch 854/5000\n",
      " - 10s - loss: 0.1486 - acc: 0.9504 - precision_m: 0.9728 - recall_m: 0.9266 - val_loss: 0.1598 - val_acc: 0.9441 - val_precision_m: 0.9779 - val_recall_m: 0.9169\n",
      "Epoch 855/5000\n",
      " - 10s - loss: 0.1514 - acc: 0.9490 - precision_m: 0.9719 - recall_m: 0.9249 - val_loss: 0.1698 - val_acc: 0.9435 - val_precision_m: 0.9502 - val_recall_m: 0.9445\n",
      "Epoch 856/5000\n",
      " - 10s - loss: 0.1568 - acc: 0.9455 - precision_m: 0.9666 - recall_m: 0.9242 - val_loss: 0.1909 - val_acc: 0.9214 - val_precision_m: 0.9870 - val_recall_m: 0.8662\n",
      "Epoch 857/5000\n",
      " - 10s - loss: 0.1655 - acc: 0.9430 - precision_m: 0.9678 - recall_m: 0.9180 - val_loss: 0.1773 - val_acc: 0.9463 - val_precision_m: 0.9423 - val_recall_m: 0.9604\n",
      "Epoch 858/5000\n",
      " - 10s - loss: 0.1547 - acc: 0.9493 - precision_m: 0.9719 - recall_m: 0.9261 - val_loss: 0.1610 - val_acc: 0.9446 - val_precision_m: 0.9747 - val_recall_m: 0.9215\n",
      "Epoch 859/5000\n",
      " - 10s - loss: 0.1474 - acc: 0.9518 - precision_m: 0.9737 - recall_m: 0.9285 - val_loss: 0.1632 - val_acc: 0.9435 - val_precision_m: 0.9768 - val_recall_m: 0.9173\n",
      "Epoch 860/5000\n",
      " - 10s - loss: 0.1554 - acc: 0.9485 - precision_m: 0.9720 - recall_m: 0.9246 - val_loss: 0.1621 - val_acc: 0.9430 - val_precision_m: 0.9626 - val_recall_m: 0.9303\n",
      "Epoch 861/5000\n",
      " - 10s - loss: 0.1521 - acc: 0.9494 - precision_m: 0.9731 - recall_m: 0.9249 - val_loss: 0.1718 - val_acc: 0.9463 - val_precision_m: 0.9458 - val_recall_m: 0.9550\n",
      "Epoch 862/5000\n",
      " - 10s - loss: 0.1517 - acc: 0.9486 - precision_m: 0.9712 - recall_m: 0.9247 - val_loss: 0.1558 - val_acc: 0.9540 - val_precision_m: 0.9674 - val_recall_m: 0.9470\n",
      "Epoch 863/5000\n",
      " - 10s - loss: 0.1528 - acc: 0.9482 - precision_m: 0.9710 - recall_m: 0.9244 - val_loss: 0.1785 - val_acc: 0.9280 - val_precision_m: 0.9815 - val_recall_m: 0.8840\n",
      "Epoch 864/5000\n",
      " - 10s - loss: 0.1515 - acc: 0.9490 - precision_m: 0.9721 - recall_m: 0.9259 - val_loss: 0.1558 - val_acc: 0.9513 - val_precision_m: 0.9722 - val_recall_m: 0.9363\n",
      "Epoch 865/5000\n",
      " - 10s - loss: 0.1487 - acc: 0.9514 - precision_m: 0.9715 - recall_m: 0.9300 - val_loss: 0.1568 - val_acc: 0.9518 - val_precision_m: 0.9721 - val_recall_m: 0.9374\n",
      "Epoch 866/5000\n",
      " - 10s - loss: 0.1532 - acc: 0.9490 - precision_m: 0.9704 - recall_m: 0.9273 - val_loss: 0.1604 - val_acc: 0.9435 - val_precision_m: 0.9777 - val_recall_m: 0.9162\n",
      "Epoch 867/5000\n",
      " - 10s - loss: 0.1530 - acc: 0.9484 - precision_m: 0.9712 - recall_m: 0.9249 - val_loss: 0.1573 - val_acc: 0.9507 - val_precision_m: 0.9704 - val_recall_m: 0.9377\n",
      "Epoch 868/5000\n",
      " - 10s - loss: 0.1566 - acc: 0.9469 - precision_m: 0.9687 - recall_m: 0.9247 - val_loss: 0.1944 - val_acc: 0.9169 - val_precision_m: 0.9833 - val_recall_m: 0.8608\n",
      "Epoch 869/5000\n",
      " - 10s - loss: 0.1536 - acc: 0.9484 - precision_m: 0.9719 - recall_m: 0.9242 - val_loss: 0.1994 - val_acc: 0.9125 - val_precision_m: 0.9832 - val_recall_m: 0.8530\n",
      "Epoch 870/5000\n",
      " - 10s - loss: 0.1490 - acc: 0.9511 - precision_m: 0.9737 - recall_m: 0.9275 - val_loss: 0.1711 - val_acc: 0.9347 - val_precision_m: 0.9798 - val_recall_m: 0.8979\n",
      "Epoch 871/5000\n",
      " - 10s - loss: 0.1464 - acc: 0.9521 - precision_m: 0.9728 - recall_m: 0.9308 - val_loss: 0.1676 - val_acc: 0.9352 - val_precision_m: 0.9807 - val_recall_m: 0.8984\n",
      "Epoch 872/5000\n",
      " - 10s - loss: 0.1582 - acc: 0.9453 - precision_m: 0.9681 - recall_m: 0.9223 - val_loss: 0.1550 - val_acc: 0.9485 - val_precision_m: 0.9771 - val_recall_m: 0.9268\n",
      "Epoch 873/5000\n",
      " - 10s - loss: 0.1534 - acc: 0.9476 - precision_m: 0.9716 - recall_m: 0.9230 - val_loss: 0.1592 - val_acc: 0.9518 - val_precision_m: 0.9655 - val_recall_m: 0.9449\n",
      "Epoch 874/5000\n",
      " - 10s - loss: 0.1476 - acc: 0.9511 - precision_m: 0.9726 - recall_m: 0.9287 - val_loss: 0.1563 - val_acc: 0.9485 - val_precision_m: 0.9811 - val_recall_m: 0.9226\n",
      "Epoch 875/5000\n",
      " - 10s - loss: 0.1458 - acc: 0.9525 - precision_m: 0.9748 - recall_m: 0.9293 - val_loss: 0.1516 - val_acc: 0.9502 - val_precision_m: 0.9772 - val_recall_m: 0.9293\n",
      "Epoch 876/5000\n",
      " - 10s - loss: 0.1440 - acc: 0.9530 - precision_m: 0.9744 - recall_m: 0.9303 - val_loss: 0.1556 - val_acc: 0.9502 - val_precision_m: 0.9613 - val_recall_m: 0.9459\n",
      "Epoch 877/5000\n",
      " - 10s - loss: 0.1500 - acc: 0.9501 - precision_m: 0.9711 - recall_m: 0.9284 - val_loss: 0.2399 - val_acc: 0.8981 - val_precision_m: 0.9913 - val_recall_m: 0.8191\n",
      "Epoch 878/5000\n",
      " - 10s - loss: 0.1538 - acc: 0.9490 - precision_m: 0.9729 - recall_m: 0.9252 - val_loss: 0.1624 - val_acc: 0.9457 - val_precision_m: 0.9736 - val_recall_m: 0.9247\n",
      "Epoch 879/5000\n",
      " - 10s - loss: 0.1493 - acc: 0.9504 - precision_m: 0.9701 - recall_m: 0.9294 - val_loss: 0.1589 - val_acc: 0.9535 - val_precision_m: 0.9714 - val_recall_m: 0.9418\n",
      "Epoch 880/5000\n",
      " - 10s - loss: 0.1497 - acc: 0.9496 - precision_m: 0.9718 - recall_m: 0.9263 - val_loss: 0.1844 - val_acc: 0.9236 - val_precision_m: 0.9860 - val_recall_m: 0.8717\n",
      "Epoch 881/5000\n",
      " - 10s - loss: 0.1501 - acc: 0.9469 - precision_m: 0.9684 - recall_m: 0.9247 - val_loss: 0.1837 - val_acc: 0.9402 - val_precision_m: 0.9292 - val_recall_m: 0.9635\n",
      "Epoch 882/5000\n",
      " - 10s - loss: 0.1450 - acc: 0.9519 - precision_m: 0.9731 - recall_m: 0.9298 - val_loss: 0.1598 - val_acc: 0.9491 - val_precision_m: 0.9584 - val_recall_m: 0.9471\n",
      "Epoch 883/5000\n",
      " - 10s - loss: 0.1439 - acc: 0.9527 - precision_m: 0.9736 - recall_m: 0.9310 - val_loss: 0.1652 - val_acc: 0.9385 - val_precision_m: 0.9820 - val_recall_m: 0.9031\n",
      "Epoch 884/5000\n",
      " - 10s - loss: 0.1551 - acc: 0.9473 - precision_m: 0.9694 - recall_m: 0.9249 - val_loss: 0.1970 - val_acc: 0.9103 - val_precision_m: 0.9843 - val_recall_m: 0.8486\n",
      "Epoch 885/5000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " - 10s - loss: 0.1456 - acc: 0.9531 - precision_m: 0.9748 - recall_m: 0.9308 - val_loss: 0.1558 - val_acc: 0.9463 - val_precision_m: 0.9728 - val_recall_m: 0.9264\n",
      "Epoch 886/5000\n",
      " - 10s - loss: 0.1417 - acc: 0.9537 - precision_m: 0.9740 - recall_m: 0.9324 - val_loss: 0.1549 - val_acc: 0.9529 - val_precision_m: 0.9674 - val_recall_m: 0.9448\n",
      "Epoch 887/5000\n",
      " - 10s - loss: 0.1437 - acc: 0.9520 - precision_m: 0.9725 - recall_m: 0.9307 - val_loss: 0.1589 - val_acc: 0.9529 - val_precision_m: 0.9685 - val_recall_m: 0.9440\n",
      "Epoch 888/5000\n",
      " - 10s - loss: 0.1457 - acc: 0.9511 - precision_m: 0.9737 - recall_m: 0.9272 - val_loss: 0.1587 - val_acc: 0.9513 - val_precision_m: 0.9565 - val_recall_m: 0.9533\n",
      "Epoch 889/5000\n",
      " - 10s - loss: 0.1636 - acc: 0.9419 - precision_m: 0.9643 - recall_m: 0.9203 - val_loss: 0.1593 - val_acc: 0.9380 - val_precision_m: 0.9830 - val_recall_m: 0.9010\n",
      "Epoch 890/5000\n",
      " - 10s - loss: 0.1433 - acc: 0.9546 - precision_m: 0.9738 - recall_m: 0.9339 - val_loss: 0.1721 - val_acc: 0.9269 - val_precision_m: 0.9794 - val_recall_m: 0.8839\n",
      "Epoch 891/5000\n",
      " - 10s - loss: 0.1481 - acc: 0.9513 - precision_m: 0.9746 - recall_m: 0.9270 - val_loss: 0.1595 - val_acc: 0.9424 - val_precision_m: 0.9777 - val_recall_m: 0.9140\n",
      "Epoch 892/5000\n",
      " - 10s - loss: 0.1516 - acc: 0.9501 - precision_m: 0.9729 - recall_m: 0.9276 - val_loss: 0.1497 - val_acc: 0.9518 - val_precision_m: 0.9793 - val_recall_m: 0.9303\n",
      "Epoch 893/5000\n",
      " - 10s - loss: 0.1469 - acc: 0.9516 - precision_m: 0.9708 - recall_m: 0.9319 - val_loss: 0.1537 - val_acc: 0.9474 - val_precision_m: 0.9749 - val_recall_m: 0.9261\n",
      "Epoch 894/5000\n",
      " - 10s - loss: 0.1493 - acc: 0.9488 - precision_m: 0.9715 - recall_m: 0.9257 - val_loss: 0.1553 - val_acc: 0.9491 - val_precision_m: 0.9792 - val_recall_m: 0.9258\n",
      "Epoch 895/5000\n",
      " - 10s - loss: 0.1454 - acc: 0.9516 - precision_m: 0.9713 - recall_m: 0.9306 - val_loss: 0.1595 - val_acc: 0.9374 - val_precision_m: 0.9809 - val_recall_m: 0.9021\n",
      "Epoch 896/5000\n",
      " - 10s - loss: 0.1434 - acc: 0.9523 - precision_m: 0.9736 - recall_m: 0.9305 - val_loss: 0.1529 - val_acc: 0.9557 - val_precision_m: 0.9732 - val_recall_m: 0.9436\n",
      "Epoch 897/5000\n",
      " - 10s - loss: 0.1396 - acc: 0.9545 - precision_m: 0.9746 - recall_m: 0.9332 - val_loss: 0.1623 - val_acc: 0.9374 - val_precision_m: 0.9833 - val_recall_m: 0.9002\n",
      "Epoch 898/5000\n",
      " - 10s - loss: 0.1432 - acc: 0.9537 - precision_m: 0.9753 - recall_m: 0.9316 - val_loss: 0.1753 - val_acc: 0.9286 - val_precision_m: 0.9815 - val_recall_m: 0.8844\n",
      "Epoch 899/5000\n",
      " - 10s - loss: 0.1428 - acc: 0.9533 - precision_m: 0.9733 - recall_m: 0.9327 - val_loss: 0.1514 - val_acc: 0.9563 - val_precision_m: 0.9650 - val_recall_m: 0.9542\n",
      "Epoch 900/5000\n",
      " - 10s - loss: 0.1459 - acc: 0.9528 - precision_m: 0.9729 - recall_m: 0.9319 - val_loss: 0.1580 - val_acc: 0.9435 - val_precision_m: 0.9798 - val_recall_m: 0.9145\n",
      "Epoch 901/5000\n",
      " - 10s - loss: 0.1470 - acc: 0.9502 - precision_m: 0.9719 - recall_m: 0.9288 - val_loss: 0.1686 - val_acc: 0.9330 - val_precision_m: 0.9786 - val_recall_m: 0.8961\n",
      "Epoch 902/5000\n",
      " - 10s - loss: 0.1460 - acc: 0.9509 - precision_m: 0.9712 - recall_m: 0.9297 - val_loss: 0.1501 - val_acc: 0.9524 - val_precision_m: 0.9743 - val_recall_m: 0.9367\n",
      "Epoch 903/5000\n",
      " - 10s - loss: 0.1410 - acc: 0.9533 - precision_m: 0.9724 - recall_m: 0.9334 - val_loss: 0.2021 - val_acc: 0.9092 - val_precision_m: 0.9856 - val_recall_m: 0.8453\n",
      "Epoch 904/5000\n",
      " - 10s - loss: 0.1416 - acc: 0.9531 - precision_m: 0.9727 - recall_m: 0.9326 - val_loss: 0.1514 - val_acc: 0.9563 - val_precision_m: 0.9695 - val_recall_m: 0.9488\n",
      "Epoch 905/5000\n",
      " - 10s - loss: 0.1469 - acc: 0.9506 - precision_m: 0.9723 - recall_m: 0.9281 - val_loss: 0.1592 - val_acc: 0.9430 - val_precision_m: 0.9477 - val_recall_m: 0.9470\n",
      "Epoch 906/5000\n",
      " - 10s - loss: 0.1418 - acc: 0.9524 - precision_m: 0.9721 - recall_m: 0.9315 - val_loss: 0.1469 - val_acc: 0.9491 - val_precision_m: 0.9750 - val_recall_m: 0.9295\n",
      "Epoch 907/5000\n",
      " - 10s - loss: 0.1410 - acc: 0.9535 - precision_m: 0.9748 - recall_m: 0.9316 - val_loss: 0.1580 - val_acc: 0.9513 - val_precision_m: 0.9560 - val_recall_m: 0.9538\n",
      "Epoch 908/5000\n",
      " - 10s - loss: 0.1385 - acc: 0.9541 - precision_m: 0.9752 - recall_m: 0.9320 - val_loss: 0.1540 - val_acc: 0.9496 - val_precision_m: 0.9692 - val_recall_m: 0.9366\n",
      "Epoch 909/5000\n",
      " - 10s - loss: 0.1395 - acc: 0.9543 - precision_m: 0.9745 - recall_m: 0.9335 - val_loss: 0.1495 - val_acc: 0.9518 - val_precision_m: 0.9692 - val_recall_m: 0.9406\n",
      "Epoch 910/5000\n",
      " - 10s - loss: 0.1390 - acc: 0.9546 - precision_m: 0.9758 - recall_m: 0.9328 - val_loss: 0.1481 - val_acc: 0.9518 - val_precision_m: 0.9722 - val_recall_m: 0.9376\n",
      "Epoch 911/5000\n",
      " - 10s - loss: 0.1406 - acc: 0.9534 - precision_m: 0.9733 - recall_m: 0.9332 - val_loss: 0.1507 - val_acc: 0.9424 - val_precision_m: 0.9768 - val_recall_m: 0.9154\n",
      "Epoch 912/5000\n",
      " - 10s - loss: 0.1398 - acc: 0.9536 - precision_m: 0.9739 - recall_m: 0.9320 - val_loss: 0.1517 - val_acc: 0.9491 - val_precision_m: 0.9753 - val_recall_m: 0.9294\n",
      "Epoch 913/5000\n",
      " - 10s - loss: 0.1419 - acc: 0.9516 - precision_m: 0.9714 - recall_m: 0.9306 - val_loss: 0.1513 - val_acc: 0.9529 - val_precision_m: 0.9695 - val_recall_m: 0.9426\n",
      "Epoch 914/5000\n",
      " - 10s - loss: 0.1395 - acc: 0.9531 - precision_m: 0.9722 - recall_m: 0.9329 - val_loss: 0.1642 - val_acc: 0.9358 - val_precision_m: 0.9863 - val_recall_m: 0.8944\n",
      "Epoch 915/5000\n",
      " - 10s - loss: 0.1398 - acc: 0.9529 - precision_m: 0.9728 - recall_m: 0.9319 - val_loss: 0.1442 - val_acc: 0.9546 - val_precision_m: 0.9674 - val_recall_m: 0.9479\n",
      "Epoch 916/5000\n",
      " - 10s - loss: 0.1399 - acc: 0.9545 - precision_m: 0.9747 - recall_m: 0.9337 - val_loss: 0.1446 - val_acc: 0.9557 - val_precision_m: 0.9763 - val_recall_m: 0.9405\n",
      "Epoch 917/5000\n",
      " - 10s - loss: 0.1414 - acc: 0.9534 - precision_m: 0.9739 - recall_m: 0.9326 - val_loss: 0.1766 - val_acc: 0.9457 - val_precision_m: 0.9362 - val_recall_m: 0.9665\n",
      "Epoch 918/5000\n",
      " - 10s - loss: 0.1435 - acc: 0.9503 - precision_m: 0.9712 - recall_m: 0.9288 - val_loss: 0.1553 - val_acc: 0.9529 - val_precision_m: 0.9712 - val_recall_m: 0.9406\n",
      "Epoch 919/5000\n",
      " - 10s - loss: 0.1446 - acc: 0.9524 - precision_m: 0.9737 - recall_m: 0.9312 - val_loss: 0.1553 - val_acc: 0.9419 - val_precision_m: 0.9821 - val_recall_m: 0.9099\n",
      "Epoch 920/5000\n",
      " - 10s - loss: 0.1423 - acc: 0.9516 - precision_m: 0.9722 - recall_m: 0.9299 - val_loss: 0.1545 - val_acc: 0.9419 - val_precision_m: 0.9822 - val_recall_m: 0.9098\n",
      "Epoch 921/5000\n",
      " - 10s - loss: 0.1373 - acc: 0.9558 - precision_m: 0.9749 - recall_m: 0.9364 - val_loss: 0.1540 - val_acc: 0.9452 - val_precision_m: 0.9782 - val_recall_m: 0.9192\n",
      "Epoch 922/5000\n",
      " - 10s - loss: 0.1404 - acc: 0.9536 - precision_m: 0.9741 - recall_m: 0.9324 - val_loss: 0.1519 - val_acc: 0.9419 - val_precision_m: 0.9798 - val_recall_m: 0.9111\n",
      "Epoch 923/5000\n",
      " - 10s - loss: 0.1341 - acc: 0.9563 - precision_m: 0.9753 - recall_m: 0.9360 - val_loss: 0.1470 - val_acc: 0.9563 - val_precision_m: 0.9675 - val_recall_m: 0.9512\n",
      "Epoch 924/5000\n",
      " - 10s - loss: 0.1452 - acc: 0.9519 - precision_m: 0.9721 - recall_m: 0.9313 - val_loss: 0.2226 - val_acc: 0.9264 - val_precision_m: 0.8936 - val_recall_m: 0.9812\n",
      "Epoch 925/5000\n",
      " - 10s - loss: 0.1493 - acc: 0.9492 - precision_m: 0.9694 - recall_m: 0.9289 - val_loss: 0.1781 - val_acc: 0.9236 - val_precision_m: 0.9850 - val_recall_m: 0.8730\n",
      "Epoch 926/5000\n",
      " - 10s - loss: 0.1370 - acc: 0.9546 - precision_m: 0.9743 - recall_m: 0.9342 - val_loss: 0.1676 - val_acc: 0.9280 - val_precision_m: 0.9826 - val_recall_m: 0.8832\n",
      "Epoch 927/5000\n",
      " - 10s - loss: 0.1341 - acc: 0.9569 - precision_m: 0.9766 - recall_m: 0.9370 - val_loss: 0.1474 - val_acc: 0.9474 - val_precision_m: 0.9811 - val_recall_m: 0.9203\n",
      "Epoch 928/5000\n",
      " - 10s - loss: 0.1351 - acc: 0.9550 - precision_m: 0.9738 - recall_m: 0.9349 - val_loss: 0.1442 - val_acc: 0.9524 - val_precision_m: 0.9802 - val_recall_m: 0.9304\n",
      "Epoch 929/5000\n",
      " - 10s - loss: 0.1329 - acc: 0.9551 - precision_m: 0.9756 - recall_m: 0.9334 - val_loss: 0.1462 - val_acc: 0.9513 - val_precision_m: 0.9672 - val_recall_m: 0.9413\n",
      "Epoch 930/5000\n",
      " - 10s - loss: 0.1492 - acc: 0.9498 - precision_m: 0.9717 - recall_m: 0.9281 - val_loss: 0.1467 - val_acc: 0.9518 - val_precision_m: 0.9762 - val_recall_m: 0.9334\n",
      "Epoch 931/5000\n",
      " - 10s - loss: 0.1422 - acc: 0.9525 - precision_m: 0.9722 - recall_m: 0.9322 - val_loss: 0.1676 - val_acc: 0.9513 - val_precision_m: 0.9497 - val_recall_m: 0.9611\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 932/5000\n",
      " - 10s - loss: 0.1370 - acc: 0.9551 - precision_m: 0.9755 - recall_m: 0.9340 - val_loss: 0.1580 - val_acc: 0.9408 - val_precision_m: 0.9831 - val_recall_m: 0.9059\n",
      "Epoch 933/5000\n",
      " - 10s - loss: 0.1345 - acc: 0.9562 - precision_m: 0.9760 - recall_m: 0.9360 - val_loss: 0.1635 - val_acc: 0.9330 - val_precision_m: 0.9829 - val_recall_m: 0.8921\n",
      "Epoch 934/5000\n",
      " - 10s - loss: 0.1358 - acc: 0.9553 - precision_m: 0.9753 - recall_m: 0.9350 - val_loss: 0.1626 - val_acc: 0.9363 - val_precision_m: 0.9830 - val_recall_m: 0.8980\n",
      "Epoch 935/5000\n",
      " - 10s - loss: 0.1371 - acc: 0.9557 - precision_m: 0.9758 - recall_m: 0.9349 - val_loss: 0.1777 - val_acc: 0.9430 - val_precision_m: 0.9260 - val_recall_m: 0.9727\n",
      "Epoch 936/5000\n",
      " - 10s - loss: 0.1459 - acc: 0.9497 - precision_m: 0.9703 - recall_m: 0.9290 - val_loss: 0.1610 - val_acc: 0.9524 - val_precision_m: 0.9521 - val_recall_m: 0.9601\n",
      "Epoch 937/5000\n",
      " - 10s - loss: 0.1372 - acc: 0.9558 - precision_m: 0.9764 - recall_m: 0.9346 - val_loss: 0.1452 - val_acc: 0.9480 - val_precision_m: 0.9770 - val_recall_m: 0.9256\n",
      "Epoch 938/5000\n",
      " - 10s - loss: 0.1408 - acc: 0.9532 - precision_m: 0.9730 - recall_m: 0.9334 - val_loss: 0.1641 - val_acc: 0.9363 - val_precision_m: 0.9820 - val_recall_m: 0.8993\n",
      "Epoch 939/5000\n",
      " - 10s - loss: 0.1333 - acc: 0.9568 - precision_m: 0.9755 - recall_m: 0.9374 - val_loss: 0.1839 - val_acc: 0.9225 - val_precision_m: 0.9880 - val_recall_m: 0.8678\n",
      "Epoch 940/5000\n",
      " - 10s - loss: 0.1417 - acc: 0.9517 - precision_m: 0.9726 - recall_m: 0.9304 - val_loss: 0.1503 - val_acc: 0.9396 - val_precision_m: 0.9767 - val_recall_m: 0.9104\n",
      "Epoch 941/5000\n",
      " - 10s - loss: 0.1321 - acc: 0.9572 - precision_m: 0.9762 - recall_m: 0.9376 - val_loss: 0.1708 - val_acc: 0.9430 - val_precision_m: 0.9281 - val_recall_m: 0.9696\n",
      "Epoch 942/5000\n",
      " - 10s - loss: 0.1338 - acc: 0.9565 - precision_m: 0.9765 - recall_m: 0.9356 - val_loss: 0.1437 - val_acc: 0.9524 - val_precision_m: 0.9595 - val_recall_m: 0.9515\n",
      "Epoch 943/5000\n",
      " - 10s - loss: 0.1336 - acc: 0.9548 - precision_m: 0.9724 - recall_m: 0.9368 - val_loss: 0.1427 - val_acc: 0.9540 - val_precision_m: 0.9753 - val_recall_m: 0.9383\n",
      "Epoch 944/5000\n",
      " - 10s - loss: 0.1339 - acc: 0.9565 - precision_m: 0.9753 - recall_m: 0.9372 - val_loss: 0.1448 - val_acc: 0.9524 - val_precision_m: 0.9569 - val_recall_m: 0.9549\n",
      "Epoch 945/5000\n",
      " - 10s - loss: 0.1356 - acc: 0.9540 - precision_m: 0.9729 - recall_m: 0.9339 - val_loss: 0.1694 - val_acc: 0.9441 - val_precision_m: 0.9327 - val_recall_m: 0.9664\n",
      "Epoch 946/5000\n",
      " - 10s - loss: 0.1338 - acc: 0.9559 - precision_m: 0.9740 - recall_m: 0.9372 - val_loss: 0.1471 - val_acc: 0.9480 - val_precision_m: 0.9813 - val_recall_m: 0.9216\n",
      "Epoch 947/5000\n",
      " - 10s - loss: 0.1314 - acc: 0.9559 - precision_m: 0.9755 - recall_m: 0.9353 - val_loss: 0.1413 - val_acc: 0.9579 - val_precision_m: 0.9707 - val_recall_m: 0.9511\n",
      "Epoch 948/5000\n",
      " - 10s - loss: 0.1338 - acc: 0.9562 - precision_m: 0.9750 - recall_m: 0.9370 - val_loss: 0.1660 - val_acc: 0.9358 - val_precision_m: 0.9828 - val_recall_m: 0.8969\n",
      "Epoch 949/5000\n",
      " - 10s - loss: 0.1395 - acc: 0.9529 - precision_m: 0.9720 - recall_m: 0.9331 - val_loss: 0.2156 - val_acc: 0.9081 - val_precision_m: 0.9888 - val_recall_m: 0.8324\n",
      "Epoch 950/5000\n",
      " - 10s - loss: 0.1319 - acc: 0.9563 - precision_m: 0.9754 - recall_m: 0.9367 - val_loss: 0.1383 - val_acc: 0.9574 - val_precision_m: 0.9715 - val_recall_m: 0.9486\n",
      "Epoch 951/5000\n",
      " - 10s - loss: 0.1328 - acc: 0.9558 - precision_m: 0.9745 - recall_m: 0.9362 - val_loss: 0.1400 - val_acc: 0.9551 - val_precision_m: 0.9715 - val_recall_m: 0.9445\n",
      "Epoch 952/5000\n",
      " - 10s - loss: 0.1298 - acc: 0.9577 - precision_m: 0.9763 - recall_m: 0.9381 - val_loss: 0.1432 - val_acc: 0.9485 - val_precision_m: 0.9771 - val_recall_m: 0.9267\n",
      "Epoch 953/5000\n",
      " - 10s - loss: 0.1290 - acc: 0.9584 - precision_m: 0.9773 - recall_m: 0.9391 - val_loss: 0.1434 - val_acc: 0.9496 - val_precision_m: 0.9753 - val_recall_m: 0.9305\n",
      "Epoch 954/5000\n",
      " - 10s - loss: 0.1338 - acc: 0.9559 - precision_m: 0.9759 - recall_m: 0.9351 - val_loss: 0.1461 - val_acc: 0.9579 - val_precision_m: 0.9708 - val_recall_m: 0.9508\n",
      "Epoch 955/5000\n",
      " - 10s - loss: 0.1364 - acc: 0.9551 - precision_m: 0.9751 - recall_m: 0.9348 - val_loss: 0.1422 - val_acc: 0.9480 - val_precision_m: 0.9802 - val_recall_m: 0.9226\n",
      "Epoch 956/5000\n",
      " - 10s - loss: 0.1325 - acc: 0.9569 - precision_m: 0.9746 - recall_m: 0.9389 - val_loss: 0.1427 - val_acc: 0.9540 - val_precision_m: 0.9805 - val_recall_m: 0.9335\n",
      "Epoch 957/5000\n",
      " - 10s - loss: 0.1302 - acc: 0.9580 - precision_m: 0.9763 - recall_m: 0.9390 - val_loss: 0.1383 - val_acc: 0.9568 - val_precision_m: 0.9723 - val_recall_m: 0.9465\n",
      "Epoch 958/5000\n",
      " - 10s - loss: 0.1366 - acc: 0.9538 - precision_m: 0.9736 - recall_m: 0.9338 - val_loss: 0.1361 - val_acc: 0.9535 - val_precision_m: 0.9775 - val_recall_m: 0.9358\n",
      "Epoch 959/5000\n",
      " - 10s - loss: 0.1373 - acc: 0.9536 - precision_m: 0.9734 - recall_m: 0.9331 - val_loss: 0.1550 - val_acc: 0.9535 - val_precision_m: 0.9525 - val_recall_m: 0.9622\n",
      "Epoch 960/5000\n",
      " - 10s - loss: 0.1336 - acc: 0.9553 - precision_m: 0.9739 - recall_m: 0.9356 - val_loss: 0.1409 - val_acc: 0.9535 - val_precision_m: 0.9774 - val_recall_m: 0.9356\n",
      "Epoch 961/5000\n",
      " - 10s - loss: 0.1299 - acc: 0.9569 - precision_m: 0.9750 - recall_m: 0.9380 - val_loss: 0.1766 - val_acc: 0.9214 - val_precision_m: 0.9848 - val_recall_m: 0.8686\n",
      "Epoch 962/5000\n",
      " - 10s - loss: 0.1309 - acc: 0.9572 - precision_m: 0.9761 - recall_m: 0.9380 - val_loss: 0.1785 - val_acc: 0.9236 - val_precision_m: 0.9824 - val_recall_m: 0.8747\n",
      "Epoch 963/5000\n",
      " - 10s - loss: 0.1355 - acc: 0.9531 - precision_m: 0.9739 - recall_m: 0.9312 - val_loss: 0.1367 - val_acc: 0.9540 - val_precision_m: 0.9803 - val_recall_m: 0.9334\n",
      "Epoch 964/5000\n",
      " - 10s - loss: 0.1288 - acc: 0.9580 - precision_m: 0.9773 - recall_m: 0.9385 - val_loss: 0.1666 - val_acc: 0.9291 - val_precision_m: 0.9836 - val_recall_m: 0.8835\n",
      "Epoch 965/5000\n",
      " - 10s - loss: 0.1291 - acc: 0.9589 - precision_m: 0.9752 - recall_m: 0.9422 - val_loss: 0.1415 - val_acc: 0.9568 - val_precision_m: 0.9608 - val_recall_m: 0.9593\n",
      "Epoch 966/5000\n",
      " - 10s - loss: 0.1296 - acc: 0.9570 - precision_m: 0.9747 - recall_m: 0.9387 - val_loss: 0.1355 - val_acc: 0.9557 - val_precision_m: 0.9753 - val_recall_m: 0.9417\n",
      "Epoch 967/5000\n",
      " - 10s - loss: 0.1313 - acc: 0.9561 - precision_m: 0.9751 - recall_m: 0.9369 - val_loss: 0.1402 - val_acc: 0.9524 - val_precision_m: 0.9773 - val_recall_m: 0.9333\n",
      "Epoch 968/5000\n",
      " - 10s - loss: 0.1265 - acc: 0.9588 - precision_m: 0.9761 - recall_m: 0.9405 - val_loss: 0.1360 - val_acc: 0.9540 - val_precision_m: 0.9763 - val_recall_m: 0.9379\n",
      "Epoch 969/5000\n",
      " - 10s - loss: 0.1286 - acc: 0.9578 - precision_m: 0.9765 - recall_m: 0.9381 - val_loss: 0.1424 - val_acc: 0.9485 - val_precision_m: 0.9812 - val_recall_m: 0.9225\n",
      "Epoch 970/5000\n",
      " - 10s - loss: 0.1300 - acc: 0.9585 - precision_m: 0.9764 - recall_m: 0.9398 - val_loss: 0.1570 - val_acc: 0.9513 - val_precision_m: 0.9477 - val_recall_m: 0.9635\n",
      "Epoch 971/5000\n",
      " - 10s - loss: 0.1302 - acc: 0.9566 - precision_m: 0.9729 - recall_m: 0.9394 - val_loss: 0.1361 - val_acc: 0.9574 - val_precision_m: 0.9794 - val_recall_m: 0.9407\n",
      "Epoch 972/5000\n",
      " - 10s - loss: 0.1302 - acc: 0.9580 - precision_m: 0.9760 - recall_m: 0.9401 - val_loss: 0.1312 - val_acc: 0.9557 - val_precision_m: 0.9682 - val_recall_m: 0.9489\n",
      "Epoch 973/5000\n",
      " - 10s - loss: 0.1256 - acc: 0.9585 - precision_m: 0.9749 - recall_m: 0.9407 - val_loss: 0.1421 - val_acc: 0.9480 - val_precision_m: 0.9822 - val_recall_m: 0.9201\n",
      "Epoch 974/5000\n",
      " - 10s - loss: 0.1282 - acc: 0.9578 - precision_m: 0.9759 - recall_m: 0.9392 - val_loss: 0.1401 - val_acc: 0.9463 - val_precision_m: 0.9812 - val_recall_m: 0.9183\n",
      "Epoch 975/5000\n",
      " - 10s - loss: 0.1257 - acc: 0.9587 - precision_m: 0.9772 - recall_m: 0.9395 - val_loss: 0.1392 - val_acc: 0.9563 - val_precision_m: 0.9674 - val_recall_m: 0.9510\n",
      "Epoch 976/5000\n",
      " - 10s - loss: 0.1325 - acc: 0.9556 - precision_m: 0.9730 - recall_m: 0.9377 - val_loss: 0.2185 - val_acc: 0.9031 - val_precision_m: 0.9902 - val_recall_m: 0.8297\n",
      "Epoch 977/5000\n",
      " - 10s - loss: 0.1351 - acc: 0.9537 - precision_m: 0.9734 - recall_m: 0.9340 - val_loss: 0.1355 - val_acc: 0.9590 - val_precision_m: 0.9718 - val_recall_m: 0.9520\n",
      "Epoch 978/5000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " - 10s - loss: 0.1240 - acc: 0.9594 - precision_m: 0.9762 - recall_m: 0.9420 - val_loss: 0.1362 - val_acc: 0.9574 - val_precision_m: 0.9757 - val_recall_m: 0.9449\n",
      "Epoch 979/5000\n",
      " - 10s - loss: 0.1324 - acc: 0.9567 - precision_m: 0.9741 - recall_m: 0.9384 - val_loss: 0.1568 - val_acc: 0.9313 - val_precision_m: 0.9838 - val_recall_m: 0.8868\n",
      "Epoch 980/5000\n",
      " - 10s - loss: 0.1243 - acc: 0.9594 - precision_m: 0.9767 - recall_m: 0.9413 - val_loss: 0.1399 - val_acc: 0.9446 - val_precision_m: 0.9790 - val_recall_m: 0.9172\n",
      "Epoch 981/5000\n",
      " - 10s - loss: 0.1340 - acc: 0.9553 - precision_m: 0.9727 - recall_m: 0.9375 - val_loss: 0.1362 - val_acc: 0.9574 - val_precision_m: 0.9796 - val_recall_m: 0.9406\n",
      "Epoch 982/5000\n",
      " - 10s - loss: 0.1314 - acc: 0.9565 - precision_m: 0.9746 - recall_m: 0.9379 - val_loss: 0.1439 - val_acc: 0.9502 - val_precision_m: 0.9814 - val_recall_m: 0.9254\n",
      "Epoch 983/5000\n",
      " - 10s - loss: 0.1248 - acc: 0.9595 - precision_m: 0.9745 - recall_m: 0.9442 - val_loss: 0.1329 - val_acc: 0.9574 - val_precision_m: 0.9784 - val_recall_m: 0.9415\n",
      "Epoch 984/5000\n",
      " - 10s - loss: 0.1286 - acc: 0.9575 - precision_m: 0.9762 - recall_m: 0.9384 - val_loss: 0.1317 - val_acc: 0.9585 - val_precision_m: 0.9764 - val_recall_m: 0.9459\n",
      "Epoch 985/5000\n",
      " - 10s - loss: 0.1262 - acc: 0.9584 - precision_m: 0.9744 - recall_m: 0.9418 - val_loss: 0.1380 - val_acc: 0.9491 - val_precision_m: 0.9802 - val_recall_m: 0.9245\n",
      "Epoch 986/5000\n",
      " - 10s - loss: 0.1263 - acc: 0.9583 - precision_m: 0.9760 - recall_m: 0.9401 - val_loss: 0.1424 - val_acc: 0.9474 - val_precision_m: 0.9823 - val_recall_m: 0.9196\n",
      "Epoch 987/5000\n",
      " - 10s - loss: 0.1237 - acc: 0.9587 - precision_m: 0.9765 - recall_m: 0.9400 - val_loss: 0.1331 - val_acc: 0.9563 - val_precision_m: 0.9704 - val_recall_m: 0.9476\n",
      "Epoch 988/5000\n",
      " - 10s - loss: 0.1350 - acc: 0.9547 - precision_m: 0.9734 - recall_m: 0.9360 - val_loss: 0.1557 - val_acc: 0.9358 - val_precision_m: 0.9853 - val_recall_m: 0.8948\n",
      "Epoch 989/5000\n",
      " - 10s - loss: 0.1269 - acc: 0.9580 - precision_m: 0.9757 - recall_m: 0.9394 - val_loss: 0.1315 - val_acc: 0.9574 - val_precision_m: 0.9776 - val_recall_m: 0.9426\n",
      "Epoch 990/5000\n",
      " - 10s - loss: 0.1288 - acc: 0.9565 - precision_m: 0.9741 - recall_m: 0.9382 - val_loss: 0.1607 - val_acc: 0.9336 - val_precision_m: 0.9875 - val_recall_m: 0.8893\n",
      "Epoch 991/5000\n",
      " - 10s - loss: 0.1273 - acc: 0.9577 - precision_m: 0.9753 - recall_m: 0.9391 - val_loss: 0.1634 - val_acc: 0.9291 - val_precision_m: 0.9850 - val_recall_m: 0.8829\n",
      "Epoch 992/5000\n",
      " - 10s - loss: 0.1256 - acc: 0.9581 - precision_m: 0.9755 - recall_m: 0.9399 - val_loss: 0.1694 - val_acc: 0.9252 - val_precision_m: 0.9861 - val_recall_m: 0.8749\n",
      "Epoch 993/5000\n",
      " - 10s - loss: 0.1269 - acc: 0.9575 - precision_m: 0.9762 - recall_m: 0.9379 - val_loss: 0.1418 - val_acc: 0.9480 - val_precision_m: 0.9845 - val_recall_m: 0.9183\n",
      "Epoch 994/5000\n",
      " - 10s - loss: 0.1224 - acc: 0.9609 - precision_m: 0.9780 - recall_m: 0.9431 - val_loss: 0.1412 - val_acc: 0.9579 - val_precision_m: 0.9608 - val_recall_m: 0.9612\n",
      "Epoch 995/5000\n",
      " - 10s - loss: 0.1275 - acc: 0.9591 - precision_m: 0.9763 - recall_m: 0.9415 - val_loss: 0.1349 - val_acc: 0.9568 - val_precision_m: 0.9764 - val_recall_m: 0.9429\n",
      "Epoch 996/5000\n",
      " - 10s - loss: 0.1226 - acc: 0.9589 - precision_m: 0.9763 - recall_m: 0.9410 - val_loss: 0.1508 - val_acc: 0.9396 - val_precision_m: 0.9865 - val_recall_m: 0.9012\n",
      "Epoch 997/5000\n",
      " - 10s - loss: 0.1230 - acc: 0.9599 - precision_m: 0.9793 - recall_m: 0.9395 - val_loss: 0.1355 - val_acc: 0.9568 - val_precision_m: 0.9655 - val_recall_m: 0.9541\n",
      "Epoch 998/5000\n",
      " - 10s - loss: 0.1221 - acc: 0.9600 - precision_m: 0.9768 - recall_m: 0.9427 - val_loss: 0.1435 - val_acc: 0.9430 - val_precision_m: 0.9834 - val_recall_m: 0.9105\n",
      "Epoch 999/5000\n",
      " - 10s - loss: 0.1258 - acc: 0.9588 - precision_m: 0.9748 - recall_m: 0.9425 - val_loss: 0.1310 - val_acc: 0.9535 - val_precision_m: 0.9785 - val_recall_m: 0.9341\n",
      "Epoch 1000/5000\n",
      " - 10s - loss: 0.1336 - acc: 0.9548 - precision_m: 0.9729 - recall_m: 0.9365 - val_loss: 0.1347 - val_acc: 0.9601 - val_precision_m: 0.9775 - val_recall_m: 0.9477\n",
      "Epoch 1001/5000\n",
      " - 10s - loss: 0.1231 - acc: 0.9603 - precision_m: 0.9764 - recall_m: 0.9435 - val_loss: 0.1363 - val_acc: 0.9612 - val_precision_m: 0.9664 - val_recall_m: 0.9619\n",
      "Epoch 1002/5000\n",
      " - 10s - loss: 0.1201 - acc: 0.9610 - precision_m: 0.9777 - recall_m: 0.9435 - val_loss: 0.1353 - val_acc: 0.9513 - val_precision_m: 0.9826 - val_recall_m: 0.9266\n",
      "Epoch 1003/5000\n",
      " - 10s - loss: 0.1218 - acc: 0.9606 - precision_m: 0.9780 - recall_m: 0.9428 - val_loss: 0.1618 - val_acc: 0.9457 - val_precision_m: 0.9386 - val_recall_m: 0.9635\n",
      "Epoch 1004/5000\n",
      " - 10s - loss: 0.1250 - acc: 0.9601 - precision_m: 0.9763 - recall_m: 0.9433 - val_loss: 0.1333 - val_acc: 0.9618 - val_precision_m: 0.9757 - val_recall_m: 0.9530\n",
      "Epoch 1005/5000\n",
      " - 10s - loss: 0.1281 - acc: 0.9585 - precision_m: 0.9757 - recall_m: 0.9411 - val_loss: 0.1411 - val_acc: 0.9535 - val_precision_m: 0.9587 - val_recall_m: 0.9551\n",
      "Epoch 1006/5000\n",
      " - 10s - loss: 0.1235 - acc: 0.9598 - precision_m: 0.9768 - recall_m: 0.9424 - val_loss: 0.1492 - val_acc: 0.9413 - val_precision_m: 0.9877 - val_recall_m: 0.9033\n",
      "Epoch 1007/5000\n",
      " - 10s - loss: 0.1292 - acc: 0.9567 - precision_m: 0.9757 - recall_m: 0.9367 - val_loss: 0.1464 - val_acc: 0.9551 - val_precision_m: 0.9526 - val_recall_m: 0.9653\n",
      "Epoch 1008/5000\n",
      " - 10s - loss: 0.1234 - acc: 0.9585 - precision_m: 0.9749 - recall_m: 0.9418 - val_loss: 0.1251 - val_acc: 0.9585 - val_precision_m: 0.9767 - val_recall_m: 0.9456\n",
      "Epoch 1009/5000\n",
      " - 10s - loss: 0.1227 - acc: 0.9589 - precision_m: 0.9761 - recall_m: 0.9414 - val_loss: 0.1337 - val_acc: 0.9551 - val_precision_m: 0.9774 - val_recall_m: 0.9389\n",
      "Epoch 1010/5000\n",
      " - 10s - loss: 0.1243 - acc: 0.9585 - precision_m: 0.9751 - recall_m: 0.9412 - val_loss: 0.2013 - val_acc: 0.9330 - val_precision_m: 0.9031 - val_recall_m: 0.9821\n",
      "Epoch 1011/5000\n",
      " - 10s - loss: 0.1248 - acc: 0.9582 - precision_m: 0.9752 - recall_m: 0.9407 - val_loss: 0.1341 - val_acc: 0.9585 - val_precision_m: 0.9795 - val_recall_m: 0.9429\n",
      "Epoch 1012/5000\n",
      " - 10s - loss: 0.1227 - acc: 0.9612 - precision_m: 0.9776 - recall_m: 0.9443 - val_loss: 0.1370 - val_acc: 0.9480 - val_precision_m: 0.9835 - val_recall_m: 0.9194\n",
      "Epoch 1013/5000\n",
      " - 10s - loss: 0.1203 - acc: 0.9594 - precision_m: 0.9763 - recall_m: 0.9420 - val_loss: 0.1333 - val_acc: 0.9529 - val_precision_m: 0.9752 - val_recall_m: 0.9368\n",
      "Epoch 1014/5000\n",
      " - 10s - loss: 0.1213 - acc: 0.9610 - precision_m: 0.9778 - recall_m: 0.9429 - val_loss: 0.1335 - val_acc: 0.9535 - val_precision_m: 0.9824 - val_recall_m: 0.9304\n",
      "Epoch 1015/5000\n",
      " - 10s - loss: 0.1223 - acc: 0.9588 - precision_m: 0.9755 - recall_m: 0.9416 - val_loss: 0.1531 - val_acc: 0.9419 - val_precision_m: 0.9842 - val_recall_m: 0.9072\n",
      "Epoch 1016/5000\n",
      " - 10s - loss: 0.1268 - acc: 0.9583 - precision_m: 0.9760 - recall_m: 0.9402 - val_loss: 0.1491 - val_acc: 0.9391 - val_precision_m: 0.9853 - val_recall_m: 0.9010\n",
      "Epoch 1017/5000\n",
      " - 10s - loss: 0.1236 - acc: 0.9591 - precision_m: 0.9754 - recall_m: 0.9423 - val_loss: 0.1344 - val_acc: 0.9496 - val_precision_m: 0.9855 - val_recall_m: 0.9202\n",
      "Epoch 1018/5000\n",
      " - 10s - loss: 0.1286 - acc: 0.9568 - precision_m: 0.9734 - recall_m: 0.9399 - val_loss: 0.1592 - val_acc: 0.9324 - val_precision_m: 0.9863 - val_recall_m: 0.8876\n",
      "Epoch 1019/5000\n",
      " - 10s - loss: 0.1192 - acc: 0.9610 - precision_m: 0.9772 - recall_m: 0.9445 - val_loss: 0.1297 - val_acc: 0.9623 - val_precision_m: 0.9747 - val_recall_m: 0.9550\n",
      "Epoch 1020/5000\n",
      " - 10s - loss: 0.1172 - acc: 0.9634 - precision_m: 0.9797 - recall_m: 0.9463 - val_loss: 0.1252 - val_acc: 0.9612 - val_precision_m: 0.9756 - val_recall_m: 0.9517\n",
      "Epoch 1021/5000\n",
      " - 10s - loss: 0.1193 - acc: 0.9605 - precision_m: 0.9770 - recall_m: 0.9433 - val_loss: 0.1306 - val_acc: 0.9590 - val_precision_m: 0.9666 - val_recall_m: 0.9570\n",
      "Epoch 1022/5000\n",
      " - 10s - loss: 0.1255 - acc: 0.9575 - precision_m: 0.9744 - recall_m: 0.9409 - val_loss: 0.1441 - val_acc: 0.9574 - val_precision_m: 0.9566 - val_recall_m: 0.9658\n",
      "Epoch 1023/5000\n",
      " - 10s - loss: 0.1245 - acc: 0.9594 - precision_m: 0.9771 - recall_m: 0.9412 - val_loss: 0.1300 - val_acc: 0.9513 - val_precision_m: 0.9815 - val_recall_m: 0.9275\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1024/5000\n",
      " - 10s - loss: 0.1190 - acc: 0.9596 - precision_m: 0.9755 - recall_m: 0.9429 - val_loss: 0.1317 - val_acc: 0.9568 - val_precision_m: 0.9725 - val_recall_m: 0.9466\n",
      "Epoch 1025/5000\n",
      " - 10s - loss: 0.1171 - acc: 0.9613 - precision_m: 0.9790 - recall_m: 0.9429 - val_loss: 0.1327 - val_acc: 0.9579 - val_precision_m: 0.9631 - val_recall_m: 0.9593\n",
      "Epoch 1026/5000\n",
      " - 10s - loss: 0.1229 - acc: 0.9591 - precision_m: 0.9768 - recall_m: 0.9411 - val_loss: 0.1264 - val_acc: 0.9601 - val_precision_m: 0.9718 - val_recall_m: 0.9545\n",
      "Epoch 1027/5000\n",
      " - 10s - loss: 0.1182 - acc: 0.9623 - precision_m: 0.9780 - recall_m: 0.9460 - val_loss: 0.1499 - val_acc: 0.9419 - val_precision_m: 0.9855 - val_recall_m: 0.9060\n",
      "Epoch 1028/5000\n",
      " - 10s - loss: 0.1201 - acc: 0.9601 - precision_m: 0.9772 - recall_m: 0.9428 - val_loss: 0.1323 - val_acc: 0.9485 - val_precision_m: 0.9804 - val_recall_m: 0.9233\n",
      "Epoch 1029/5000\n",
      " - 10s - loss: 0.1188 - acc: 0.9615 - precision_m: 0.9767 - recall_m: 0.9459 - val_loss: 0.1473 - val_acc: 0.9419 - val_precision_m: 0.9866 - val_recall_m: 0.9048\n",
      "Epoch 1030/5000\n",
      " - 10s - loss: 0.1152 - acc: 0.9633 - precision_m: 0.9794 - recall_m: 0.9467 - val_loss: 0.1300 - val_acc: 0.9585 - val_precision_m: 0.9804 - val_recall_m: 0.9412\n",
      "Epoch 1031/5000\n",
      " - 10s - loss: 0.1171 - acc: 0.9620 - precision_m: 0.9774 - recall_m: 0.9460 - val_loss: 0.1309 - val_acc: 0.9551 - val_precision_m: 0.9746 - val_recall_m: 0.9418\n",
      "Epoch 1032/5000\n",
      " - 10s - loss: 0.1221 - acc: 0.9591 - precision_m: 0.9732 - recall_m: 0.9446 - val_loss: 0.1479 - val_acc: 0.9408 - val_precision_m: 0.9833 - val_recall_m: 0.9066\n",
      "Epoch 1033/5000\n",
      " - 10s - loss: 0.1187 - acc: 0.9616 - precision_m: 0.9769 - recall_m: 0.9460 - val_loss: 0.1282 - val_acc: 0.9529 - val_precision_m: 0.9835 - val_recall_m: 0.9287\n",
      "Epoch 1034/5000\n",
      " - 10s - loss: 0.1238 - acc: 0.9580 - precision_m: 0.9733 - recall_m: 0.9424 - val_loss: 0.1283 - val_acc: 0.9651 - val_precision_m: 0.9779 - val_recall_m: 0.9576\n",
      "Epoch 1035/5000\n",
      " - 10s - loss: 0.1157 - acc: 0.9627 - precision_m: 0.9771 - recall_m: 0.9473 - val_loss: 0.1244 - val_acc: 0.9607 - val_precision_m: 0.9817 - val_recall_m: 0.9450\n",
      "Epoch 1036/5000\n",
      " - 10s - loss: 0.1193 - acc: 0.9600 - precision_m: 0.9764 - recall_m: 0.9431 - val_loss: 0.1412 - val_acc: 0.9435 - val_precision_m: 0.9833 - val_recall_m: 0.9113\n",
      "Epoch 1037/5000\n",
      " - 10s - loss: 0.1178 - acc: 0.9626 - precision_m: 0.9775 - recall_m: 0.9472 - val_loss: 0.1313 - val_acc: 0.9557 - val_precision_m: 0.9815 - val_recall_m: 0.9356\n",
      "Epoch 1038/5000\n",
      " - 10s - loss: 0.1279 - acc: 0.9570 - precision_m: 0.9758 - recall_m: 0.9384 - val_loss: 0.1315 - val_acc: 0.9502 - val_precision_m: 0.9823 - val_recall_m: 0.9245\n",
      "Epoch 1039/5000\n",
      " - 10s - loss: 0.1175 - acc: 0.9607 - precision_m: 0.9767 - recall_m: 0.9443 - val_loss: 0.1221 - val_acc: 0.9551 - val_precision_m: 0.9795 - val_recall_m: 0.9368\n",
      "Epoch 1040/5000\n",
      " - 10s - loss: 0.1261 - acc: 0.9570 - precision_m: 0.9735 - recall_m: 0.9412 - val_loss: 0.1446 - val_acc: 0.9408 - val_precision_m: 0.9841 - val_recall_m: 0.9054\n",
      "Epoch 1041/5000\n",
      " - 10s - loss: 0.1153 - acc: 0.9623 - precision_m: 0.9775 - recall_m: 0.9466 - val_loss: 0.1277 - val_acc: 0.9535 - val_precision_m: 0.9744 - val_recall_m: 0.9386\n",
      "Epoch 1042/5000\n",
      " - 10s - loss: 0.1121 - acc: 0.9641 - precision_m: 0.9792 - recall_m: 0.9481 - val_loss: 0.1235 - val_acc: 0.9590 - val_precision_m: 0.9828 - val_recall_m: 0.9408\n",
      "Epoch 1043/5000\n",
      " - 10s - loss: 0.1154 - acc: 0.9626 - precision_m: 0.9779 - recall_m: 0.9465 - val_loss: 0.1297 - val_acc: 0.9496 - val_precision_m: 0.9824 - val_recall_m: 0.9232\n",
      "Epoch 1044/5000\n",
      " - 10s - loss: 0.1098 - acc: 0.9655 - precision_m: 0.9812 - recall_m: 0.9493 - val_loss: 0.1349 - val_acc: 0.9496 - val_precision_m: 0.9802 - val_recall_m: 0.9255\n",
      "Epoch 1045/5000\n",
      " - 10s - loss: 0.1164 - acc: 0.9617 - precision_m: 0.9777 - recall_m: 0.9451 - val_loss: 0.1258 - val_acc: 0.9618 - val_precision_m: 0.9757 - val_recall_m: 0.9531\n",
      "Epoch 1046/5000\n",
      " - 10s - loss: 0.1199 - acc: 0.9609 - precision_m: 0.9764 - recall_m: 0.9448 - val_loss: 0.1204 - val_acc: 0.9585 - val_precision_m: 0.9755 - val_recall_m: 0.9470\n",
      "Epoch 1047/5000\n",
      " - 10s - loss: 0.1119 - acc: 0.9642 - precision_m: 0.9804 - recall_m: 0.9477 - val_loss: 0.1283 - val_acc: 0.9618 - val_precision_m: 0.9727 - val_recall_m: 0.9559\n",
      "Epoch 1048/5000\n",
      " - 10s - loss: 0.1120 - acc: 0.9647 - precision_m: 0.9802 - recall_m: 0.9485 - val_loss: 0.1661 - val_acc: 0.9258 - val_precision_m: 0.9860 - val_recall_m: 0.8754\n",
      "Epoch 1049/5000\n",
      " - 10s - loss: 0.1163 - acc: 0.9609 - precision_m: 0.9773 - recall_m: 0.9441 - val_loss: 0.1345 - val_acc: 0.9446 - val_precision_m: 0.9833 - val_recall_m: 0.9133\n",
      "Epoch 1050/5000\n",
      " - 10s - loss: 0.1179 - acc: 0.9613 - precision_m: 0.9760 - recall_m: 0.9463 - val_loss: 0.1463 - val_acc: 0.9441 - val_precision_m: 0.9788 - val_recall_m: 0.9168\n",
      "Epoch 1051/5000\n",
      " - 10s - loss: 0.1202 - acc: 0.9599 - precision_m: 0.9765 - recall_m: 0.9428 - val_loss: 0.1237 - val_acc: 0.9607 - val_precision_m: 0.9787 - val_recall_m: 0.9479\n",
      "Epoch 1052/5000\n",
      " - 10s - loss: 0.1178 - acc: 0.9602 - precision_m: 0.9772 - recall_m: 0.9431 - val_loss: 0.1306 - val_acc: 0.9596 - val_precision_m: 0.9609 - val_recall_m: 0.9644\n",
      "Epoch 1053/5000\n",
      " - 10s - loss: 0.1137 - acc: 0.9633 - precision_m: 0.9784 - recall_m: 0.9477 - val_loss: 0.1376 - val_acc: 0.9452 - val_precision_m: 0.9845 - val_recall_m: 0.9137\n",
      "Epoch 1054/5000\n",
      " - 10s - loss: 0.1184 - acc: 0.9613 - precision_m: 0.9790 - recall_m: 0.9430 - val_loss: 0.1238 - val_acc: 0.9579 - val_precision_m: 0.9746 - val_recall_m: 0.9466\n",
      "Epoch 1055/5000\n",
      " - 10s - loss: 0.1134 - acc: 0.9623 - precision_m: 0.9761 - recall_m: 0.9475 - val_loss: 0.1246 - val_acc: 0.9574 - val_precision_m: 0.9797 - val_recall_m: 0.9411\n",
      "Epoch 1056/5000\n",
      " - 10s - loss: 0.1129 - acc: 0.9635 - precision_m: 0.9769 - recall_m: 0.9498 - val_loss: 0.1531 - val_acc: 0.9513 - val_precision_m: 0.9459 - val_recall_m: 0.9658\n",
      "Epoch 1057/5000\n",
      " - 10s - loss: 0.1201 - acc: 0.9602 - precision_m: 0.9756 - recall_m: 0.9444 - val_loss: 0.1278 - val_acc: 0.9585 - val_precision_m: 0.9785 - val_recall_m: 0.9438\n",
      "Epoch 1058/5000\n",
      " - 10s - loss: 0.1174 - acc: 0.9606 - precision_m: 0.9760 - recall_m: 0.9449 - val_loss: 0.1400 - val_acc: 0.9463 - val_precision_m: 0.9864 - val_recall_m: 0.9131\n",
      "Epoch 1059/5000\n",
      " - 10s - loss: 0.1107 - acc: 0.9660 - precision_m: 0.9803 - recall_m: 0.9509 - val_loss: 0.1223 - val_acc: 0.9585 - val_precision_m: 0.9776 - val_recall_m: 0.9448\n",
      "Epoch 1060/5000\n",
      " - 10s - loss: 0.1221 - acc: 0.9599 - precision_m: 0.9761 - recall_m: 0.9440 - val_loss: 0.1223 - val_acc: 0.9623 - val_precision_m: 0.9692 - val_recall_m: 0.9614\n",
      "Epoch 1061/5000\n",
      " - 10s - loss: 0.1143 - acc: 0.9621 - precision_m: 0.9775 - recall_m: 0.9467 - val_loss: 0.1352 - val_acc: 0.9441 - val_precision_m: 0.9842 - val_recall_m: 0.9113\n",
      "Epoch 1062/5000\n",
      " - 10s - loss: 0.1159 - acc: 0.9614 - precision_m: 0.9775 - recall_m: 0.9451 - val_loss: 0.1273 - val_acc: 0.9623 - val_precision_m: 0.9721 - val_recall_m: 0.9583\n",
      "Epoch 1063/5000\n",
      " - 10s - loss: 0.1122 - acc: 0.9644 - precision_m: 0.9799 - recall_m: 0.9484 - val_loss: 0.1212 - val_acc: 0.9629 - val_precision_m: 0.9827 - val_recall_m: 0.9480\n",
      "Epoch 1064/5000\n",
      " - 10s - loss: 0.1155 - acc: 0.9623 - precision_m: 0.9771 - recall_m: 0.9467 - val_loss: 0.1292 - val_acc: 0.9662 - val_precision_m: 0.9701 - val_recall_m: 0.9677\n",
      "Epoch 1065/5000\n",
      " - 10s - loss: 0.1115 - acc: 0.9633 - precision_m: 0.9787 - recall_m: 0.9472 - val_loss: 0.1217 - val_acc: 0.9635 - val_precision_m: 0.9796 - val_recall_m: 0.9520\n",
      "Epoch 1066/5000\n",
      " - 10s - loss: 0.1149 - acc: 0.9616 - precision_m: 0.9765 - recall_m: 0.9464 - val_loss: 0.1260 - val_acc: 0.9513 - val_precision_m: 0.9813 - val_recall_m: 0.9275\n",
      "Epoch 1067/5000\n",
      " - 10s - loss: 0.1152 - acc: 0.9613 - precision_m: 0.9762 - recall_m: 0.9460 - val_loss: 0.1356 - val_acc: 0.9468 - val_precision_m: 0.9832 - val_recall_m: 0.9174\n",
      "Epoch 1068/5000\n",
      " - 10s - loss: 0.1163 - acc: 0.9627 - precision_m: 0.9772 - recall_m: 0.9475 - val_loss: 0.1195 - val_acc: 0.9657 - val_precision_m: 0.9758 - val_recall_m: 0.9604\n",
      "Epoch 1069/5000\n",
      " - 10s - loss: 0.1106 - acc: 0.9625 - precision_m: 0.9780 - recall_m: 0.9465 - val_loss: 0.1211 - val_acc: 0.9623 - val_precision_m: 0.9659 - val_recall_m: 0.9642\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1070/5000\n",
      " - 10s - loss: 0.1129 - acc: 0.9618 - precision_m: 0.9775 - recall_m: 0.9450 - val_loss: 0.1436 - val_acc: 0.9441 - val_precision_m: 0.9876 - val_recall_m: 0.9079\n",
      "Epoch 1071/5000\n",
      " - 10s - loss: 0.1133 - acc: 0.9633 - precision_m: 0.9793 - recall_m: 0.9476 - val_loss: 0.1216 - val_acc: 0.9646 - val_precision_m: 0.9778 - val_recall_m: 0.9560\n",
      "Epoch 1072/5000\n",
      " - 10s - loss: 0.1126 - acc: 0.9637 - precision_m: 0.9783 - recall_m: 0.9482 - val_loss: 0.1234 - val_acc: 0.9563 - val_precision_m: 0.9777 - val_recall_m: 0.9407\n",
      "Epoch 1073/5000\n",
      " - 10s - loss: 0.1107 - acc: 0.9651 - precision_m: 0.9792 - recall_m: 0.9502 - val_loss: 0.1233 - val_acc: 0.9590 - val_precision_m: 0.9815 - val_recall_m: 0.9414\n",
      "Epoch 1074/5000\n",
      " - 10s - loss: 0.1099 - acc: 0.9647 - precision_m: 0.9793 - recall_m: 0.9500 - val_loss: 0.1266 - val_acc: 0.9540 - val_precision_m: 0.9726 - val_recall_m: 0.9416\n",
      "Epoch 1075/5000\n",
      " - 10s - loss: 0.1112 - acc: 0.9646 - precision_m: 0.9797 - recall_m: 0.9492 - val_loss: 0.1219 - val_acc: 0.9585 - val_precision_m: 0.9808 - val_recall_m: 0.9417\n",
      "Epoch 1076/5000\n",
      " - 10s - loss: 0.1079 - acc: 0.9654 - precision_m: 0.9802 - recall_m: 0.9503 - val_loss: 0.1264 - val_acc: 0.9623 - val_precision_m: 0.9623 - val_recall_m: 0.9684\n",
      "Epoch 1077/5000\n",
      " - 10s - loss: 0.1085 - acc: 0.9644 - precision_m: 0.9777 - recall_m: 0.9507 - val_loss: 0.1291 - val_acc: 0.9485 - val_precision_m: 0.9825 - val_recall_m: 0.9216\n",
      "Epoch 1078/5000\n",
      " - 10s - loss: 0.1111 - acc: 0.9633 - precision_m: 0.9786 - recall_m: 0.9476 - val_loss: 0.1872 - val_acc: 0.9147 - val_precision_m: 0.9892 - val_recall_m: 0.8523\n",
      "Epoch 1079/5000\n",
      " - 10s - loss: 0.1189 - acc: 0.9602 - precision_m: 0.9755 - recall_m: 0.9446 - val_loss: 0.1204 - val_acc: 0.9651 - val_precision_m: 0.9738 - val_recall_m: 0.9611\n",
      "Epoch 1080/5000\n",
      " - 10s - loss: 0.1119 - acc: 0.9619 - precision_m: 0.9789 - recall_m: 0.9447 - val_loss: 0.1204 - val_acc: 0.9635 - val_precision_m: 0.9699 - val_recall_m: 0.9621\n",
      "Epoch 1081/5000\n",
      " - 10s - loss: 0.1092 - acc: 0.9656 - precision_m: 0.9795 - recall_m: 0.9512 - val_loss: 0.1244 - val_acc: 0.9551 - val_precision_m: 0.9804 - val_recall_m: 0.9353\n",
      "Epoch 1082/5000\n",
      " - 10s - loss: 0.1101 - acc: 0.9632 - precision_m: 0.9787 - recall_m: 0.9471 - val_loss: 0.1492 - val_acc: 0.9385 - val_precision_m: 0.9852 - val_recall_m: 0.8999\n",
      "Epoch 1083/5000\n",
      " - 10s - loss: 0.1145 - acc: 0.9629 - precision_m: 0.9778 - recall_m: 0.9481 - val_loss: 0.1200 - val_acc: 0.9596 - val_precision_m: 0.9817 - val_recall_m: 0.9431\n",
      "Epoch 1084/5000\n",
      " - 10s - loss: 0.1051 - acc: 0.9668 - precision_m: 0.9797 - recall_m: 0.9532 - val_loss: 0.1207 - val_acc: 0.9640 - val_precision_m: 0.9710 - val_recall_m: 0.9624\n",
      "Epoch 1085/5000\n",
      " - 10s - loss: 0.1080 - acc: 0.9646 - precision_m: 0.9786 - recall_m: 0.9505 - val_loss: 0.1533 - val_acc: 0.9369 - val_precision_m: 0.9896 - val_recall_m: 0.8932\n",
      "Epoch 1086/5000\n",
      " - 10s - loss: 0.1144 - acc: 0.9605 - precision_m: 0.9755 - recall_m: 0.9452 - val_loss: 0.1212 - val_acc: 0.9551 - val_precision_m: 0.9763 - val_recall_m: 0.9397\n",
      "Epoch 1087/5000\n",
      " - 10s - loss: 0.1144 - acc: 0.9623 - precision_m: 0.9779 - recall_m: 0.9462 - val_loss: 0.1190 - val_acc: 0.9612 - val_precision_m: 0.9775 - val_recall_m: 0.9506\n",
      "Epoch 1088/5000\n",
      " - 10s - loss: 0.1095 - acc: 0.9630 - precision_m: 0.9786 - recall_m: 0.9472 - val_loss: 0.1235 - val_acc: 0.9629 - val_precision_m: 0.9681 - val_recall_m: 0.9635\n",
      "Epoch 1089/5000\n",
      " - 10s - loss: 0.1097 - acc: 0.9626 - precision_m: 0.9783 - recall_m: 0.9466 - val_loss: 0.1303 - val_acc: 0.9618 - val_precision_m: 0.9615 - val_recall_m: 0.9689\n",
      "Epoch 1090/5000\n",
      " - 10s - loss: 0.1087 - acc: 0.9651 - precision_m: 0.9780 - recall_m: 0.9517 - val_loss: 0.1225 - val_acc: 0.9507 - val_precision_m: 0.9834 - val_recall_m: 0.9245\n",
      "Epoch 1091/5000\n",
      " - 10s - loss: 0.1105 - acc: 0.9639 - precision_m: 0.9795 - recall_m: 0.9473 - val_loss: 0.1164 - val_acc: 0.9640 - val_precision_m: 0.9787 - val_recall_m: 0.9543\n",
      "Epoch 1092/5000\n",
      " - 10s - loss: 0.1086 - acc: 0.9636 - precision_m: 0.9778 - recall_m: 0.9490 - val_loss: 0.1297 - val_acc: 0.9485 - val_precision_m: 0.9834 - val_recall_m: 0.9202\n",
      "Epoch 1093/5000\n",
      " - 10s - loss: 0.1132 - acc: 0.9633 - precision_m: 0.9780 - recall_m: 0.9483 - val_loss: 0.1343 - val_acc: 0.9551 - val_precision_m: 0.9649 - val_recall_m: 0.9521\n",
      "Epoch 1094/5000\n",
      " - 10s - loss: 0.1149 - acc: 0.9606 - precision_m: 0.9757 - recall_m: 0.9446 - val_loss: 0.1284 - val_acc: 0.9612 - val_precision_m: 0.9597 - val_recall_m: 0.9698\n",
      "Epoch 1095/5000\n",
      " - 10s - loss: 0.1115 - acc: 0.9628 - precision_m: 0.9766 - recall_m: 0.9481 - val_loss: 0.1170 - val_acc: 0.9635 - val_precision_m: 0.9757 - val_recall_m: 0.9560\n",
      "Epoch 1096/5000\n",
      " - 10s - loss: 0.1089 - acc: 0.9651 - precision_m: 0.9803 - recall_m: 0.9500 - val_loss: 0.1190 - val_acc: 0.9574 - val_precision_m: 0.9818 - val_recall_m: 0.9384\n",
      "Epoch 1097/5000\n",
      " - 10s - loss: 0.1057 - acc: 0.9652 - precision_m: 0.9789 - recall_m: 0.9510 - val_loss: 0.1189 - val_acc: 0.9673 - val_precision_m: 0.9779 - val_recall_m: 0.9610\n",
      "Epoch 1098/5000\n",
      " - 10s - loss: 0.1062 - acc: 0.9654 - precision_m: 0.9812 - recall_m: 0.9496 - val_loss: 0.1153 - val_acc: 0.9623 - val_precision_m: 0.9808 - val_recall_m: 0.9492\n",
      "Epoch 1099/5000\n",
      " - 10s - loss: 0.1067 - acc: 0.9654 - precision_m: 0.9797 - recall_m: 0.9504 - val_loss: 0.1154 - val_acc: 0.9635 - val_precision_m: 0.9807 - val_recall_m: 0.9509\n",
      "Epoch 1100/5000\n",
      " - 10s - loss: 0.1141 - acc: 0.9611 - precision_m: 0.9763 - recall_m: 0.9461 - val_loss: 0.1244 - val_acc: 0.9635 - val_precision_m: 0.9613 - val_recall_m: 0.9717\n",
      "Epoch 1101/5000\n",
      " - 10s - loss: 0.1062 - acc: 0.9655 - precision_m: 0.9792 - recall_m: 0.9516 - val_loss: 0.1345 - val_acc: 0.9640 - val_precision_m: 0.9596 - val_recall_m: 0.9752\n",
      "Epoch 1102/5000\n",
      " - 10s - loss: 0.1126 - acc: 0.9629 - precision_m: 0.9784 - recall_m: 0.9477 - val_loss: 0.1119 - val_acc: 0.9662 - val_precision_m: 0.9801 - val_recall_m: 0.9569\n",
      "Epoch 1103/5000\n",
      " - 10s - loss: 0.1086 - acc: 0.9654 - precision_m: 0.9799 - recall_m: 0.9508 - val_loss: 0.1327 - val_acc: 0.9485 - val_precision_m: 0.9857 - val_recall_m: 0.9188\n",
      "Epoch 1104/5000\n",
      " - 10s - loss: 0.1073 - acc: 0.9645 - precision_m: 0.9791 - recall_m: 0.9498 - val_loss: 0.1172 - val_acc: 0.9607 - val_precision_m: 0.9827 - val_recall_m: 0.9441\n",
      "Epoch 1105/5000\n",
      " - 10s - loss: 0.1058 - acc: 0.9654 - precision_m: 0.9805 - recall_m: 0.9495 - val_loss: 0.1360 - val_acc: 0.9551 - val_precision_m: 0.9506 - val_recall_m: 0.9677\n",
      "Epoch 1106/5000\n",
      " - 10s - loss: 0.1095 - acc: 0.9636 - precision_m: 0.9776 - recall_m: 0.9491 - val_loss: 0.1645 - val_acc: 0.9297 - val_precision_m: 0.9883 - val_recall_m: 0.8803\n",
      "Epoch 1107/5000\n",
      " - 10s - loss: 0.1052 - acc: 0.9653 - precision_m: 0.9793 - recall_m: 0.9511 - val_loss: 0.1178 - val_acc: 0.9657 - val_precision_m: 0.9770 - val_recall_m: 0.9592\n",
      "Epoch 1108/5000\n",
      " - 10s - loss: 0.1048 - acc: 0.9654 - precision_m: 0.9797 - recall_m: 0.9506 - val_loss: 0.1170 - val_acc: 0.9618 - val_precision_m: 0.9786 - val_recall_m: 0.9499\n",
      "Epoch 1109/5000\n",
      " - 10s - loss: 0.1043 - acc: 0.9661 - precision_m: 0.9796 - recall_m: 0.9519 - val_loss: 0.1139 - val_acc: 0.9651 - val_precision_m: 0.9809 - val_recall_m: 0.9545\n",
      "Epoch 1110/5000\n",
      " - 10s - loss: 0.1017 - acc: 0.9667 - precision_m: 0.9801 - recall_m: 0.9530 - val_loss: 0.1158 - val_acc: 0.9568 - val_precision_m: 0.9817 - val_recall_m: 0.9377\n",
      "Epoch 1111/5000\n",
      " - 10s - loss: 0.1021 - acc: 0.9661 - precision_m: 0.9805 - recall_m: 0.9513 - val_loss: 0.1184 - val_acc: 0.9651 - val_precision_m: 0.9739 - val_recall_m: 0.9613\n",
      "Epoch 1112/5000\n",
      " - 10s - loss: 0.1030 - acc: 0.9677 - precision_m: 0.9823 - recall_m: 0.9526 - val_loss: 0.1145 - val_acc: 0.9612 - val_precision_m: 0.9766 - val_recall_m: 0.9510\n",
      "Epoch 1113/5000\n",
      " - 10s - loss: 0.1108 - acc: 0.9625 - precision_m: 0.9750 - recall_m: 0.9497 - val_loss: 0.1117 - val_acc: 0.9646 - val_precision_m: 0.9711 - val_recall_m: 0.9637\n",
      "Epoch 1114/5000\n",
      " - 10s - loss: 0.1055 - acc: 0.9655 - precision_m: 0.9800 - recall_m: 0.9500 - val_loss: 0.1321 - val_acc: 0.9446 - val_precision_m: 0.9836 - val_recall_m: 0.9134\n",
      "Epoch 1115/5000\n",
      " - 10s - loss: 0.1077 - acc: 0.9641 - precision_m: 0.9783 - recall_m: 0.9497 - val_loss: 0.1090 - val_acc: 0.9657 - val_precision_m: 0.9790 - val_recall_m: 0.9575\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1116/5000\n",
      " - 10s - loss: 0.1037 - acc: 0.9667 - precision_m: 0.9779 - recall_m: 0.9549 - val_loss: 0.1182 - val_acc: 0.9585 - val_precision_m: 0.9768 - val_recall_m: 0.9464\n",
      "Epoch 1117/5000\n",
      " - 10s - loss: 0.1015 - acc: 0.9662 - precision_m: 0.9809 - recall_m: 0.9509 - val_loss: 0.1285 - val_acc: 0.9513 - val_precision_m: 0.9833 - val_recall_m: 0.9253\n",
      "Epoch 1118/5000\n",
      " - 10s - loss: 0.1059 - acc: 0.9645 - precision_m: 0.9795 - recall_m: 0.9488 - val_loss: 0.1282 - val_acc: 0.9646 - val_precision_m: 0.9621 - val_recall_m: 0.9726\n",
      "Epoch 1119/5000\n",
      " - 10s - loss: 0.1070 - acc: 0.9656 - precision_m: 0.9782 - recall_m: 0.9524 - val_loss: 0.1131 - val_acc: 0.9612 - val_precision_m: 0.9768 - val_recall_m: 0.9511\n",
      "Epoch 1120/5000\n",
      " - 10s - loss: 0.1052 - acc: 0.9655 - precision_m: 0.9807 - recall_m: 0.9504 - val_loss: 0.1219 - val_acc: 0.9579 - val_precision_m: 0.9836 - val_recall_m: 0.9376\n",
      "Epoch 1121/5000\n",
      " - 10s - loss: 0.1037 - acc: 0.9649 - precision_m: 0.9780 - recall_m: 0.9517 - val_loss: 0.1120 - val_acc: 0.9623 - val_precision_m: 0.9780 - val_recall_m: 0.9519\n",
      "Epoch 1122/5000\n",
      " - 10s - loss: 0.1072 - acc: 0.9645 - precision_m: 0.9781 - recall_m: 0.9506 - val_loss: 0.1370 - val_acc: 0.9413 - val_precision_m: 0.9854 - val_recall_m: 0.9048\n",
      "Epoch 1123/5000\n",
      " - 10s - loss: 0.1134 - acc: 0.9622 - precision_m: 0.9774 - recall_m: 0.9468 - val_loss: 0.1197 - val_acc: 0.9623 - val_precision_m: 0.9681 - val_recall_m: 0.9623\n",
      "Epoch 1124/5000\n",
      " - 10s - loss: 0.1022 - acc: 0.9676 - precision_m: 0.9814 - recall_m: 0.9535 - val_loss: 0.1243 - val_acc: 0.9640 - val_precision_m: 0.9578 - val_recall_m: 0.9770\n",
      "Epoch 1125/5000\n",
      " - 10s - loss: 0.1010 - acc: 0.9668 - precision_m: 0.9808 - recall_m: 0.9526 - val_loss: 0.1178 - val_acc: 0.9618 - val_precision_m: 0.9766 - val_recall_m: 0.9520\n",
      "Epoch 1126/5000\n",
      " - 10s - loss: 0.1016 - acc: 0.9666 - precision_m: 0.9799 - recall_m: 0.9528 - val_loss: 0.1372 - val_acc: 0.9574 - val_precision_m: 0.9457 - val_recall_m: 0.9781\n",
      "Epoch 1127/5000\n",
      " - 10s - loss: 0.1048 - acc: 0.9665 - precision_m: 0.9800 - recall_m: 0.9527 - val_loss: 0.1295 - val_acc: 0.9563 - val_precision_m: 0.9605 - val_recall_m: 0.9586\n",
      "Epoch 1128/5000\n",
      " - 10s - loss: 0.1051 - acc: 0.9659 - precision_m: 0.9802 - recall_m: 0.9515 - val_loss: 0.1165 - val_acc: 0.9635 - val_precision_m: 0.9807 - val_recall_m: 0.9509\n",
      "Epoch 1129/5000\n",
      " - 10s - loss: 0.1030 - acc: 0.9669 - precision_m: 0.9807 - recall_m: 0.9528 - val_loss: 0.1117 - val_acc: 0.9635 - val_precision_m: 0.9820 - val_recall_m: 0.9500\n",
      "Epoch 1130/5000\n",
      " - 10s - loss: 0.1014 - acc: 0.9664 - precision_m: 0.9793 - recall_m: 0.9532 - val_loss: 0.1268 - val_acc: 0.9524 - val_precision_m: 0.9835 - val_recall_m: 0.9275\n",
      "Epoch 1131/5000\n",
      " - 10s - loss: 0.1052 - acc: 0.9649 - precision_m: 0.9797 - recall_m: 0.9504 - val_loss: 0.1148 - val_acc: 0.9651 - val_precision_m: 0.9769 - val_recall_m: 0.9586\n",
      "Epoch 1132/5000\n",
      " - 10s - loss: 0.1046 - acc: 0.9642 - precision_m: 0.9784 - recall_m: 0.9500 - val_loss: 0.1652 - val_acc: 0.9468 - val_precision_m: 0.9209 - val_recall_m: 0.9870\n",
      "Epoch 1133/5000\n",
      " - 10s - loss: 0.1057 - acc: 0.9661 - precision_m: 0.9803 - recall_m: 0.9522 - val_loss: 0.1160 - val_acc: 0.9684 - val_precision_m: 0.9742 - val_recall_m: 0.9678\n",
      "Epoch 1134/5000\n",
      " - 10s - loss: 0.1033 - acc: 0.9658 - precision_m: 0.9791 - recall_m: 0.9524 - val_loss: 0.1157 - val_acc: 0.9574 - val_precision_m: 0.9847 - val_recall_m: 0.9358\n",
      "Epoch 1135/5000\n",
      " - 10s - loss: 0.0984 - acc: 0.9689 - precision_m: 0.9818 - recall_m: 0.9556 - val_loss: 0.1146 - val_acc: 0.9623 - val_precision_m: 0.9806 - val_recall_m: 0.9492\n",
      "Epoch 1136/5000\n",
      " - 10s - loss: 0.0985 - acc: 0.9693 - precision_m: 0.9811 - recall_m: 0.9571 - val_loss: 0.1082 - val_acc: 0.9640 - val_precision_m: 0.9819 - val_recall_m: 0.9511\n",
      "Epoch 1137/5000\n",
      " - 10s - loss: 0.1054 - acc: 0.9653 - precision_m: 0.9781 - recall_m: 0.9523 - val_loss: 0.1104 - val_acc: 0.9601 - val_precision_m: 0.9797 - val_recall_m: 0.9459\n",
      "Epoch 1138/5000\n",
      " - 10s - loss: 0.1087 - acc: 0.9628 - precision_m: 0.9752 - recall_m: 0.9506 - val_loss: 0.1141 - val_acc: 0.9574 - val_precision_m: 0.9732 - val_recall_m: 0.9474\n",
      "Epoch 1139/5000\n",
      " - 10s - loss: 0.1039 - acc: 0.9663 - precision_m: 0.9786 - recall_m: 0.9535 - val_loss: 0.1210 - val_acc: 0.9518 - val_precision_m: 0.9835 - val_recall_m: 0.9267\n",
      "Epoch 1140/5000\n",
      " - 10s - loss: 0.0996 - acc: 0.9678 - precision_m: 0.9803 - recall_m: 0.9547 - val_loss: 0.1102 - val_acc: 0.9673 - val_precision_m: 0.9799 - val_recall_m: 0.9594\n",
      "Epoch 1141/5000\n",
      " - 10s - loss: 0.1027 - acc: 0.9662 - precision_m: 0.9805 - recall_m: 0.9516 - val_loss: 0.1201 - val_acc: 0.9684 - val_precision_m: 0.9662 - val_recall_m: 0.9764\n",
      "Epoch 1142/5000\n",
      " - 10s - loss: 0.1026 - acc: 0.9668 - precision_m: 0.9794 - recall_m: 0.9540 - val_loss: 0.1138 - val_acc: 0.9712 - val_precision_m: 0.9715 - val_recall_m: 0.9759\n",
      "Epoch 1143/5000\n",
      " - 10s - loss: 0.0988 - acc: 0.9691 - precision_m: 0.9829 - recall_m: 0.9544 - val_loss: 0.1139 - val_acc: 0.9579 - val_precision_m: 0.9848 - val_recall_m: 0.9368\n",
      "Epoch 1144/5000\n",
      " - 10s - loss: 0.1008 - acc: 0.9661 - precision_m: 0.9803 - recall_m: 0.9520 - val_loss: 0.1103 - val_acc: 0.9612 - val_precision_m: 0.9788 - val_recall_m: 0.9494\n",
      "Epoch 1145/5000\n",
      " - 10s - loss: 0.0978 - acc: 0.9671 - precision_m: 0.9809 - recall_m: 0.9527 - val_loss: 0.1131 - val_acc: 0.9684 - val_precision_m: 0.9656 - val_recall_m: 0.9767\n",
      "Epoch 1146/5000\n",
      " - 10s - loss: 0.1013 - acc: 0.9671 - precision_m: 0.9805 - recall_m: 0.9531 - val_loss: 0.1162 - val_acc: 0.9612 - val_precision_m: 0.9757 - val_recall_m: 0.9522\n",
      "Epoch 1147/5000\n",
      " - 10s - loss: 0.1010 - acc: 0.9658 - precision_m: 0.9787 - recall_m: 0.9526 - val_loss: 0.1179 - val_acc: 0.9607 - val_precision_m: 0.9840 - val_recall_m: 0.9430\n",
      "Epoch 1148/5000\n",
      " - 10s - loss: 0.0972 - acc: 0.9692 - precision_m: 0.9817 - recall_m: 0.9566 - val_loss: 0.1307 - val_acc: 0.9601 - val_precision_m: 0.9568 - val_recall_m: 0.9709\n",
      "Epoch 1149/5000\n",
      " - 10s - loss: 0.1078 - acc: 0.9642 - precision_m: 0.9789 - recall_m: 0.9495 - val_loss: 0.1110 - val_acc: 0.9679 - val_precision_m: 0.9748 - val_recall_m: 0.9656\n",
      "Epoch 1150/5000\n",
      " - 10s - loss: 0.0967 - acc: 0.9698 - precision_m: 0.9817 - recall_m: 0.9569 - val_loss: 0.1047 - val_acc: 0.9640 - val_precision_m: 0.9795 - val_recall_m: 0.9531\n",
      "Epoch 1151/5000\n",
      " - 10s - loss: 0.0994 - acc: 0.9669 - precision_m: 0.9803 - recall_m: 0.9532 - val_loss: 0.1183 - val_acc: 0.9557 - val_precision_m: 0.9869 - val_recall_m: 0.9312\n",
      "Epoch 1152/5000\n",
      " - 10s - loss: 0.1084 - acc: 0.9620 - precision_m: 0.9752 - recall_m: 0.9490 - val_loss: 0.1325 - val_acc: 0.9568 - val_precision_m: 0.9482 - val_recall_m: 0.9741\n",
      "Epoch 1153/5000\n",
      " - 10s - loss: 0.0988 - acc: 0.9684 - precision_m: 0.9806 - recall_m: 0.9564 - val_loss: 0.1136 - val_acc: 0.9612 - val_precision_m: 0.9837 - val_recall_m: 0.9436\n",
      "Epoch 1154/5000\n",
      " - 10s - loss: 0.0978 - acc: 0.9690 - precision_m: 0.9823 - recall_m: 0.9555 - val_loss: 0.1317 - val_acc: 0.9618 - val_precision_m: 0.9486 - val_recall_m: 0.9831\n",
      "Epoch 1155/5000\n",
      " - 10s - loss: 0.0985 - acc: 0.9677 - precision_m: 0.9796 - recall_m: 0.9552 - val_loss: 0.1316 - val_acc: 0.9485 - val_precision_m: 0.9888 - val_recall_m: 0.9151\n",
      "Epoch 1156/5000\n",
      " - 10s - loss: 0.0964 - acc: 0.9671 - precision_m: 0.9803 - recall_m: 0.9535 - val_loss: 0.1366 - val_acc: 0.9585 - val_precision_m: 0.9465 - val_recall_m: 0.9790\n",
      "Epoch 1157/5000\n",
      " - 10s - loss: 0.0969 - acc: 0.9687 - precision_m: 0.9809 - recall_m: 0.9561 - val_loss: 0.1096 - val_acc: 0.9635 - val_precision_m: 0.9757 - val_recall_m: 0.9562\n",
      "Epoch 1158/5000\n",
      " - 10s - loss: 0.1006 - acc: 0.9660 - precision_m: 0.9796 - recall_m: 0.9520 - val_loss: 0.1094 - val_acc: 0.9679 - val_precision_m: 0.9721 - val_recall_m: 0.9687\n",
      "Epoch 1159/5000\n",
      " - 10s - loss: 0.1029 - acc: 0.9663 - precision_m: 0.9786 - recall_m: 0.9539 - val_loss: 0.1098 - val_acc: 0.9551 - val_precision_m: 0.9804 - val_recall_m: 0.9363\n",
      "Epoch 1160/5000\n",
      " - 10s - loss: 0.1003 - acc: 0.9668 - precision_m: 0.9791 - recall_m: 0.9540 - val_loss: 0.1110 - val_acc: 0.9596 - val_precision_m: 0.9837 - val_recall_m: 0.9408\n",
      "Epoch 1161/5000\n",
      " - 10s - loss: 0.0988 - acc: 0.9661 - precision_m: 0.9779 - recall_m: 0.9541 - val_loss: 0.1101 - val_acc: 0.9673 - val_precision_m: 0.9711 - val_recall_m: 0.9688\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1162/5000\n",
      " - 10s - loss: 0.1004 - acc: 0.9669 - precision_m: 0.9795 - recall_m: 0.9542 - val_loss: 0.1158 - val_acc: 0.9635 - val_precision_m: 0.9726 - val_recall_m: 0.9596\n",
      "Epoch 1163/5000\n",
      " - 10s - loss: 0.1030 - acc: 0.9664 - precision_m: 0.9794 - recall_m: 0.9536 - val_loss: 0.1040 - val_acc: 0.9651 - val_precision_m: 0.9789 - val_recall_m: 0.9560\n",
      "Epoch 1164/5000\n",
      " - 10s - loss: 0.1070 - acc: 0.9629 - precision_m: 0.9774 - recall_m: 0.9482 - val_loss: 0.1155 - val_acc: 0.9662 - val_precision_m: 0.9695 - val_recall_m: 0.9687\n",
      "Epoch 1165/5000\n",
      " - 10s - loss: 0.0980 - acc: 0.9668 - precision_m: 0.9792 - recall_m: 0.9538 - val_loss: 0.1151 - val_acc: 0.9646 - val_precision_m: 0.9672 - val_recall_m: 0.9677\n",
      "Epoch 1166/5000\n",
      " - 10s - loss: 0.1036 - acc: 0.9650 - precision_m: 0.9789 - recall_m: 0.9513 - val_loss: 0.1049 - val_acc: 0.9679 - val_precision_m: 0.9731 - val_recall_m: 0.9678\n",
      "Epoch 1167/5000\n",
      " - 10s - loss: 0.0949 - acc: 0.9697 - precision_m: 0.9832 - recall_m: 0.9559 - val_loss: 0.1101 - val_acc: 0.9640 - val_precision_m: 0.9827 - val_recall_m: 0.9500\n",
      "Epoch 1168/5000\n",
      " - 10s - loss: 0.0932 - acc: 0.9697 - precision_m: 0.9813 - recall_m: 0.9578 - val_loss: 0.1101 - val_acc: 0.9568 - val_precision_m: 0.9816 - val_recall_m: 0.9377\n",
      "Epoch 1169/5000\n",
      " - 10s - loss: 0.0956 - acc: 0.9686 - precision_m: 0.9814 - recall_m: 0.9553 - val_loss: 0.1144 - val_acc: 0.9574 - val_precision_m: 0.9807 - val_recall_m: 0.9400\n",
      "Epoch 1170/5000\n",
      " - 10s - loss: 0.0969 - acc: 0.9683 - precision_m: 0.9815 - recall_m: 0.9545 - val_loss: 0.1072 - val_acc: 0.9668 - val_precision_m: 0.9762 - val_recall_m: 0.9623\n",
      "Epoch 1171/5000\n",
      " - 10s - loss: 0.0999 - acc: 0.9672 - precision_m: 0.9799 - recall_m: 0.9547 - val_loss: 0.1092 - val_acc: 0.9651 - val_precision_m: 0.9788 - val_recall_m: 0.9564\n",
      "Epoch 1172/5000\n",
      " - 10s - loss: 0.0945 - acc: 0.9694 - precision_m: 0.9810 - recall_m: 0.9572 - val_loss: 0.1086 - val_acc: 0.9623 - val_precision_m: 0.9756 - val_recall_m: 0.9540\n",
      "Epoch 1173/5000\n",
      " - 10s - loss: 0.0942 - acc: 0.9698 - precision_m: 0.9807 - recall_m: 0.9589 - val_loss: 0.1225 - val_acc: 0.9524 - val_precision_m: 0.9825 - val_recall_m: 0.9283\n",
      "Epoch 1174/5000\n",
      " - 10s - loss: 0.0977 - acc: 0.9679 - precision_m: 0.9791 - recall_m: 0.9565 - val_loss: 0.1263 - val_acc: 0.9441 - val_precision_m: 0.9855 - val_recall_m: 0.9099\n",
      "Epoch 1175/5000\n",
      " - 10s - loss: 0.0937 - acc: 0.9697 - precision_m: 0.9813 - recall_m: 0.9578 - val_loss: 0.1092 - val_acc: 0.9701 - val_precision_m: 0.9733 - val_recall_m: 0.9716\n",
      "Epoch 1176/5000\n",
      " - 10s - loss: 0.0974 - acc: 0.9679 - precision_m: 0.9800 - recall_m: 0.9551 - val_loss: 0.1066 - val_acc: 0.9690 - val_precision_m: 0.9722 - val_recall_m: 0.9706\n",
      "Epoch 1177/5000\n",
      " - 10s - loss: 0.0960 - acc: 0.9681 - precision_m: 0.9796 - recall_m: 0.9567 - val_loss: 0.1053 - val_acc: 0.9662 - val_precision_m: 0.9757 - val_recall_m: 0.9613\n",
      "Epoch 1178/5000\n",
      " - 10s - loss: 0.0953 - acc: 0.9698 - precision_m: 0.9824 - recall_m: 0.9574 - val_loss: 0.1103 - val_acc: 0.9601 - val_precision_m: 0.9837 - val_recall_m: 0.9418\n",
      "Epoch 1179/5000\n",
      " - 10s - loss: 0.0980 - acc: 0.9689 - precision_m: 0.9807 - recall_m: 0.9567 - val_loss: 0.1039 - val_acc: 0.9646 - val_precision_m: 0.9828 - val_recall_m: 0.9511\n",
      "Epoch 1180/5000\n",
      " - 10s - loss: 0.0982 - acc: 0.9679 - precision_m: 0.9801 - recall_m: 0.9550 - val_loss: 0.1155 - val_acc: 0.9640 - val_precision_m: 0.9587 - val_recall_m: 0.9763\n",
      "Epoch 1181/5000\n",
      " - 10s - loss: 0.1030 - acc: 0.9642 - precision_m: 0.9769 - recall_m: 0.9517 - val_loss: 0.1153 - val_acc: 0.9657 - val_precision_m: 0.9750 - val_recall_m: 0.9617\n",
      "Epoch 1182/5000\n",
      " - 10s - loss: 0.0924 - acc: 0.9698 - precision_m: 0.9814 - recall_m: 0.9581 - val_loss: 0.1044 - val_acc: 0.9701 - val_precision_m: 0.9739 - val_recall_m: 0.9707\n",
      "Epoch 1183/5000\n",
      " - 10s - loss: 0.0978 - acc: 0.9684 - precision_m: 0.9813 - recall_m: 0.9556 - val_loss: 0.1103 - val_acc: 0.9640 - val_precision_m: 0.9820 - val_recall_m: 0.9511\n",
      "Epoch 1184/5000\n",
      " - 10s - loss: 0.0976 - acc: 0.9674 - precision_m: 0.9783 - recall_m: 0.9559 - val_loss: 0.1067 - val_acc: 0.9601 - val_precision_m: 0.9838 - val_recall_m: 0.9419\n",
      "Epoch 1185/5000\n",
      " - 10s - loss: 0.0960 - acc: 0.9690 - precision_m: 0.9809 - recall_m: 0.9565 - val_loss: 0.1034 - val_acc: 0.9673 - val_precision_m: 0.9790 - val_recall_m: 0.9599\n",
      "Epoch 1186/5000\n",
      " - 10s - loss: 0.0902 - acc: 0.9713 - precision_m: 0.9825 - recall_m: 0.9594 - val_loss: 0.1127 - val_acc: 0.9596 - val_precision_m: 0.9816 - val_recall_m: 0.9428\n",
      "Epoch 1187/5000\n",
      " - 10s - loss: 0.0952 - acc: 0.9697 - precision_m: 0.9818 - recall_m: 0.9573 - val_loss: 0.1044 - val_acc: 0.9679 - val_precision_m: 0.9768 - val_recall_m: 0.9635\n",
      "Epoch 1188/5000\n",
      " - 10s - loss: 0.0936 - acc: 0.9689 - precision_m: 0.9806 - recall_m: 0.9569 - val_loss: 0.1194 - val_acc: 0.9507 - val_precision_m: 0.9868 - val_recall_m: 0.9216\n",
      "Epoch 1189/5000\n",
      " - 10s - loss: 0.0955 - acc: 0.9675 - precision_m: 0.9793 - recall_m: 0.9559 - val_loss: 0.1249 - val_acc: 0.9480 - val_precision_m: 0.9877 - val_recall_m: 0.9153\n",
      "Epoch 1190/5000\n",
      " - 10s - loss: 0.0938 - acc: 0.9687 - precision_m: 0.9801 - recall_m: 0.9571 - val_loss: 0.1095 - val_acc: 0.9690 - val_precision_m: 0.9695 - val_recall_m: 0.9739\n",
      "Epoch 1191/5000\n",
      " - 10s - loss: 0.0945 - acc: 0.9695 - precision_m: 0.9806 - recall_m: 0.9580 - val_loss: 0.1164 - val_acc: 0.9662 - val_precision_m: 0.9617 - val_recall_m: 0.9770\n",
      "Epoch 1192/5000\n",
      " - 10s - loss: 0.0920 - acc: 0.9707 - precision_m: 0.9826 - recall_m: 0.9584 - val_loss: 0.1034 - val_acc: 0.9646 - val_precision_m: 0.9829 - val_recall_m: 0.9514\n",
      "Epoch 1193/5000\n",
      " - 10s - loss: 0.1003 - acc: 0.9662 - precision_m: 0.9792 - recall_m: 0.9537 - val_loss: 0.1359 - val_acc: 0.9551 - val_precision_m: 0.9430 - val_recall_m: 0.9770\n",
      "Epoch 1194/5000\n",
      " - 10s - loss: 0.0987 - acc: 0.9664 - precision_m: 0.9799 - recall_m: 0.9525 - val_loss: 0.1000 - val_acc: 0.9701 - val_precision_m: 0.9801 - val_recall_m: 0.9648\n",
      "Epoch 1195/5000\n",
      " - 10s - loss: 0.0935 - acc: 0.9693 - precision_m: 0.9817 - recall_m: 0.9570 - val_loss: 0.1060 - val_acc: 0.9707 - val_precision_m: 0.9725 - val_recall_m: 0.9740\n",
      "Epoch 1196/5000\n",
      " - 10s - loss: 0.0918 - acc: 0.9701 - precision_m: 0.9813 - recall_m: 0.9584 - val_loss: 0.1021 - val_acc: 0.9640 - val_precision_m: 0.9816 - val_recall_m: 0.9511\n",
      "Epoch 1197/5000\n",
      " - 10s - loss: 0.0923 - acc: 0.9691 - precision_m: 0.9812 - recall_m: 0.9562 - val_loss: 0.1085 - val_acc: 0.9662 - val_precision_m: 0.9862 - val_recall_m: 0.9510\n",
      "Epoch 1198/5000\n",
      " - 10s - loss: 0.0940 - acc: 0.9700 - precision_m: 0.9808 - recall_m: 0.9587 - val_loss: 0.1105 - val_acc: 0.9540 - val_precision_m: 0.9848 - val_recall_m: 0.9292\n",
      "Epoch 1199/5000\n",
      " - 10s - loss: 0.1009 - acc: 0.9660 - precision_m: 0.9791 - recall_m: 0.9528 - val_loss: 0.2161 - val_acc: 0.9158 - val_precision_m: 0.8735 - val_recall_m: 0.9892\n",
      "Epoch 1200/5000\n",
      " - 10s - loss: 0.1006 - acc: 0.9663 - precision_m: 0.9776 - recall_m: 0.9547 - val_loss: 0.1101 - val_acc: 0.9657 - val_precision_m: 0.9672 - val_recall_m: 0.9700\n",
      "Epoch 1201/5000\n",
      " - 10s - loss: 0.1009 - acc: 0.9664 - precision_m: 0.9795 - recall_m: 0.9534 - val_loss: 0.1021 - val_acc: 0.9718 - val_precision_m: 0.9802 - val_recall_m: 0.9679\n",
      "Epoch 1202/5000\n",
      " - 10s - loss: 0.0935 - acc: 0.9693 - precision_m: 0.9819 - recall_m: 0.9567 - val_loss: 0.1044 - val_acc: 0.9695 - val_precision_m: 0.9770 - val_recall_m: 0.9665\n",
      "Epoch 1203/5000\n",
      " - 10s - loss: 0.0929 - acc: 0.9708 - precision_m: 0.9830 - recall_m: 0.9581 - val_loss: 0.1228 - val_acc: 0.9491 - val_precision_m: 0.9869 - val_recall_m: 0.9185\n",
      "Epoch 1204/5000\n",
      " - 10s - loss: 0.1028 - acc: 0.9653 - precision_m: 0.9789 - recall_m: 0.9522 - val_loss: 0.1096 - val_acc: 0.9651 - val_precision_m: 0.9711 - val_recall_m: 0.9643\n",
      "Epoch 1205/5000\n",
      " - 10s - loss: 0.0925 - acc: 0.9695 - precision_m: 0.9818 - recall_m: 0.9568 - val_loss: 0.1038 - val_acc: 0.9695 - val_precision_m: 0.9758 - val_recall_m: 0.9675\n",
      "Epoch 1206/5000\n",
      " - 10s - loss: 0.0911 - acc: 0.9702 - precision_m: 0.9827 - recall_m: 0.9573 - val_loss: 0.1087 - val_acc: 0.9623 - val_precision_m: 0.9838 - val_recall_m: 0.9459\n",
      "Epoch 1207/5000\n",
      " - 10s - loss: 0.0896 - acc: 0.9699 - precision_m: 0.9812 - recall_m: 0.9578 - val_loss: 0.1049 - val_acc: 0.9684 - val_precision_m: 0.9720 - val_recall_m: 0.9696\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1208/5000\n",
      " - 10s - loss: 0.0891 - acc: 0.9708 - precision_m: 0.9822 - recall_m: 0.9593 - val_loss: 0.1240 - val_acc: 0.9496 - val_precision_m: 0.9878 - val_recall_m: 0.9182\n",
      "Epoch 1209/5000\n",
      " - 10s - loss: 0.0889 - acc: 0.9717 - precision_m: 0.9841 - recall_m: 0.9596 - val_loss: 0.1134 - val_acc: 0.9574 - val_precision_m: 0.9890 - val_recall_m: 0.9320\n",
      "Epoch 1210/5000\n",
      " - 10s - loss: 0.0930 - acc: 0.9684 - precision_m: 0.9804 - recall_m: 0.9564 - val_loss: 0.1084 - val_acc: 0.9684 - val_precision_m: 0.9672 - val_recall_m: 0.9750\n",
      "Epoch 1211/5000\n",
      " - 10s - loss: 0.0929 - acc: 0.9698 - precision_m: 0.9818 - recall_m: 0.9579 - val_loss: 0.0992 - val_acc: 0.9695 - val_precision_m: 0.9811 - val_recall_m: 0.9624\n",
      "Epoch 1212/5000\n",
      " - 10s - loss: 0.0905 - acc: 0.9713 - precision_m: 0.9816 - recall_m: 0.9602 - val_loss: 0.1018 - val_acc: 0.9712 - val_precision_m: 0.9823 - val_recall_m: 0.9648\n",
      "Epoch 1213/5000\n",
      " - 10s - loss: 0.0932 - acc: 0.9692 - precision_m: 0.9808 - recall_m: 0.9577 - val_loss: 0.1046 - val_acc: 0.9640 - val_precision_m: 0.9807 - val_recall_m: 0.9519\n",
      "Epoch 1214/5000\n",
      " - 10s - loss: 0.0909 - acc: 0.9704 - precision_m: 0.9818 - recall_m: 0.9584 - val_loss: 0.1045 - val_acc: 0.9701 - val_precision_m: 0.9853 - val_recall_m: 0.9594\n",
      "Epoch 1215/5000\n",
      " - 10s - loss: 0.0918 - acc: 0.9704 - precision_m: 0.9824 - recall_m: 0.9580 - val_loss: 0.1008 - val_acc: 0.9668 - val_precision_m: 0.9750 - val_recall_m: 0.9640\n",
      "Epoch 1216/5000\n",
      " - 10s - loss: 0.0928 - acc: 0.9695 - precision_m: 0.9821 - recall_m: 0.9571 - val_loss: 0.1083 - val_acc: 0.9662 - val_precision_m: 0.9663 - val_recall_m: 0.9716\n",
      "Epoch 1217/5000\n",
      " - 10s - loss: 0.0904 - acc: 0.9709 - precision_m: 0.9809 - recall_m: 0.9607 - val_loss: 0.1069 - val_acc: 0.9557 - val_precision_m: 0.9773 - val_recall_m: 0.9399\n",
      "Epoch 1218/5000\n",
      " - 10s - loss: 0.0910 - acc: 0.9692 - precision_m: 0.9815 - recall_m: 0.9565 - val_loss: 0.1116 - val_acc: 0.9646 - val_precision_m: 0.9691 - val_recall_m: 0.9656\n",
      "Epoch 1219/5000\n",
      " - 10s - loss: 0.0947 - acc: 0.9693 - precision_m: 0.9806 - recall_m: 0.9578 - val_loss: 0.1015 - val_acc: 0.9629 - val_precision_m: 0.9818 - val_recall_m: 0.9490\n",
      "Epoch 1220/5000\n",
      " - 10s - loss: 0.0881 - acc: 0.9712 - precision_m: 0.9824 - recall_m: 0.9592 - val_loss: 0.1033 - val_acc: 0.9629 - val_precision_m: 0.9828 - val_recall_m: 0.9483\n",
      "Epoch 1221/5000\n",
      " - 10s - loss: 0.0893 - acc: 0.9716 - precision_m: 0.9837 - recall_m: 0.9589 - val_loss: 0.1058 - val_acc: 0.9679 - val_precision_m: 0.9840 - val_recall_m: 0.9565\n",
      "Epoch 1222/5000\n",
      " - 10s - loss: 0.0964 - acc: 0.9671 - precision_m: 0.9782 - recall_m: 0.9561 - val_loss: 0.1175 - val_acc: 0.9640 - val_precision_m: 0.9548 - val_recall_m: 0.9801\n",
      "Epoch 1223/5000\n",
      " - 10s - loss: 0.0957 - acc: 0.9679 - precision_m: 0.9802 - recall_m: 0.9567 - val_loss: 0.1091 - val_acc: 0.9707 - val_precision_m: 0.9704 - val_recall_m: 0.9758\n",
      "Epoch 1224/5000\n",
      " - 10s - loss: 0.0882 - acc: 0.9711 - precision_m: 0.9826 - recall_m: 0.9592 - val_loss: 0.1084 - val_acc: 0.9585 - val_precision_m: 0.9860 - val_recall_m: 0.9369\n",
      "Epoch 1225/5000\n",
      " - 10s - loss: 0.0888 - acc: 0.9706 - precision_m: 0.9830 - recall_m: 0.9572 - val_loss: 0.1022 - val_acc: 0.9635 - val_precision_m: 0.9809 - val_recall_m: 0.9512\n",
      "Epoch 1226/5000\n",
      " - 10s - loss: 0.0997 - acc: 0.9658 - precision_m: 0.9779 - recall_m: 0.9541 - val_loss: 0.1111 - val_acc: 0.9540 - val_precision_m: 0.9837 - val_recall_m: 0.9307\n",
      "Epoch 1227/5000\n",
      " - 10s - loss: 0.0925 - acc: 0.9682 - precision_m: 0.9806 - recall_m: 0.9557 - val_loss: 0.1140 - val_acc: 0.9607 - val_precision_m: 0.9645 - val_recall_m: 0.9629\n",
      "Epoch 1228/5000\n",
      " - 10s - loss: 0.0936 - acc: 0.9689 - precision_m: 0.9809 - recall_m: 0.9567 - val_loss: 0.1006 - val_acc: 0.9707 - val_precision_m: 0.9781 - val_recall_m: 0.9677\n",
      "Epoch 1229/5000\n",
      " - 10s - loss: 0.0880 - acc: 0.9706 - precision_m: 0.9814 - recall_m: 0.9596 - val_loss: 0.1012 - val_acc: 0.9740 - val_precision_m: 0.9735 - val_recall_m: 0.9789\n",
      "Epoch 1230/5000\n",
      " - 10s - loss: 0.0895 - acc: 0.9712 - precision_m: 0.9830 - recall_m: 0.9594 - val_loss: 0.1028 - val_acc: 0.9657 - val_precision_m: 0.9779 - val_recall_m: 0.9583\n",
      "Epoch 1231/5000\n",
      " - 10s - loss: 0.0851 - acc: 0.9728 - precision_m: 0.9824 - recall_m: 0.9627 - val_loss: 0.1487 - val_acc: 0.9341 - val_precision_m: 0.9907 - val_recall_m: 0.8868\n",
      "Epoch 1232/5000\n",
      " - 10s - loss: 0.0866 - acc: 0.9727 - precision_m: 0.9838 - recall_m: 0.9615 - val_loss: 0.1034 - val_acc: 0.9695 - val_precision_m: 0.9752 - val_recall_m: 0.9685\n",
      "Epoch 1233/5000\n",
      " - 10s - loss: 0.0906 - acc: 0.9703 - precision_m: 0.9815 - recall_m: 0.9588 - val_loss: 0.1040 - val_acc: 0.9690 - val_precision_m: 0.9783 - val_recall_m: 0.9646\n",
      "Epoch 1234/5000\n",
      " - 10s - loss: 0.0945 - acc: 0.9691 - precision_m: 0.9808 - recall_m: 0.9572 - val_loss: 0.1080 - val_acc: 0.9695 - val_precision_m: 0.9636 - val_recall_m: 0.9811\n",
      "Epoch 1235/5000\n",
      " - 10s - loss: 0.0913 - acc: 0.9706 - precision_m: 0.9826 - recall_m: 0.9584 - val_loss: 0.1009 - val_acc: 0.9712 - val_precision_m: 0.9772 - val_recall_m: 0.9698\n",
      "Epoch 1236/5000\n",
      " - 10s - loss: 0.0873 - acc: 0.9716 - precision_m: 0.9829 - recall_m: 0.9596 - val_loss: 0.1034 - val_acc: 0.9601 - val_precision_m: 0.9870 - val_recall_m: 0.9387\n",
      "Epoch 1237/5000\n",
      " - 10s - loss: 0.0904 - acc: 0.9685 - precision_m: 0.9797 - recall_m: 0.9571 - val_loss: 0.1200 - val_acc: 0.9668 - val_precision_m: 0.9618 - val_recall_m: 0.9780\n",
      "Epoch 1238/5000\n",
      " - 10s - loss: 0.0902 - acc: 0.9707 - precision_m: 0.9820 - recall_m: 0.9596 - val_loss: 0.1028 - val_acc: 0.9668 - val_precision_m: 0.9691 - val_recall_m: 0.9696\n",
      "Epoch 1239/5000\n",
      " - 10s - loss: 0.0843 - acc: 0.9728 - precision_m: 0.9831 - recall_m: 0.9623 - val_loss: 0.1015 - val_acc: 0.9673 - val_precision_m: 0.9789 - val_recall_m: 0.9602\n",
      "Epoch 1240/5000\n",
      " - 10s - loss: 0.0958 - acc: 0.9677 - precision_m: 0.9792 - recall_m: 0.9561 - val_loss: 0.1043 - val_acc: 0.9612 - val_precision_m: 0.9828 - val_recall_m: 0.9449\n",
      "Epoch 1241/5000\n",
      " - 10s - loss: 0.0849 - acc: 0.9730 - precision_m: 0.9831 - recall_m: 0.9626 - val_loss: 0.0984 - val_acc: 0.9734 - val_precision_m: 0.9754 - val_recall_m: 0.9760\n",
      "Epoch 1242/5000\n",
      " - 10s - loss: 0.0848 - acc: 0.9730 - precision_m: 0.9816 - recall_m: 0.9639 - val_loss: 0.1057 - val_acc: 0.9607 - val_precision_m: 0.9837 - val_recall_m: 0.9424\n",
      "Epoch 1243/5000\n",
      " - 10s - loss: 0.0896 - acc: 0.9703 - precision_m: 0.9821 - recall_m: 0.9581 - val_loss: 0.0945 - val_acc: 0.9707 - val_precision_m: 0.9801 - val_recall_m: 0.9654\n",
      "Epoch 1244/5000\n",
      " - 10s - loss: 0.0882 - acc: 0.9707 - precision_m: 0.9819 - recall_m: 0.9594 - val_loss: 0.0992 - val_acc: 0.9723 - val_precision_m: 0.9751 - val_recall_m: 0.9739\n",
      "Epoch 1245/5000\n",
      " - 10s - loss: 0.0860 - acc: 0.9737 - precision_m: 0.9844 - recall_m: 0.9627 - val_loss: 0.1167 - val_acc: 0.9662 - val_precision_m: 0.9580 - val_recall_m: 0.9812\n",
      "Epoch 1246/5000\n",
      " - 10s - loss: 0.0878 - acc: 0.9707 - precision_m: 0.9813 - recall_m: 0.9601 - val_loss: 0.1095 - val_acc: 0.9563 - val_precision_m: 0.9837 - val_recall_m: 0.9346\n",
      "Epoch 1247/5000\n",
      " - 10s - loss: 0.0879 - acc: 0.9716 - precision_m: 0.9829 - recall_m: 0.9599 - val_loss: 0.0984 - val_acc: 0.9718 - val_precision_m: 0.9821 - val_recall_m: 0.9657\n",
      "Epoch 1248/5000\n",
      " - 10s - loss: 0.0872 - acc: 0.9716 - precision_m: 0.9815 - recall_m: 0.9611 - val_loss: 0.0997 - val_acc: 0.9723 - val_precision_m: 0.9781 - val_recall_m: 0.9708\n",
      "Epoch 1249/5000\n",
      " - 10s - loss: 0.0880 - acc: 0.9712 - precision_m: 0.9819 - recall_m: 0.9602 - val_loss: 0.1012 - val_acc: 0.9601 - val_precision_m: 0.9837 - val_recall_m: 0.9427\n",
      "Epoch 1250/5000\n",
      " - 10s - loss: 0.0872 - acc: 0.9722 - precision_m: 0.9826 - recall_m: 0.9617 - val_loss: 0.1056 - val_acc: 0.9601 - val_precision_m: 0.9860 - val_recall_m: 0.9396\n",
      "Epoch 1251/5000\n",
      " - 10s - loss: 0.0982 - acc: 0.9670 - precision_m: 0.9790 - recall_m: 0.9554 - val_loss: 0.1001 - val_acc: 0.9723 - val_precision_m: 0.9762 - val_recall_m: 0.9725\n",
      "Epoch 1252/5000\n",
      " - 10s - loss: 0.0884 - acc: 0.9716 - precision_m: 0.9820 - recall_m: 0.9608 - val_loss: 0.1422 - val_acc: 0.9374 - val_precision_m: 0.9898 - val_recall_m: 0.8936\n",
      "Epoch 1253/5000\n",
      " - 10s - loss: 0.0997 - acc: 0.9655 - precision_m: 0.9781 - recall_m: 0.9534 - val_loss: 0.0965 - val_acc: 0.9629 - val_precision_m: 0.9828 - val_recall_m: 0.9477\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1254/5000\n",
      " - 10s - loss: 0.0850 - acc: 0.9732 - precision_m: 0.9840 - recall_m: 0.9620 - val_loss: 0.1025 - val_acc: 0.9574 - val_precision_m: 0.9815 - val_recall_m: 0.9386\n",
      "Epoch 1255/5000\n",
      " - 10s - loss: 0.0859 - acc: 0.9714 - precision_m: 0.9830 - recall_m: 0.9597 - val_loss: 0.1170 - val_acc: 0.9518 - val_precision_m: 0.9868 - val_recall_m: 0.9236\n",
      "Epoch 1256/5000\n",
      " - 10s - loss: 0.0842 - acc: 0.9727 - precision_m: 0.9829 - recall_m: 0.9623 - val_loss: 0.1104 - val_acc: 0.9579 - val_precision_m: 0.9848 - val_recall_m: 0.9370\n",
      "Epoch 1257/5000\n",
      " - 10s - loss: 0.0885 - acc: 0.9708 - precision_m: 0.9825 - recall_m: 0.9589 - val_loss: 0.1019 - val_acc: 0.9657 - val_precision_m: 0.9732 - val_recall_m: 0.9634\n",
      "Epoch 1258/5000\n",
      " - 10s - loss: 0.0835 - acc: 0.9732 - precision_m: 0.9835 - recall_m: 0.9629 - val_loss: 0.0979 - val_acc: 0.9723 - val_precision_m: 0.9735 - val_recall_m: 0.9760\n",
      "Epoch 1259/5000\n",
      " - 10s - loss: 0.0866 - acc: 0.9712 - precision_m: 0.9833 - recall_m: 0.9592 - val_loss: 0.0979 - val_acc: 0.9701 - val_precision_m: 0.9831 - val_recall_m: 0.9617\n",
      "Epoch 1260/5000\n",
      " - 10s - loss: 0.0887 - acc: 0.9703 - precision_m: 0.9807 - recall_m: 0.9595 - val_loss: 0.1224 - val_acc: 0.9491 - val_precision_m: 0.9900 - val_recall_m: 0.9151\n",
      "Epoch 1261/5000\n",
      " - 10s - loss: 0.0832 - acc: 0.9727 - precision_m: 0.9837 - recall_m: 0.9613 - val_loss: 0.1120 - val_acc: 0.9535 - val_precision_m: 0.9869 - val_recall_m: 0.9270\n",
      "Epoch 1262/5000\n",
      " - 10s - loss: 0.0830 - acc: 0.9722 - precision_m: 0.9822 - recall_m: 0.9617 - val_loss: 0.0945 - val_acc: 0.9657 - val_precision_m: 0.9798 - val_recall_m: 0.9568\n",
      "Epoch 1263/5000\n",
      " - 10s - loss: 0.0831 - acc: 0.9737 - precision_m: 0.9846 - recall_m: 0.9627 - val_loss: 0.1082 - val_acc: 0.9673 - val_precision_m: 0.9588 - val_recall_m: 0.9822\n",
      "Epoch 1264/5000\n",
      " - 10s - loss: 0.0828 - acc: 0.9731 - precision_m: 0.9830 - recall_m: 0.9630 - val_loss: 0.1063 - val_acc: 0.9701 - val_precision_m: 0.9647 - val_recall_m: 0.9811\n",
      "Epoch 1265/5000\n",
      " - 10s - loss: 0.0841 - acc: 0.9725 - precision_m: 0.9826 - recall_m: 0.9621 - val_loss: 0.0954 - val_acc: 0.9707 - val_precision_m: 0.9801 - val_recall_m: 0.9657\n",
      "Epoch 1266/5000\n",
      " - 10s - loss: 0.0841 - acc: 0.9724 - precision_m: 0.9829 - recall_m: 0.9616 - val_loss: 0.1129 - val_acc: 0.9646 - val_precision_m: 0.9550 - val_recall_m: 0.9810\n",
      "Epoch 1267/5000\n",
      " - 10s - loss: 0.0941 - acc: 0.9682 - precision_m: 0.9797 - recall_m: 0.9570 - val_loss: 0.0999 - val_acc: 0.9668 - val_precision_m: 0.9843 - val_recall_m: 0.9542\n",
      "Epoch 1268/5000\n",
      " - 10s - loss: 0.0860 - acc: 0.9725 - precision_m: 0.9824 - recall_m: 0.9628 - val_loss: 0.0986 - val_acc: 0.9640 - val_precision_m: 0.9798 - val_recall_m: 0.9532\n",
      "Epoch 1269/5000\n",
      " - 10s - loss: 0.0841 - acc: 0.9730 - precision_m: 0.9826 - recall_m: 0.9631 - val_loss: 0.0945 - val_acc: 0.9718 - val_precision_m: 0.9820 - val_recall_m: 0.9657\n",
      "Epoch 1270/5000\n",
      " - 10s - loss: 0.0812 - acc: 0.9736 - precision_m: 0.9840 - recall_m: 0.9632 - val_loss: 0.1115 - val_acc: 0.9684 - val_precision_m: 0.9618 - val_recall_m: 0.9813\n",
      "Epoch 1271/5000\n",
      " - 10s - loss: 0.0814 - acc: 0.9737 - precision_m: 0.9836 - recall_m: 0.9637 - val_loss: 0.0919 - val_acc: 0.9745 - val_precision_m: 0.9823 - val_recall_m: 0.9708\n",
      "Epoch 1272/5000\n",
      " - 10s - loss: 0.0871 - acc: 0.9709 - precision_m: 0.9829 - recall_m: 0.9588 - val_loss: 0.0927 - val_acc: 0.9668 - val_precision_m: 0.9799 - val_recall_m: 0.9588\n",
      "Epoch 1273/5000\n",
      " - 10s - loss: 0.0800 - acc: 0.9754 - precision_m: 0.9845 - recall_m: 0.9661 - val_loss: 0.0938 - val_acc: 0.9657 - val_precision_m: 0.9850 - val_recall_m: 0.9512\n",
      "Epoch 1274/5000\n",
      " - 10s - loss: 0.0801 - acc: 0.9742 - precision_m: 0.9834 - recall_m: 0.9649 - val_loss: 0.0947 - val_acc: 0.9695 - val_precision_m: 0.9841 - val_recall_m: 0.9594\n",
      "Epoch 1275/5000\n",
      " - 10s - loss: 0.0869 - acc: 0.9712 - precision_m: 0.9814 - recall_m: 0.9611 - val_loss: 0.0919 - val_acc: 0.9712 - val_precision_m: 0.9789 - val_recall_m: 0.9682\n",
      "Epoch 1276/5000\n",
      " - 10s - loss: 0.0816 - acc: 0.9737 - precision_m: 0.9839 - recall_m: 0.9634 - val_loss: 0.1029 - val_acc: 0.9745 - val_precision_m: 0.9716 - val_recall_m: 0.9822\n",
      "Epoch 1277/5000\n",
      " - 10s - loss: 0.0885 - acc: 0.9708 - precision_m: 0.9818 - recall_m: 0.9597 - val_loss: 0.1011 - val_acc: 0.9629 - val_precision_m: 0.9882 - val_recall_m: 0.9431\n",
      "Epoch 1278/5000\n",
      " - 10s - loss: 0.0869 - acc: 0.9715 - precision_m: 0.9828 - recall_m: 0.9601 - val_loss: 0.0951 - val_acc: 0.9701 - val_precision_m: 0.9811 - val_recall_m: 0.9640\n",
      "Epoch 1279/5000\n",
      " - 10s - loss: 0.0878 - acc: 0.9714 - precision_m: 0.9826 - recall_m: 0.9605 - val_loss: 0.0944 - val_acc: 0.9635 - val_precision_m: 0.9787 - val_recall_m: 0.9530\n",
      "Epoch 1280/5000\n",
      " - 10s - loss: 0.0860 - acc: 0.9714 - precision_m: 0.9815 - recall_m: 0.9609 - val_loss: 0.0972 - val_acc: 0.9729 - val_precision_m: 0.9723 - val_recall_m: 0.9779\n",
      "Epoch 1281/5000\n",
      " - 10s - loss: 0.0812 - acc: 0.9742 - precision_m: 0.9838 - recall_m: 0.9641 - val_loss: 0.1009 - val_acc: 0.9734 - val_precision_m: 0.9714 - val_recall_m: 0.9801\n",
      "Epoch 1282/5000\n",
      " - 10s - loss: 0.0916 - acc: 0.9693 - precision_m: 0.9806 - recall_m: 0.9584 - val_loss: 0.1082 - val_acc: 0.9563 - val_precision_m: 0.9858 - val_recall_m: 0.9328\n",
      "Epoch 1283/5000\n",
      " - 10s - loss: 0.0815 - acc: 0.9740 - precision_m: 0.9840 - recall_m: 0.9635 - val_loss: 0.0932 - val_acc: 0.9723 - val_precision_m: 0.9812 - val_recall_m: 0.9676\n",
      "Epoch 1284/5000\n",
      " - 10s - loss: 0.0814 - acc: 0.9735 - precision_m: 0.9837 - recall_m: 0.9631 - val_loss: 0.0992 - val_acc: 0.9729 - val_precision_m: 0.9751 - val_recall_m: 0.9751\n",
      "Epoch 1285/5000\n",
      " - 10s - loss: 0.0792 - acc: 0.9748 - precision_m: 0.9843 - recall_m: 0.9652 - val_loss: 0.1088 - val_acc: 0.9579 - val_precision_m: 0.9891 - val_recall_m: 0.9330\n",
      "Epoch 1286/5000\n",
      " - 10s - loss: 0.0898 - acc: 0.9691 - precision_m: 0.9808 - recall_m: 0.9579 - val_loss: 0.0964 - val_acc: 0.9723 - val_precision_m: 0.9734 - val_recall_m: 0.9762\n",
      "Epoch 1287/5000\n",
      " - 10s - loss: 0.0799 - acc: 0.9738 - precision_m: 0.9840 - recall_m: 0.9635 - val_loss: 0.0955 - val_acc: 0.9640 - val_precision_m: 0.9808 - val_recall_m: 0.9522\n",
      "Epoch 1288/5000\n",
      " - 10s - loss: 0.0822 - acc: 0.9733 - precision_m: 0.9827 - recall_m: 0.9640 - val_loss: 0.0944 - val_acc: 0.9729 - val_precision_m: 0.9737 - val_recall_m: 0.9772\n",
      "Epoch 1289/5000\n",
      " - 10s - loss: 0.0841 - acc: 0.9723 - precision_m: 0.9827 - recall_m: 0.9615 - val_loss: 0.0941 - val_acc: 0.9707 - val_precision_m: 0.9782 - val_recall_m: 0.9678\n",
      "Epoch 1290/5000\n",
      " - 10s - loss: 0.0845 - acc: 0.9716 - precision_m: 0.9821 - recall_m: 0.9609 - val_loss: 0.0953 - val_acc: 0.9734 - val_precision_m: 0.9745 - val_recall_m: 0.9769\n",
      "Epoch 1291/5000\n",
      " - 10s - loss: 0.0804 - acc: 0.9749 - precision_m: 0.9834 - recall_m: 0.9661 - val_loss: 0.0993 - val_acc: 0.9607 - val_precision_m: 0.9848 - val_recall_m: 0.9425\n",
      "Epoch 1292/5000\n",
      " - 10s - loss: 0.0872 - acc: 0.9705 - precision_m: 0.9817 - recall_m: 0.9593 - val_loss: 0.1077 - val_acc: 0.9529 - val_precision_m: 0.9848 - val_recall_m: 0.9276\n",
      "Epoch 1293/5000\n",
      " - 10s - loss: 0.0772 - acc: 0.9742 - precision_m: 0.9840 - recall_m: 0.9637 - val_loss: 0.1475 - val_acc: 0.9380 - val_precision_m: 0.9965 - val_recall_m: 0.8890\n",
      "Epoch 1294/5000\n",
      " - 10s - loss: 0.0795 - acc: 0.9746 - precision_m: 0.9847 - recall_m: 0.9647 - val_loss: 0.1129 - val_acc: 0.9546 - val_precision_m: 0.9869 - val_recall_m: 0.9287\n",
      "Epoch 1295/5000\n",
      " - 10s - loss: 0.0797 - acc: 0.9736 - precision_m: 0.9840 - recall_m: 0.9632 - val_loss: 0.0932 - val_acc: 0.9729 - val_precision_m: 0.9812 - val_recall_m: 0.9687\n",
      "Epoch 1296/5000\n",
      " - 10s - loss: 0.0842 - acc: 0.9721 - precision_m: 0.9818 - recall_m: 0.9623 - val_loss: 0.1112 - val_acc: 0.9662 - val_precision_m: 0.9572 - val_recall_m: 0.9822\n",
      "Epoch 1297/5000\n",
      " - 10s - loss: 0.0915 - acc: 0.9686 - precision_m: 0.9780 - recall_m: 0.9590 - val_loss: 0.1181 - val_acc: 0.9546 - val_precision_m: 0.9901 - val_recall_m: 0.9255\n",
      "Epoch 1298/5000\n",
      " - 10s - loss: 0.0813 - acc: 0.9734 - precision_m: 0.9839 - recall_m: 0.9626 - val_loss: 0.1405 - val_acc: 0.9596 - val_precision_m: 0.9393 - val_recall_m: 0.9901\n",
      "Epoch 1299/5000\n",
      " - 10s - loss: 0.0824 - acc: 0.9737 - precision_m: 0.9838 - recall_m: 0.9640 - val_loss: 0.0912 - val_acc: 0.9740 - val_precision_m: 0.9782 - val_recall_m: 0.9739\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1300/5000\n",
      " - 10s - loss: 0.0777 - acc: 0.9749 - precision_m: 0.9841 - recall_m: 0.9653 - val_loss: 0.1084 - val_acc: 0.9524 - val_precision_m: 0.9847 - val_recall_m: 0.9269\n",
      "Epoch 1301/5000\n",
      " - 10s - loss: 0.0791 - acc: 0.9754 - precision_m: 0.9847 - recall_m: 0.9655 - val_loss: 0.0927 - val_acc: 0.9740 - val_precision_m: 0.9717 - val_recall_m: 0.9809\n",
      "Epoch 1302/5000\n",
      " - 10s - loss: 0.0870 - acc: 0.9709 - precision_m: 0.9806 - recall_m: 0.9611 - val_loss: 0.0979 - val_acc: 0.9646 - val_precision_m: 0.9809 - val_recall_m: 0.9530\n",
      "Epoch 1303/5000\n",
      " - 10s - loss: 0.0796 - acc: 0.9745 - precision_m: 0.9855 - recall_m: 0.9627 - val_loss: 0.0914 - val_acc: 0.9729 - val_precision_m: 0.9793 - val_recall_m: 0.9706\n",
      "Epoch 1304/5000\n",
      " - 10s - loss: 0.0898 - acc: 0.9692 - precision_m: 0.9798 - recall_m: 0.9583 - val_loss: 0.0919 - val_acc: 0.9673 - val_precision_m: 0.9789 - val_recall_m: 0.9604\n",
      "Epoch 1305/5000\n",
      " - 10s - loss: 0.0852 - acc: 0.9719 - precision_m: 0.9840 - recall_m: 0.9598 - val_loss: 0.0990 - val_acc: 0.9684 - val_precision_m: 0.9764 - val_recall_m: 0.9655\n",
      "Epoch 1306/5000\n",
      " - 10s - loss: 0.0808 - acc: 0.9735 - precision_m: 0.9835 - recall_m: 0.9628 - val_loss: 0.0903 - val_acc: 0.9718 - val_precision_m: 0.9801 - val_recall_m: 0.9677\n",
      "Epoch 1307/5000\n",
      " - 10s - loss: 0.0778 - acc: 0.9751 - precision_m: 0.9834 - recall_m: 0.9667 - val_loss: 0.0917 - val_acc: 0.9673 - val_precision_m: 0.9829 - val_recall_m: 0.9563\n",
      "Epoch 1308/5000\n",
      " - 10s - loss: 0.0781 - acc: 0.9739 - precision_m: 0.9841 - recall_m: 0.9641 - val_loss: 0.1131 - val_acc: 0.9524 - val_precision_m: 0.9878 - val_recall_m: 0.9236\n",
      "Epoch 1309/5000\n",
      " - 10s - loss: 0.0844 - acc: 0.9720 - precision_m: 0.9811 - recall_m: 0.9636 - val_loss: 0.0950 - val_acc: 0.9701 - val_precision_m: 0.9772 - val_recall_m: 0.9677\n",
      "Epoch 1310/5000\n",
      " - 10s - loss: 0.0764 - acc: 0.9753 - precision_m: 0.9859 - recall_m: 0.9642 - val_loss: 0.1289 - val_acc: 0.9568 - val_precision_m: 0.9409 - val_recall_m: 0.9821\n",
      "Epoch 1311/5000\n",
      " - 10s - loss: 0.0835 - acc: 0.9721 - precision_m: 0.9804 - recall_m: 0.9637 - val_loss: 0.0886 - val_acc: 0.9723 - val_precision_m: 0.9697 - val_recall_m: 0.9801\n",
      "Epoch 1312/5000\n",
      " - 10s - loss: 0.0772 - acc: 0.9759 - precision_m: 0.9847 - recall_m: 0.9673 - val_loss: 0.0885 - val_acc: 0.9740 - val_precision_m: 0.9736 - val_recall_m: 0.9791\n",
      "Epoch 1313/5000\n",
      " - 10s - loss: 0.0793 - acc: 0.9736 - precision_m: 0.9830 - recall_m: 0.9639 - val_loss: 0.0884 - val_acc: 0.9729 - val_precision_m: 0.9753 - val_recall_m: 0.9752\n",
      "Epoch 1314/5000\n",
      " - 10s - loss: 0.0788 - acc: 0.9747 - precision_m: 0.9843 - recall_m: 0.9650 - val_loss: 0.1107 - val_acc: 0.9662 - val_precision_m: 0.9559 - val_recall_m: 0.9830\n",
      "Epoch 1315/5000\n",
      " - 10s - loss: 0.0780 - acc: 0.9756 - precision_m: 0.9849 - recall_m: 0.9660 - val_loss: 0.0930 - val_acc: 0.9673 - val_precision_m: 0.9862 - val_recall_m: 0.9533\n",
      "Epoch 1316/5000\n",
      " - 10s - loss: 0.0826 - acc: 0.9718 - precision_m: 0.9825 - recall_m: 0.9611 - val_loss: 0.0991 - val_acc: 0.9618 - val_precision_m: 0.9871 - val_recall_m: 0.9418\n",
      "Epoch 1317/5000\n",
      " - 10s - loss: 0.0834 - acc: 0.9733 - precision_m: 0.9832 - recall_m: 0.9636 - val_loss: 0.0967 - val_acc: 0.9729 - val_precision_m: 0.9744 - val_recall_m: 0.9760\n",
      "Epoch 1318/5000\n",
      " - 10s - loss: 0.0756 - acc: 0.9774 - precision_m: 0.9863 - recall_m: 0.9681 - val_loss: 0.0980 - val_acc: 0.9590 - val_precision_m: 0.9826 - val_recall_m: 0.9409\n",
      "Epoch 1319/5000\n",
      " - 10s - loss: 0.0802 - acc: 0.9730 - precision_m: 0.9830 - recall_m: 0.9630 - val_loss: 0.0932 - val_acc: 0.9723 - val_precision_m: 0.9771 - val_recall_m: 0.9718\n",
      "Epoch 1320/5000\n",
      " - 10s - loss: 0.0818 - acc: 0.9735 - precision_m: 0.9829 - recall_m: 0.9640 - val_loss: 0.0981 - val_acc: 0.9629 - val_precision_m: 0.9819 - val_recall_m: 0.9493\n",
      "Epoch 1321/5000\n",
      " - 10s - loss: 0.0771 - acc: 0.9754 - precision_m: 0.9852 - recall_m: 0.9660 - val_loss: 0.1123 - val_acc: 0.9668 - val_precision_m: 0.9523 - val_recall_m: 0.9880\n",
      "Epoch 1322/5000\n",
      " - 10s - loss: 0.0849 - acc: 0.9721 - precision_m: 0.9820 - recall_m: 0.9624 - val_loss: 0.0892 - val_acc: 0.9745 - val_precision_m: 0.9813 - val_recall_m: 0.9720\n",
      "Epoch 1323/5000\n",
      " - 10s - loss: 0.0775 - acc: 0.9749 - precision_m: 0.9840 - recall_m: 0.9661 - val_loss: 0.0954 - val_acc: 0.9684 - val_precision_m: 0.9850 - val_recall_m: 0.9566\n",
      "Epoch 1324/5000\n",
      " - 10s - loss: 0.0757 - acc: 0.9765 - precision_m: 0.9848 - recall_m: 0.9681 - val_loss: 0.0891 - val_acc: 0.9723 - val_precision_m: 0.9812 - val_recall_m: 0.9675\n",
      "Epoch 1325/5000\n",
      " - 10s - loss: 0.0767 - acc: 0.9761 - precision_m: 0.9851 - recall_m: 0.9670 - val_loss: 0.1047 - val_acc: 0.9590 - val_precision_m: 0.9891 - val_recall_m: 0.9346\n",
      "Epoch 1326/5000\n",
      " - 10s - loss: 0.0935 - acc: 0.9673 - precision_m: 0.9775 - recall_m: 0.9580 - val_loss: 0.1203 - val_acc: 0.9640 - val_precision_m: 0.9530 - val_recall_m: 0.9819\n",
      "Epoch 1327/5000\n",
      " - 10s - loss: 0.0804 - acc: 0.9741 - precision_m: 0.9835 - recall_m: 0.9645 - val_loss: 0.0925 - val_acc: 0.9695 - val_precision_m: 0.9851 - val_recall_m: 0.9584\n",
      "Epoch 1328/5000\n",
      " - 10s - loss: 0.0818 - acc: 0.9733 - precision_m: 0.9832 - recall_m: 0.9634 - val_loss: 0.0904 - val_acc: 0.9734 - val_precision_m: 0.9781 - val_recall_m: 0.9729\n",
      "Epoch 1329/5000\n",
      " - 10s - loss: 0.0752 - acc: 0.9765 - precision_m: 0.9856 - recall_m: 0.9672 - val_loss: 0.0918 - val_acc: 0.9734 - val_precision_m: 0.9745 - val_recall_m: 0.9770\n",
      "Epoch 1330/5000\n",
      " - 10s - loss: 0.0760 - acc: 0.9759 - precision_m: 0.9858 - recall_m: 0.9658 - val_loss: 0.0875 - val_acc: 0.9734 - val_precision_m: 0.9780 - val_recall_m: 0.9730\n",
      "Epoch 1331/5000\n",
      " - 10s - loss: 0.0769 - acc: 0.9745 - precision_m: 0.9845 - recall_m: 0.9641 - val_loss: 0.1079 - val_acc: 0.9563 - val_precision_m: 0.9868 - val_recall_m: 0.9317\n",
      "Epoch 1332/5000\n",
      " - 10s - loss: 0.0783 - acc: 0.9750 - precision_m: 0.9851 - recall_m: 0.9648 - val_loss: 0.0923 - val_acc: 0.9651 - val_precision_m: 0.9851 - val_recall_m: 0.9503\n",
      "Epoch 1333/5000\n",
      " - 10s - loss: 0.0775 - acc: 0.9749 - precision_m: 0.9838 - recall_m: 0.9653 - val_loss: 0.0957 - val_acc: 0.9629 - val_precision_m: 0.9881 - val_recall_m: 0.9430\n",
      "Epoch 1334/5000\n",
      " - 10s - loss: 0.0752 - acc: 0.9763 - precision_m: 0.9857 - recall_m: 0.9669 - val_loss: 0.0874 - val_acc: 0.9740 - val_precision_m: 0.9792 - val_recall_m: 0.9729\n",
      "Epoch 1335/5000\n",
      " - 10s - loss: 0.0764 - acc: 0.9748 - precision_m: 0.9853 - recall_m: 0.9646 - val_loss: 0.0931 - val_acc: 0.9718 - val_precision_m: 0.9752 - val_recall_m: 0.9727\n",
      "Epoch 1336/5000\n",
      " - 10s - loss: 0.0760 - acc: 0.9768 - precision_m: 0.9856 - recall_m: 0.9678 - val_loss: 0.0928 - val_acc: 0.9734 - val_precision_m: 0.9696 - val_recall_m: 0.9822\n",
      "Epoch 1337/5000\n",
      " - 10s - loss: 0.0818 - acc: 0.9730 - precision_m: 0.9815 - recall_m: 0.9649 - val_loss: 0.0907 - val_acc: 0.9712 - val_precision_m: 0.9780 - val_recall_m: 0.9688\n",
      "Epoch 1338/5000\n",
      " - 10s - loss: 0.0741 - acc: 0.9758 - precision_m: 0.9852 - recall_m: 0.9664 - val_loss: 0.1017 - val_acc: 0.9668 - val_precision_m: 0.9692 - val_recall_m: 0.9700\n",
      "Epoch 1339/5000\n",
      " - 10s - loss: 0.0785 - acc: 0.9738 - precision_m: 0.9835 - recall_m: 0.9645 - val_loss: 0.0955 - val_acc: 0.9646 - val_precision_m: 0.9892 - val_recall_m: 0.9452\n",
      "Epoch 1340/5000\n",
      " - 10s - loss: 0.0772 - acc: 0.9747 - precision_m: 0.9836 - recall_m: 0.9658 - val_loss: 0.0882 - val_acc: 0.9718 - val_precision_m: 0.9841 - val_recall_m: 0.9636\n",
      "Epoch 1341/5000\n",
      " - 10s - loss: 0.0783 - acc: 0.9728 - precision_m: 0.9831 - recall_m: 0.9625 - val_loss: 0.0879 - val_acc: 0.9718 - val_precision_m: 0.9695 - val_recall_m: 0.9790\n",
      "Epoch 1342/5000\n",
      " - 10s - loss: 0.0836 - acc: 0.9717 - precision_m: 0.9819 - recall_m: 0.9617 - val_loss: 0.0890 - val_acc: 0.9756 - val_precision_m: 0.9784 - val_recall_m: 0.9771\n",
      "Epoch 1343/5000\n",
      " - 10s - loss: 0.0761 - acc: 0.9761 - precision_m: 0.9852 - recall_m: 0.9664 - val_loss: 0.0881 - val_acc: 0.9718 - val_precision_m: 0.9772 - val_recall_m: 0.9706\n",
      "Epoch 1344/5000\n",
      " - 10s - loss: 0.0748 - acc: 0.9759 - precision_m: 0.9847 - recall_m: 0.9670 - val_loss: 0.1473 - val_acc: 0.9336 - val_precision_m: 0.9942 - val_recall_m: 0.8823\n",
      "Epoch 1345/5000\n",
      " - 10s - loss: 0.0784 - acc: 0.9745 - precision_m: 0.9842 - recall_m: 0.9651 - val_loss: 0.0895 - val_acc: 0.9734 - val_precision_m: 0.9811 - val_recall_m: 0.9699\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1346/5000\n",
      " - 10s - loss: 0.0746 - acc: 0.9765 - precision_m: 0.9842 - recall_m: 0.9685 - val_loss: 0.1039 - val_acc: 0.9618 - val_precision_m: 0.9902 - val_recall_m: 0.9390\n",
      "Epoch 1347/5000\n",
      " - 10s - loss: 0.0757 - acc: 0.9765 - precision_m: 0.9869 - recall_m: 0.9657 - val_loss: 0.1047 - val_acc: 0.9679 - val_precision_m: 0.9693 - val_recall_m: 0.9718\n",
      "Epoch 1348/5000\n",
      " - 10s - loss: 0.0773 - acc: 0.9754 - precision_m: 0.9843 - recall_m: 0.9665 - val_loss: 0.0887 - val_acc: 0.9729 - val_precision_m: 0.9754 - val_recall_m: 0.9748\n",
      "Epoch 1349/5000\n",
      " - 10s - loss: 0.0771 - acc: 0.9733 - precision_m: 0.9831 - recall_m: 0.9634 - val_loss: 0.0903 - val_acc: 0.9756 - val_precision_m: 0.9753 - val_recall_m: 0.9803\n",
      "Epoch 1350/5000\n",
      " - 10s - loss: 0.0748 - acc: 0.9755 - precision_m: 0.9848 - recall_m: 0.9663 - val_loss: 0.0940 - val_acc: 0.9718 - val_precision_m: 0.9684 - val_recall_m: 0.9803\n",
      "Epoch 1351/5000\n",
      " - 10s - loss: 0.0817 - acc: 0.9725 - precision_m: 0.9830 - recall_m: 0.9621 - val_loss: 0.1130 - val_acc: 0.9629 - val_precision_m: 0.9506 - val_recall_m: 0.9831\n",
      "Epoch 1352/5000\n",
      " - 10s - loss: 0.0767 - acc: 0.9748 - precision_m: 0.9834 - recall_m: 0.9660 - val_loss: 0.0883 - val_acc: 0.9712 - val_precision_m: 0.9753 - val_recall_m: 0.9719\n",
      "Epoch 1353/5000\n",
      " - 10s - loss: 0.0779 - acc: 0.9739 - precision_m: 0.9831 - recall_m: 0.9643 - val_loss: 0.0883 - val_acc: 0.9707 - val_precision_m: 0.9821 - val_recall_m: 0.9638\n",
      "Epoch 1354/5000\n",
      " - 10s - loss: 0.0816 - acc: 0.9717 - precision_m: 0.9805 - recall_m: 0.9632 - val_loss: 0.0944 - val_acc: 0.9690 - val_precision_m: 0.9790 - val_recall_m: 0.9640\n",
      "Epoch 1355/5000\n",
      " - 10s - loss: 0.0779 - acc: 0.9745 - precision_m: 0.9839 - recall_m: 0.9650 - val_loss: 0.1020 - val_acc: 0.9707 - val_precision_m: 0.9629 - val_recall_m: 0.9843\n",
      "Epoch 1356/5000\n",
      " - 10s - loss: 0.0751 - acc: 0.9762 - precision_m: 0.9848 - recall_m: 0.9676 - val_loss: 0.0900 - val_acc: 0.9740 - val_precision_m: 0.9725 - val_recall_m: 0.9802\n",
      "Epoch 1357/5000\n",
      " - 10s - loss: 0.0787 - acc: 0.9733 - precision_m: 0.9835 - recall_m: 0.9628 - val_loss: 0.1205 - val_acc: 0.9468 - val_precision_m: 0.9911 - val_recall_m: 0.9103\n",
      "Epoch 1358/5000\n",
      " - 10s - loss: 0.0748 - acc: 0.9752 - precision_m: 0.9844 - recall_m: 0.9654 - val_loss: 0.1013 - val_acc: 0.9612 - val_precision_m: 0.9881 - val_recall_m: 0.9400\n",
      "Epoch 1359/5000\n",
      " - 10s - loss: 0.0928 - acc: 0.9680 - precision_m: 0.9790 - recall_m: 0.9576 - val_loss: 0.1198 - val_acc: 0.9468 - val_precision_m: 0.9933 - val_recall_m: 0.9085\n",
      "Epoch 1360/5000\n",
      " - 10s - loss: 0.0753 - acc: 0.9757 - precision_m: 0.9858 - recall_m: 0.9656 - val_loss: 0.0892 - val_acc: 0.9723 - val_precision_m: 0.9864 - val_recall_m: 0.9624\n",
      "Epoch 1361/5000\n",
      " - 10s - loss: 0.0706 - acc: 0.9764 - precision_m: 0.9857 - recall_m: 0.9670 - val_loss: 0.0878 - val_acc: 0.9745 - val_precision_m: 0.9783 - val_recall_m: 0.9750\n",
      "Epoch 1362/5000\n",
      " - 10s - loss: 0.0687 - acc: 0.9776 - precision_m: 0.9857 - recall_m: 0.9692 - val_loss: 0.0793 - val_acc: 0.9767 - val_precision_m: 0.9853 - val_recall_m: 0.9719\n",
      "Epoch 1363/5000\n",
      " - 10s - loss: 0.0735 - acc: 0.9759 - precision_m: 0.9852 - recall_m: 0.9663 - val_loss: 0.0890 - val_acc: 0.9651 - val_precision_m: 0.9860 - val_recall_m: 0.9497\n",
      "Epoch 1364/5000\n",
      " - 10s - loss: 0.0746 - acc: 0.9756 - precision_m: 0.9841 - recall_m: 0.9669 - val_loss: 0.0869 - val_acc: 0.9756 - val_precision_m: 0.9794 - val_recall_m: 0.9758\n",
      "Epoch 1365/5000\n",
      " - 10s - loss: 0.0696 - acc: 0.9779 - precision_m: 0.9854 - recall_m: 0.9704 - val_loss: 0.1082 - val_acc: 0.9684 - val_precision_m: 0.9577 - val_recall_m: 0.9863\n",
      "Epoch 1366/5000\n",
      " - 10s - loss: 0.0738 - acc: 0.9771 - precision_m: 0.9844 - recall_m: 0.9696 - val_loss: 0.0807 - val_acc: 0.9762 - val_precision_m: 0.9842 - val_recall_m: 0.9720\n",
      "Epoch 1367/5000\n",
      " - 10s - loss: 0.0729 - acc: 0.9773 - precision_m: 0.9862 - recall_m: 0.9683 - val_loss: 0.0858 - val_acc: 0.9690 - val_precision_m: 0.9850 - val_recall_m: 0.9578\n",
      "Epoch 1368/5000\n",
      " - 10s - loss: 0.0724 - acc: 0.9765 - precision_m: 0.9859 - recall_m: 0.9672 - val_loss: 0.0852 - val_acc: 0.9751 - val_precision_m: 0.9801 - val_recall_m: 0.9741\n",
      "Epoch 1369/5000\n",
      " - 10s - loss: 0.0760 - acc: 0.9741 - precision_m: 0.9833 - recall_m: 0.9647 - val_loss: 0.0884 - val_acc: 0.9729 - val_precision_m: 0.9755 - val_recall_m: 0.9750\n",
      "Epoch 1370/5000\n",
      " - 10s - loss: 0.0766 - acc: 0.9746 - precision_m: 0.9839 - recall_m: 0.9655 - val_loss: 0.0888 - val_acc: 0.9751 - val_precision_m: 0.9761 - val_recall_m: 0.9779\n",
      "Epoch 1371/5000\n",
      " - 10s - loss: 0.0851 - acc: 0.9717 - precision_m: 0.9823 - recall_m: 0.9618 - val_loss: 0.1058 - val_acc: 0.9695 - val_precision_m: 0.9594 - val_recall_m: 0.9861\n",
      "Epoch 1372/5000\n",
      " - 10s - loss: 0.0713 - acc: 0.9769 - precision_m: 0.9855 - recall_m: 0.9688 - val_loss: 0.0883 - val_acc: 0.9701 - val_precision_m: 0.9740 - val_recall_m: 0.9709\n",
      "Epoch 1373/5000\n",
      " - 10s - loss: 0.0712 - acc: 0.9777 - precision_m: 0.9849 - recall_m: 0.9699 - val_loss: 0.0844 - val_acc: 0.9684 - val_precision_m: 0.9862 - val_recall_m: 0.9553\n",
      "Epoch 1374/5000\n",
      " - 10s - loss: 0.0711 - acc: 0.9785 - precision_m: 0.9870 - recall_m: 0.9703 - val_loss: 0.0897 - val_acc: 0.9751 - val_precision_m: 0.9754 - val_recall_m: 0.9791\n",
      "Epoch 1375/5000\n",
      " - 10s - loss: 0.0758 - acc: 0.9742 - precision_m: 0.9836 - recall_m: 0.9648 - val_loss: 0.1046 - val_acc: 0.9668 - val_precision_m: 0.9554 - val_recall_m: 0.9852\n",
      "Epoch 1376/5000\n",
      " - 10s - loss: 0.0771 - acc: 0.9754 - precision_m: 0.9837 - recall_m: 0.9671 - val_loss: 0.0821 - val_acc: 0.9773 - val_precision_m: 0.9823 - val_recall_m: 0.9761\n",
      "Epoch 1377/5000\n",
      " - 10s - loss: 0.0723 - acc: 0.9767 - precision_m: 0.9858 - recall_m: 0.9674 - val_loss: 0.0887 - val_acc: 0.9712 - val_precision_m: 0.9704 - val_recall_m: 0.9770\n",
      "Epoch 1378/5000\n",
      " - 10s - loss: 0.0727 - acc: 0.9764 - precision_m: 0.9846 - recall_m: 0.9682 - val_loss: 0.0794 - val_acc: 0.9745 - val_precision_m: 0.9842 - val_recall_m: 0.9687\n",
      "Epoch 1379/5000\n",
      " - 10s - loss: 0.0738 - acc: 0.9759 - precision_m: 0.9839 - recall_m: 0.9678 - val_loss: 0.0866 - val_acc: 0.9767 - val_precision_m: 0.9793 - val_recall_m: 0.9782\n",
      "Epoch 1380/5000\n",
      " - 10s - loss: 0.0741 - acc: 0.9761 - precision_m: 0.9843 - recall_m: 0.9679 - val_loss: 0.0808 - val_acc: 0.9762 - val_precision_m: 0.9823 - val_recall_m: 0.9738\n",
      "Epoch 1381/5000\n",
      " - 10s - loss: 0.0818 - acc: 0.9729 - precision_m: 0.9829 - recall_m: 0.9631 - val_loss: 0.0937 - val_acc: 0.9618 - val_precision_m: 0.9870 - val_recall_m: 0.9422\n",
      "Epoch 1382/5000\n",
      " - 10s - loss: 0.0704 - acc: 0.9777 - precision_m: 0.9855 - recall_m: 0.9699 - val_loss: 0.0865 - val_acc: 0.9651 - val_precision_m: 0.9849 - val_recall_m: 0.9501\n",
      "Epoch 1383/5000\n",
      " - 10s - loss: 0.0727 - acc: 0.9758 - precision_m: 0.9848 - recall_m: 0.9669 - val_loss: 0.0827 - val_acc: 0.9762 - val_precision_m: 0.9775 - val_recall_m: 0.9791\n",
      "Epoch 1384/5000\n",
      " - 10s - loss: 0.0726 - acc: 0.9765 - precision_m: 0.9851 - recall_m: 0.9678 - val_loss: 0.0908 - val_acc: 0.9662 - val_precision_m: 0.9862 - val_recall_m: 0.9514\n",
      "Epoch 1385/5000\n",
      " - 10s - loss: 0.0702 - acc: 0.9778 - precision_m: 0.9859 - recall_m: 0.9694 - val_loss: 0.0879 - val_acc: 0.9718 - val_precision_m: 0.9811 - val_recall_m: 0.9667\n",
      "Epoch 1386/5000\n",
      " - 10s - loss: 0.0728 - acc: 0.9760 - precision_m: 0.9830 - recall_m: 0.9689 - val_loss: 0.0858 - val_acc: 0.9734 - val_precision_m: 0.9863 - val_recall_m: 0.9649\n",
      "Epoch 1387/5000\n",
      " - 10s - loss: 0.0699 - acc: 0.9778 - precision_m: 0.9853 - recall_m: 0.9702 - val_loss: 0.0894 - val_acc: 0.9723 - val_precision_m: 0.9762 - val_recall_m: 0.9733\n",
      "Epoch 1388/5000\n",
      " - 10s - loss: 0.0722 - acc: 0.9761 - precision_m: 0.9855 - recall_m: 0.9669 - val_loss: 0.0836 - val_acc: 0.9745 - val_precision_m: 0.9716 - val_recall_m: 0.9823\n",
      "Epoch 1389/5000\n",
      " - 10s - loss: 0.0705 - acc: 0.9777 - precision_m: 0.9863 - recall_m: 0.9689 - val_loss: 0.0859 - val_acc: 0.9712 - val_precision_m: 0.9725 - val_recall_m: 0.9752\n",
      "Epoch 1390/5000\n",
      " - 10s - loss: 0.0741 - acc: 0.9759 - precision_m: 0.9845 - recall_m: 0.9677 - val_loss: 0.1280 - val_acc: 0.9430 - val_precision_m: 0.9922 - val_recall_m: 0.9015\n",
      "Epoch 1391/5000\n",
      " - 10s - loss: 0.0741 - acc: 0.9744 - precision_m: 0.9840 - recall_m: 0.9647 - val_loss: 0.0852 - val_acc: 0.9745 - val_precision_m: 0.9794 - val_recall_m: 0.9738\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1392/5000\n",
      " - 10s - loss: 0.0701 - acc: 0.9774 - precision_m: 0.9852 - recall_m: 0.9691 - val_loss: 0.0828 - val_acc: 0.9756 - val_precision_m: 0.9832 - val_recall_m: 0.9719\n",
      "Epoch 1393/5000\n",
      " - 10s - loss: 0.0742 - acc: 0.9762 - precision_m: 0.9847 - recall_m: 0.9677 - val_loss: 0.0974 - val_acc: 0.9612 - val_precision_m: 0.9870 - val_recall_m: 0.9407\n",
      "Epoch 1394/5000\n",
      " - 10s - loss: 0.0774 - acc: 0.9754 - precision_m: 0.9845 - recall_m: 0.9667 - val_loss: 0.0837 - val_acc: 0.9723 - val_precision_m: 0.9801 - val_recall_m: 0.9689\n",
      "Epoch 1395/5000\n",
      " - 10s - loss: 0.0698 - acc: 0.9777 - precision_m: 0.9865 - recall_m: 0.9688 - val_loss: 0.0778 - val_acc: 0.9767 - val_precision_m: 0.9834 - val_recall_m: 0.9739\n",
      "Epoch 1396/5000\n",
      " - 10s - loss: 0.0679 - acc: 0.9777 - precision_m: 0.9849 - recall_m: 0.9702 - val_loss: 0.0979 - val_acc: 0.9601 - val_precision_m: 0.9901 - val_recall_m: 0.9355\n",
      "Epoch 1397/5000\n",
      " - 10s - loss: 0.0697 - acc: 0.9764 - precision_m: 0.9844 - recall_m: 0.9686 - val_loss: 0.0846 - val_acc: 0.9729 - val_precision_m: 0.9884 - val_recall_m: 0.9614\n",
      "Epoch 1398/5000\n",
      " - 10s - loss: 0.0684 - acc: 0.9778 - precision_m: 0.9867 - recall_m: 0.9689 - val_loss: 0.0848 - val_acc: 0.9695 - val_precision_m: 0.9831 - val_recall_m: 0.9608\n",
      "Epoch 1399/5000\n",
      " - 10s - loss: 0.0702 - acc: 0.9769 - precision_m: 0.9848 - recall_m: 0.9687 - val_loss: 0.0926 - val_acc: 0.9718 - val_precision_m: 0.9769 - val_recall_m: 0.9711\n",
      "Epoch 1400/5000\n",
      " - 10s - loss: 0.0698 - acc: 0.9778 - precision_m: 0.9864 - recall_m: 0.9693 - val_loss: 0.0815 - val_acc: 0.9762 - val_precision_m: 0.9785 - val_recall_m: 0.9781\n",
      "Epoch 1401/5000\n",
      " - 10s - loss: 0.0720 - acc: 0.9766 - precision_m: 0.9856 - recall_m: 0.9675 - val_loss: 0.0815 - val_acc: 0.9723 - val_precision_m: 0.9752 - val_recall_m: 0.9739\n",
      "Epoch 1402/5000\n",
      " - 10s - loss: 0.0679 - acc: 0.9769 - precision_m: 0.9850 - recall_m: 0.9687 - val_loss: 0.0834 - val_acc: 0.9729 - val_precision_m: 0.9812 - val_recall_m: 0.9691\n",
      "Epoch 1403/5000\n",
      " - 10s - loss: 0.0760 - acc: 0.9748 - precision_m: 0.9834 - recall_m: 0.9665 - val_loss: 0.1207 - val_acc: 0.9623 - val_precision_m: 0.9433 - val_recall_m: 0.9901\n",
      "Epoch 1404/5000\n",
      " - 10s - loss: 0.0707 - acc: 0.9776 - precision_m: 0.9867 - recall_m: 0.9679 - val_loss: 0.0833 - val_acc: 0.9695 - val_precision_m: 0.9840 - val_recall_m: 0.9595\n",
      "Epoch 1405/5000\n",
      " - 10s - loss: 0.0674 - acc: 0.9798 - precision_m: 0.9876 - recall_m: 0.9717 - val_loss: 0.0876 - val_acc: 0.9668 - val_precision_m: 0.9830 - val_recall_m: 0.9553\n",
      "Epoch 1406/5000\n",
      " - 10s - loss: 0.0667 - acc: 0.9791 - precision_m: 0.9856 - recall_m: 0.9723 - val_loss: 0.0822 - val_acc: 0.9762 - val_precision_m: 0.9727 - val_recall_m: 0.9842\n",
      "Epoch 1407/5000\n",
      " - 10s - loss: 0.0755 - acc: 0.9747 - precision_m: 0.9844 - recall_m: 0.9650 - val_loss: 0.0783 - val_acc: 0.9745 - val_precision_m: 0.9803 - val_recall_m: 0.9729\n",
      "Epoch 1408/5000\n",
      " - 10s - loss: 0.0722 - acc: 0.9754 - precision_m: 0.9832 - recall_m: 0.9678 - val_loss: 0.0886 - val_acc: 0.9679 - val_precision_m: 0.9840 - val_recall_m: 0.9568\n",
      "Epoch 1409/5000\n",
      " - 10s - loss: 0.0753 - acc: 0.9749 - precision_m: 0.9835 - recall_m: 0.9662 - val_loss: 0.1005 - val_acc: 0.9684 - val_precision_m: 0.9664 - val_recall_m: 0.9760\n",
      "Epoch 1410/5000\n",
      " - 10s - loss: 0.0735 - acc: 0.9762 - precision_m: 0.9847 - recall_m: 0.9679 - val_loss: 0.0805 - val_acc: 0.9751 - val_precision_m: 0.9822 - val_recall_m: 0.9720\n",
      "Epoch 1411/5000\n",
      " - 10s - loss: 0.0682 - acc: 0.9790 - precision_m: 0.9878 - recall_m: 0.9705 - val_loss: 0.0834 - val_acc: 0.9701 - val_precision_m: 0.9852 - val_recall_m: 0.9599\n",
      "Epoch 1412/5000\n",
      " - 10s - loss: 0.0670 - acc: 0.9783 - precision_m: 0.9855 - recall_m: 0.9709 - val_loss: 0.0848 - val_acc: 0.9745 - val_precision_m: 0.9793 - val_recall_m: 0.9740\n",
      "Epoch 1413/5000\n",
      " - 10s - loss: 0.0659 - acc: 0.9795 - precision_m: 0.9869 - recall_m: 0.9717 - val_loss: 0.0881 - val_acc: 0.9679 - val_precision_m: 0.9872 - val_recall_m: 0.9533\n",
      "Epoch 1414/5000\n",
      " - 10s - loss: 0.0656 - acc: 0.9798 - precision_m: 0.9865 - recall_m: 0.9729 - val_loss: 0.1065 - val_acc: 0.9524 - val_precision_m: 0.9901 - val_recall_m: 0.9215\n",
      "Epoch 1415/5000\n",
      " - 10s - loss: 0.0752 - acc: 0.9735 - precision_m: 0.9828 - recall_m: 0.9647 - val_loss: 0.1142 - val_acc: 0.9601 - val_precision_m: 0.9460 - val_recall_m: 0.9831\n",
      "Epoch 1416/5000\n",
      " - 10s - loss: 0.0722 - acc: 0.9762 - precision_m: 0.9846 - recall_m: 0.9677 - val_loss: 0.0932 - val_acc: 0.9618 - val_precision_m: 0.9911 - val_recall_m: 0.9381\n",
      "Epoch 1417/5000\n",
      " - 10s - loss: 0.0666 - acc: 0.9783 - precision_m: 0.9853 - recall_m: 0.9713 - val_loss: 0.0806 - val_acc: 0.9723 - val_precision_m: 0.9813 - val_recall_m: 0.9678\n",
      "Epoch 1418/5000\n",
      " - 10s - loss: 0.0691 - acc: 0.9773 - precision_m: 0.9855 - recall_m: 0.9692 - val_loss: 0.0815 - val_acc: 0.9729 - val_precision_m: 0.9842 - val_recall_m: 0.9658\n",
      "Epoch 1419/5000\n",
      " - 10s - loss: 0.0663 - acc: 0.9786 - precision_m: 0.9861 - recall_m: 0.9711 - val_loss: 0.1025 - val_acc: 0.9596 - val_precision_m: 0.9881 - val_recall_m: 0.9367\n",
      "Epoch 1420/5000\n",
      " - 10s - loss: 0.0700 - acc: 0.9773 - precision_m: 0.9852 - recall_m: 0.9691 - val_loss: 0.1949 - val_acc: 0.9120 - val_precision_m: 0.9976 - val_recall_m: 0.8400\n",
      "Epoch 1421/5000\n",
      " - 10s - loss: 0.0707 - acc: 0.9769 - precision_m: 0.9852 - recall_m: 0.9682 - val_loss: 0.0786 - val_acc: 0.9751 - val_precision_m: 0.9852 - val_recall_m: 0.9688\n",
      "Epoch 1422/5000\n",
      " - 10s - loss: 0.0683 - acc: 0.9780 - precision_m: 0.9873 - recall_m: 0.9688 - val_loss: 0.1249 - val_acc: 0.9424 - val_precision_m: 0.9909 - val_recall_m: 0.9018\n",
      "Epoch 1423/5000\n",
      " - 10s - loss: 0.0726 - acc: 0.9749 - precision_m: 0.9840 - recall_m: 0.9658 - val_loss: 0.0833 - val_acc: 0.9745 - val_precision_m: 0.9764 - val_recall_m: 0.9771\n",
      "Epoch 1424/5000\n",
      " - 10s - loss: 0.0707 - acc: 0.9771 - precision_m: 0.9854 - recall_m: 0.9686 - val_loss: 0.0945 - val_acc: 0.9707 - val_precision_m: 0.9619 - val_recall_m: 0.9853\n",
      "Epoch 1425/5000\n",
      " - 10s - loss: 0.0721 - acc: 0.9769 - precision_m: 0.9854 - recall_m: 0.9685 - val_loss: 0.0877 - val_acc: 0.9701 - val_precision_m: 0.9829 - val_recall_m: 0.9614\n",
      "Epoch 1426/5000\n",
      " - 10s - loss: 0.0684 - acc: 0.9772 - precision_m: 0.9860 - recall_m: 0.9681 - val_loss: 0.0832 - val_acc: 0.9723 - val_precision_m: 0.9873 - val_recall_m: 0.9617\n",
      "Epoch 1427/5000\n",
      " - 10s - loss: 0.0681 - acc: 0.9778 - precision_m: 0.9859 - recall_m: 0.9690 - val_loss: 0.0800 - val_acc: 0.9745 - val_precision_m: 0.9832 - val_recall_m: 0.9698\n",
      "Epoch 1428/5000\n",
      " - 10s - loss: 0.0665 - acc: 0.9791 - precision_m: 0.9867 - recall_m: 0.9715 - val_loss: 0.0884 - val_acc: 0.9701 - val_precision_m: 0.9832 - val_recall_m: 0.9616\n",
      "Epoch 1429/5000\n",
      " - 10s - loss: 0.0642 - acc: 0.9789 - precision_m: 0.9862 - recall_m: 0.9712 - val_loss: 0.0878 - val_acc: 0.9756 - val_precision_m: 0.9700 - val_recall_m: 0.9862\n",
      "Epoch 1430/5000\n",
      " - 10s - loss: 0.0725 - acc: 0.9763 - precision_m: 0.9844 - recall_m: 0.9684 - val_loss: 0.0807 - val_acc: 0.9734 - val_precision_m: 0.9812 - val_recall_m: 0.9698\n",
      "Epoch 1431/5000\n",
      " - 10s - loss: 0.0698 - acc: 0.9768 - precision_m: 0.9841 - recall_m: 0.9692 - val_loss: 0.0861 - val_acc: 0.9668 - val_precision_m: 0.9880 - val_recall_m: 0.9502\n",
      "Epoch 1432/5000\n",
      " - 10s - loss: 0.0670 - acc: 0.9777 - precision_m: 0.9852 - recall_m: 0.9699 - val_loss: 0.0869 - val_acc: 0.9734 - val_precision_m: 0.9696 - val_recall_m: 0.9823\n",
      "Epoch 1433/5000\n",
      " - 10s - loss: 0.0706 - acc: 0.9771 - precision_m: 0.9853 - recall_m: 0.9693 - val_loss: 0.0909 - val_acc: 0.9629 - val_precision_m: 0.9871 - val_recall_m: 0.9439\n",
      "Epoch 1434/5000\n",
      " - 10s - loss: 0.0701 - acc: 0.9768 - precision_m: 0.9858 - recall_m: 0.9669 - val_loss: 0.0755 - val_acc: 0.9751 - val_precision_m: 0.9821 - val_recall_m: 0.9717\n",
      "Epoch 1435/5000\n",
      " - 10s - loss: 0.0763 - acc: 0.9745 - precision_m: 0.9838 - recall_m: 0.9651 - val_loss: 0.0820 - val_acc: 0.9745 - val_precision_m: 0.9843 - val_recall_m: 0.9691\n",
      "Epoch 1436/5000\n",
      " - 10s - loss: 0.0643 - acc: 0.9796 - precision_m: 0.9874 - recall_m: 0.9718 - val_loss: 0.0867 - val_acc: 0.9673 - val_precision_m: 0.9912 - val_recall_m: 0.9480\n",
      "Epoch 1437/5000\n",
      " - 10s - loss: 0.0790 - acc: 0.9727 - precision_m: 0.9818 - recall_m: 0.9646 - val_loss: 0.1321 - val_acc: 0.9596 - val_precision_m: 0.9424 - val_recall_m: 0.9863\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1438/5000\n",
      " - 10s - loss: 0.0667 - acc: 0.9783 - precision_m: 0.9865 - recall_m: 0.9698 - val_loss: 0.0812 - val_acc: 0.9729 - val_precision_m: 0.9824 - val_recall_m: 0.9678\n",
      "Epoch 1439/5000\n",
      " - 10s - loss: 0.0654 - acc: 0.9788 - precision_m: 0.9862 - recall_m: 0.9714 - val_loss: 0.0847 - val_acc: 0.9679 - val_precision_m: 0.9891 - val_recall_m: 0.9514\n",
      "Epoch 1440/5000\n",
      " - 10s - loss: 0.0672 - acc: 0.9779 - precision_m: 0.9861 - recall_m: 0.9701 - val_loss: 0.0823 - val_acc: 0.9734 - val_precision_m: 0.9773 - val_recall_m: 0.9740\n",
      "Epoch 1441/5000\n",
      " - 10s - loss: 0.0709 - acc: 0.9775 - precision_m: 0.9857 - recall_m: 0.9693 - val_loss: 0.0847 - val_acc: 0.9729 - val_precision_m: 0.9873 - val_recall_m: 0.9626\n",
      "Epoch 1442/5000\n",
      " - 10s - loss: 0.0693 - acc: 0.9774 - precision_m: 0.9843 - recall_m: 0.9704 - val_loss: 0.0855 - val_acc: 0.9695 - val_precision_m: 0.9863 - val_recall_m: 0.9574\n",
      "Epoch 1443/5000\n",
      " - 10s - loss: 0.0668 - acc: 0.9786 - precision_m: 0.9861 - recall_m: 0.9711 - val_loss: 0.0791 - val_acc: 0.9751 - val_precision_m: 0.9737 - val_recall_m: 0.9812\n",
      "Epoch 1444/5000\n",
      " - 10s - loss: 0.0621 - acc: 0.9806 - precision_m: 0.9875 - recall_m: 0.9736 - val_loss: 0.0781 - val_acc: 0.9767 - val_precision_m: 0.9792 - val_recall_m: 0.9778\n",
      "Epoch 1445/5000\n",
      " - 10s - loss: 0.0662 - acc: 0.9785 - precision_m: 0.9848 - recall_m: 0.9722 - val_loss: 0.0784 - val_acc: 0.9756 - val_precision_m: 0.9842 - val_recall_m: 0.9708\n",
      "Epoch 1446/5000\n",
      " - 10s - loss: 0.0687 - acc: 0.9785 - precision_m: 0.9863 - recall_m: 0.9707 - val_loss: 0.0807 - val_acc: 0.9762 - val_precision_m: 0.9745 - val_recall_m: 0.9822\n",
      "Epoch 1447/5000\n",
      " - 10s - loss: 0.0652 - acc: 0.9785 - precision_m: 0.9856 - recall_m: 0.9714 - val_loss: 0.0788 - val_acc: 0.9740 - val_precision_m: 0.9812 - val_recall_m: 0.9709\n",
      "Epoch 1448/5000\n",
      " - 10s - loss: 0.0656 - acc: 0.9783 - precision_m: 0.9860 - recall_m: 0.9707 - val_loss: 0.0912 - val_acc: 0.9635 - val_precision_m: 0.9892 - val_recall_m: 0.9432\n",
      "Epoch 1449/5000\n",
      " - 10s - loss: 0.0695 - acc: 0.9769 - precision_m: 0.9850 - recall_m: 0.9690 - val_loss: 0.1026 - val_acc: 0.9607 - val_precision_m: 0.9891 - val_recall_m: 0.9381\n",
      "Epoch 1450/5000\n",
      " - 10s - loss: 0.0675 - acc: 0.9784 - precision_m: 0.9856 - recall_m: 0.9712 - val_loss: 0.0852 - val_acc: 0.9734 - val_precision_m: 0.9706 - val_recall_m: 0.9812\n",
      "Epoch 1451/5000\n",
      " - 10s - loss: 0.0709 - acc: 0.9778 - precision_m: 0.9850 - recall_m: 0.9706 - val_loss: 0.0826 - val_acc: 0.9718 - val_precision_m: 0.9883 - val_recall_m: 0.9596\n",
      "Epoch 1452/5000\n",
      " - 10s - loss: 0.0622 - acc: 0.9801 - precision_m: 0.9872 - recall_m: 0.9731 - val_loss: 0.0873 - val_acc: 0.9629 - val_precision_m: 0.9890 - val_recall_m: 0.9420\n",
      "Epoch 1453/5000\n",
      " - 10s - loss: 0.0686 - acc: 0.9769 - precision_m: 0.9852 - recall_m: 0.9688 - val_loss: 0.0825 - val_acc: 0.9690 - val_precision_m: 0.9863 - val_recall_m: 0.9565\n",
      "Epoch 1454/5000\n",
      " - 10s - loss: 0.0633 - acc: 0.9796 - precision_m: 0.9885 - recall_m: 0.9710 - val_loss: 0.0813 - val_acc: 0.9773 - val_precision_m: 0.9775 - val_recall_m: 0.9811\n",
      "Epoch 1455/5000\n",
      " - 10s - loss: 0.0675 - acc: 0.9780 - precision_m: 0.9862 - recall_m: 0.9697 - val_loss: 0.0746 - val_acc: 0.9773 - val_precision_m: 0.9832 - val_recall_m: 0.9749\n",
      "Epoch 1456/5000\n",
      " - 10s - loss: 0.0645 - acc: 0.9795 - precision_m: 0.9867 - recall_m: 0.9725 - val_loss: 0.0791 - val_acc: 0.9773 - val_precision_m: 0.9744 - val_recall_m: 0.9841\n",
      "Epoch 1457/5000\n",
      " - 10s - loss: 0.0699 - acc: 0.9780 - precision_m: 0.9850 - recall_m: 0.9709 - val_loss: 0.0786 - val_acc: 0.9734 - val_precision_m: 0.9883 - val_recall_m: 0.9628\n",
      "Epoch 1458/5000\n",
      " - 10s - loss: 0.0673 - acc: 0.9781 - precision_m: 0.9859 - recall_m: 0.9707 - val_loss: 0.0884 - val_acc: 0.9712 - val_precision_m: 0.9630 - val_recall_m: 0.9852\n",
      "Epoch 1459/5000\n",
      " - 10s - loss: 0.0703 - acc: 0.9775 - precision_m: 0.9854 - recall_m: 0.9696 - val_loss: 0.0877 - val_acc: 0.9745 - val_precision_m: 0.9661 - val_recall_m: 0.9883\n",
      "Epoch 1460/5000\n",
      " - 10s - loss: 0.0655 - acc: 0.9786 - precision_m: 0.9861 - recall_m: 0.9710 - val_loss: 0.0804 - val_acc: 0.9695 - val_precision_m: 0.9830 - val_recall_m: 0.9608\n",
      "Epoch 1461/5000\n",
      " - 10s - loss: 0.0631 - acc: 0.9807 - precision_m: 0.9888 - recall_m: 0.9727 - val_loss: 0.0996 - val_acc: 0.9679 - val_precision_m: 0.9562 - val_recall_m: 0.9861\n",
      "Epoch 1462/5000\n",
      " - 10s - loss: 0.0622 - acc: 0.9804 - precision_m: 0.9870 - recall_m: 0.9737 - val_loss: 0.0750 - val_acc: 0.9762 - val_precision_m: 0.9764 - val_recall_m: 0.9800\n",
      "Epoch 1463/5000\n",
      " - 10s - loss: 0.0606 - acc: 0.9809 - precision_m: 0.9872 - recall_m: 0.9746 - val_loss: 0.0833 - val_acc: 0.9684 - val_precision_m: 0.9883 - val_recall_m: 0.9534\n",
      "Epoch 1464/5000\n",
      " - 10s - loss: 0.0635 - acc: 0.9798 - precision_m: 0.9861 - recall_m: 0.9733 - val_loss: 0.0934 - val_acc: 0.9740 - val_precision_m: 0.9685 - val_recall_m: 0.9840\n",
      "Epoch 1465/5000\n",
      " - 10s - loss: 0.0654 - acc: 0.9777 - precision_m: 0.9858 - recall_m: 0.9697 - val_loss: 0.0736 - val_acc: 0.9767 - val_precision_m: 0.9845 - val_recall_m: 0.9728\n",
      "Epoch 1466/5000\n",
      " - 10s - loss: 0.0680 - acc: 0.9781 - precision_m: 0.9854 - recall_m: 0.9707 - val_loss: 0.0757 - val_acc: 0.9773 - val_precision_m: 0.9863 - val_recall_m: 0.9720\n",
      "Epoch 1467/5000\n",
      " - 10s - loss: 0.0615 - acc: 0.9806 - precision_m: 0.9883 - recall_m: 0.9726 - val_loss: 0.0704 - val_acc: 0.9767 - val_precision_m: 0.9845 - val_recall_m: 0.9729\n",
      "Epoch 1468/5000\n",
      " - 10s - loss: 0.0669 - acc: 0.9777 - precision_m: 0.9841 - recall_m: 0.9715 - val_loss: 0.0753 - val_acc: 0.9756 - val_precision_m: 0.9863 - val_recall_m: 0.9687\n",
      "Epoch 1469/5000\n",
      " - 10s - loss: 0.0628 - acc: 0.9794 - precision_m: 0.9861 - recall_m: 0.9726 - val_loss: 0.0774 - val_acc: 0.9767 - val_precision_m: 0.9813 - val_recall_m: 0.9761\n",
      "Epoch 1470/5000\n",
      " - 10s - loss: 0.0677 - acc: 0.9783 - precision_m: 0.9854 - recall_m: 0.9717 - val_loss: 0.0863 - val_acc: 0.9784 - val_precision_m: 0.9755 - val_recall_m: 0.9851\n",
      "Epoch 1471/5000\n",
      " - 10s - loss: 0.0651 - acc: 0.9780 - precision_m: 0.9858 - recall_m: 0.9699 - val_loss: 0.0779 - val_acc: 0.9690 - val_precision_m: 0.9873 - val_recall_m: 0.9554\n",
      "Epoch 1472/5000\n",
      " - 10s - loss: 0.0686 - acc: 0.9781 - precision_m: 0.9847 - recall_m: 0.9717 - val_loss: 0.0828 - val_acc: 0.9762 - val_precision_m: 0.9756 - val_recall_m: 0.9811\n",
      "Epoch 1473/5000\n",
      " - 10s - loss: 0.0685 - acc: 0.9783 - precision_m: 0.9858 - recall_m: 0.9705 - val_loss: 0.0728 - val_acc: 0.9751 - val_precision_m: 0.9841 - val_recall_m: 0.9696\n",
      "Epoch 1474/5000\n",
      " - 10s - loss: 0.0591 - acc: 0.9813 - precision_m: 0.9882 - recall_m: 0.9744 - val_loss: 0.0850 - val_acc: 0.9734 - val_precision_m: 0.9678 - val_recall_m: 0.9842\n",
      "Epoch 1475/5000\n",
      " - 10s - loss: 0.0583 - acc: 0.9822 - precision_m: 0.9885 - recall_m: 0.9760 - val_loss: 0.0752 - val_acc: 0.9767 - val_precision_m: 0.9865 - val_recall_m: 0.9708\n",
      "Epoch 1476/5000\n",
      " - 10s - loss: 0.0627 - acc: 0.9798 - precision_m: 0.9879 - recall_m: 0.9718 - val_loss: 0.0733 - val_acc: 0.9790 - val_precision_m: 0.9774 - val_recall_m: 0.9840\n",
      "Epoch 1477/5000\n",
      " - 10s - loss: 0.0615 - acc: 0.9798 - precision_m: 0.9867 - recall_m: 0.9731 - val_loss: 0.0810 - val_acc: 0.9756 - val_precision_m: 0.9735 - val_recall_m: 0.9821\n",
      "Epoch 1478/5000\n",
      " - 10s - loss: 0.0590 - acc: 0.9807 - precision_m: 0.9863 - recall_m: 0.9750 - val_loss: 0.0741 - val_acc: 0.9779 - val_precision_m: 0.9844 - val_recall_m: 0.9749\n",
      "Epoch 1479/5000\n",
      " - 10s - loss: 0.0614 - acc: 0.9793 - precision_m: 0.9860 - recall_m: 0.9722 - val_loss: 0.0900 - val_acc: 0.9668 - val_precision_m: 0.9811 - val_recall_m: 0.9575\n",
      "Epoch 1480/5000\n",
      " - 10s - loss: 0.0658 - acc: 0.9790 - precision_m: 0.9860 - recall_m: 0.9718 - val_loss: 0.0844 - val_acc: 0.9657 - val_precision_m: 0.9830 - val_recall_m: 0.9537\n",
      "Epoch 1481/5000\n",
      " - 10s - loss: 0.0660 - acc: 0.9792 - precision_m: 0.9866 - recall_m: 0.9720 - val_loss: 0.0808 - val_acc: 0.9707 - val_precision_m: 0.9883 - val_recall_m: 0.9576\n",
      "Epoch 1482/5000\n",
      " - 10s - loss: 0.0651 - acc: 0.9781 - precision_m: 0.9857 - recall_m: 0.9710 - val_loss: 0.0856 - val_acc: 0.9751 - val_precision_m: 0.9707 - val_recall_m: 0.9841\n",
      "Epoch 1483/5000\n",
      " - 10s - loss: 0.0611 - acc: 0.9806 - precision_m: 0.9879 - recall_m: 0.9734 - val_loss: 0.0698 - val_acc: 0.9806 - val_precision_m: 0.9874 - val_recall_m: 0.9769\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1484/5000\n",
      " - 10s - loss: 0.0621 - acc: 0.9796 - precision_m: 0.9865 - recall_m: 0.9725 - val_loss: 0.0896 - val_acc: 0.9623 - val_precision_m: 0.9913 - val_recall_m: 0.9390\n",
      "Epoch 1485/5000\n",
      " - 10s - loss: 0.0605 - acc: 0.9807 - precision_m: 0.9863 - recall_m: 0.9750 - val_loss: 0.0734 - val_acc: 0.9806 - val_precision_m: 0.9823 - val_recall_m: 0.9822\n",
      "Epoch 1486/5000\n",
      " - 10s - loss: 0.0632 - acc: 0.9799 - precision_m: 0.9868 - recall_m: 0.9729 - val_loss: 0.0966 - val_acc: 0.9623 - val_precision_m: 0.9924 - val_recall_m: 0.9377\n",
      "Epoch 1487/5000\n",
      " - 10s - loss: 0.0613 - acc: 0.9799 - precision_m: 0.9870 - recall_m: 0.9728 - val_loss: 0.0743 - val_acc: 0.9773 - val_precision_m: 0.9833 - val_recall_m: 0.9753\n",
      "Epoch 1488/5000\n",
      " - 10s - loss: 0.0647 - acc: 0.9780 - precision_m: 0.9864 - recall_m: 0.9696 - val_loss: 0.0901 - val_acc: 0.9740 - val_precision_m: 0.9732 - val_recall_m: 0.9789\n",
      "Epoch 1489/5000\n",
      " - 10s - loss: 0.0638 - acc: 0.9798 - precision_m: 0.9865 - recall_m: 0.9734 - val_loss: 0.0977 - val_acc: 0.9635 - val_precision_m: 0.9870 - val_recall_m: 0.9450\n",
      "Epoch 1490/5000\n",
      " - 10s - loss: 0.0607 - acc: 0.9804 - precision_m: 0.9877 - recall_m: 0.9730 - val_loss: 0.0853 - val_acc: 0.9646 - val_precision_m: 0.9871 - val_recall_m: 0.9469\n",
      "Epoch 1491/5000\n",
      " - 10s - loss: 0.0614 - acc: 0.9805 - precision_m: 0.9875 - recall_m: 0.9737 - val_loss: 0.0829 - val_acc: 0.9762 - val_precision_m: 0.9734 - val_recall_m: 0.9831\n",
      "Epoch 1492/5000\n",
      " - 10s - loss: 0.0611 - acc: 0.9799 - precision_m: 0.9858 - recall_m: 0.9737 - val_loss: 0.0713 - val_acc: 0.9806 - val_precision_m: 0.9885 - val_recall_m: 0.9759\n",
      "Epoch 1493/5000\n",
      " - 10s - loss: 0.0612 - acc: 0.9804 - precision_m: 0.9877 - recall_m: 0.9729 - val_loss: 0.0727 - val_acc: 0.9762 - val_precision_m: 0.9873 - val_recall_m: 0.9688\n",
      "Epoch 1494/5000\n",
      " - 10s - loss: 0.0613 - acc: 0.9803 - precision_m: 0.9880 - recall_m: 0.9727 - val_loss: 0.0743 - val_acc: 0.9745 - val_precision_m: 0.9873 - val_recall_m: 0.9659\n",
      "Epoch 1495/5000\n",
      " - 10s - loss: 0.0621 - acc: 0.9794 - precision_m: 0.9858 - recall_m: 0.9730 - val_loss: 0.1018 - val_acc: 0.9585 - val_precision_m: 0.9922 - val_recall_m: 0.9305\n",
      "Epoch 1496/5000\n",
      " - 10s - loss: 0.0610 - acc: 0.9802 - precision_m: 0.9861 - recall_m: 0.9742 - val_loss: 0.0873 - val_acc: 0.9657 - val_precision_m: 0.9882 - val_recall_m: 0.9484\n",
      "Epoch 1497/5000\n",
      " - 10s - loss: 0.0678 - acc: 0.9777 - precision_m: 0.9854 - recall_m: 0.9701 - val_loss: 0.0745 - val_acc: 0.9734 - val_precision_m: 0.9841 - val_recall_m: 0.9671\n",
      "Epoch 1498/5000\n",
      " - 10s - loss: 0.0665 - acc: 0.9791 - precision_m: 0.9868 - recall_m: 0.9720 - val_loss: 0.1018 - val_acc: 0.9502 - val_precision_m: 0.9922 - val_recall_m: 0.9155\n",
      "Epoch 1499/5000\n",
      " - 10s - loss: 0.0662 - acc: 0.9780 - precision_m: 0.9861 - recall_m: 0.9702 - val_loss: 0.0761 - val_acc: 0.9745 - val_precision_m: 0.9844 - val_recall_m: 0.9688\n",
      "Epoch 1500/5000\n",
      " - 10s - loss: 0.0638 - acc: 0.9800 - precision_m: 0.9883 - recall_m: 0.9716 - val_loss: 0.0720 - val_acc: 0.9756 - val_precision_m: 0.9803 - val_recall_m: 0.9750\n",
      "Epoch 1501/5000\n",
      " - 10s - loss: 0.0722 - acc: 0.9746 - precision_m: 0.9828 - recall_m: 0.9664 - val_loss: 0.0758 - val_acc: 0.9767 - val_precision_m: 0.9793 - val_recall_m: 0.9782\n",
      "Epoch 1502/5000\n",
      " - 10s - loss: 0.0608 - acc: 0.9809 - precision_m: 0.9873 - recall_m: 0.9746 - val_loss: 0.0722 - val_acc: 0.9767 - val_precision_m: 0.9815 - val_recall_m: 0.9761\n",
      "Epoch 1503/5000\n",
      " - 10s - loss: 0.0616 - acc: 0.9794 - precision_m: 0.9862 - recall_m: 0.9728 - val_loss: 0.0761 - val_acc: 0.9751 - val_precision_m: 0.9833 - val_recall_m: 0.9709\n",
      "Epoch 1504/5000\n",
      " - 10s - loss: 0.0587 - acc: 0.9807 - precision_m: 0.9879 - recall_m: 0.9731 - val_loss: 0.0737 - val_acc: 0.9751 - val_precision_m: 0.9772 - val_recall_m: 0.9769\n",
      "Epoch 1505/5000\n",
      " - 10s - loss: 0.0670 - acc: 0.9774 - precision_m: 0.9860 - recall_m: 0.9687 - val_loss: 0.1443 - val_acc: 0.9551 - val_precision_m: 0.9292 - val_recall_m: 0.9932\n",
      "Epoch 1506/5000\n",
      " - 10s - loss: 0.0620 - acc: 0.9815 - precision_m: 0.9872 - recall_m: 0.9759 - val_loss: 0.0964 - val_acc: 0.9563 - val_precision_m: 0.9901 - val_recall_m: 0.9283\n",
      "Epoch 1507/5000\n",
      " - 10s - loss: 0.0575 - acc: 0.9814 - precision_m: 0.9875 - recall_m: 0.9751 - val_loss: 0.0799 - val_acc: 0.9718 - val_precision_m: 0.9904 - val_recall_m: 0.9576\n",
      "Epoch 1508/5000\n",
      " - 10s - loss: 0.0621 - acc: 0.9795 - precision_m: 0.9873 - recall_m: 0.9718 - val_loss: 0.0719 - val_acc: 0.9806 - val_precision_m: 0.9834 - val_recall_m: 0.9808\n",
      "Epoch 1509/5000\n",
      " - 10s - loss: 0.0606 - acc: 0.9801 - precision_m: 0.9863 - recall_m: 0.9738 - val_loss: 0.0684 - val_acc: 0.9801 - val_precision_m: 0.9843 - val_recall_m: 0.9792\n",
      "Epoch 1510/5000\n",
      " - 10s - loss: 0.0605 - acc: 0.9799 - precision_m: 0.9876 - recall_m: 0.9719 - val_loss: 0.0694 - val_acc: 0.9790 - val_precision_m: 0.9803 - val_recall_m: 0.9812\n",
      "Epoch 1511/5000\n",
      " - 10s - loss: 0.0600 - acc: 0.9806 - precision_m: 0.9868 - recall_m: 0.9743 - val_loss: 0.0782 - val_acc: 0.9712 - val_precision_m: 0.9891 - val_recall_m: 0.9578\n",
      "Epoch 1512/5000\n",
      " - 10s - loss: 0.0645 - acc: 0.9783 - precision_m: 0.9863 - recall_m: 0.9706 - val_loss: 0.0810 - val_acc: 0.9712 - val_precision_m: 0.9892 - val_recall_m: 0.9573\n",
      "Epoch 1513/5000\n",
      " - 10s - loss: 0.0592 - acc: 0.9806 - precision_m: 0.9876 - recall_m: 0.9737 - val_loss: 0.0755 - val_acc: 0.9779 - val_precision_m: 0.9774 - val_recall_m: 0.9821\n",
      "Epoch 1514/5000\n",
      " - 10s - loss: 0.0589 - acc: 0.9813 - precision_m: 0.9874 - recall_m: 0.9750 - val_loss: 0.0736 - val_acc: 0.9779 - val_precision_m: 0.9755 - val_recall_m: 0.9842\n",
      "Epoch 1515/5000\n",
      " - 10s - loss: 0.0578 - acc: 0.9823 - precision_m: 0.9892 - recall_m: 0.9753 - val_loss: 0.0953 - val_acc: 0.9585 - val_precision_m: 0.9946 - val_recall_m: 0.9283\n",
      "Epoch 1516/5000\n",
      " - 10s - loss: 0.0622 - acc: 0.9803 - precision_m: 0.9867 - recall_m: 0.9740 - val_loss: 0.0719 - val_acc: 0.9779 - val_precision_m: 0.9756 - val_recall_m: 0.9843\n",
      "Epoch 1517/5000\n",
      " - 10s - loss: 0.0578 - acc: 0.9815 - precision_m: 0.9880 - recall_m: 0.9751 - val_loss: 0.0793 - val_acc: 0.9790 - val_precision_m: 0.9765 - val_recall_m: 0.9851\n",
      "Epoch 1518/5000\n",
      " - 10s - loss: 0.0573 - acc: 0.9813 - precision_m: 0.9879 - recall_m: 0.9744 - val_loss: 0.0686 - val_acc: 0.9801 - val_precision_m: 0.9854 - val_recall_m: 0.9781\n",
      "Epoch 1519/5000\n",
      " - 10s - loss: 0.0618 - acc: 0.9795 - precision_m: 0.9871 - recall_m: 0.9720 - val_loss: 0.0789 - val_acc: 0.9751 - val_precision_m: 0.9812 - val_recall_m: 0.9729\n",
      "Epoch 1520/5000\n",
      " - 10s - loss: 0.0646 - acc: 0.9791 - precision_m: 0.9866 - recall_m: 0.9718 - val_loss: 0.0775 - val_acc: 0.9773 - val_precision_m: 0.9736 - val_recall_m: 0.9851\n",
      "Epoch 1521/5000\n",
      " - 10s - loss: 0.0588 - acc: 0.9815 - precision_m: 0.9883 - recall_m: 0.9745 - val_loss: 0.0789 - val_acc: 0.9701 - val_precision_m: 0.9741 - val_recall_m: 0.9707\n",
      "Epoch 1522/5000\n",
      " - 10s - loss: 0.0560 - acc: 0.9822 - precision_m: 0.9883 - recall_m: 0.9761 - val_loss: 0.0660 - val_acc: 0.9806 - val_precision_m: 0.9834 - val_recall_m: 0.9812\n",
      "Epoch 1523/5000\n",
      " - 10s - loss: 0.0626 - acc: 0.9798 - precision_m: 0.9872 - recall_m: 0.9725 - val_loss: 0.0873 - val_acc: 0.9745 - val_precision_m: 0.9696 - val_recall_m: 0.9839\n",
      "Epoch 1524/5000\n",
      " - 10s - loss: 0.0620 - acc: 0.9789 - precision_m: 0.9854 - recall_m: 0.9727 - val_loss: 0.0691 - val_acc: 0.9751 - val_precision_m: 0.9851 - val_recall_m: 0.9692\n",
      "Epoch 1525/5000\n",
      " - 10s - loss: 0.0597 - acc: 0.9810 - precision_m: 0.9874 - recall_m: 0.9745 - val_loss: 0.0714 - val_acc: 0.9762 - val_precision_m: 0.9863 - val_recall_m: 0.9698\n",
      "Epoch 1526/5000\n",
      " - 10s - loss: 0.0589 - acc: 0.9811 - precision_m: 0.9877 - recall_m: 0.9743 - val_loss: 0.0774 - val_acc: 0.9740 - val_precision_m: 0.9884 - val_recall_m: 0.9639\n",
      "Epoch 1527/5000\n",
      " - 10s - loss: 0.0591 - acc: 0.9805 - precision_m: 0.9878 - recall_m: 0.9731 - val_loss: 0.0723 - val_acc: 0.9773 - val_precision_m: 0.9825 - val_recall_m: 0.9759\n",
      "Epoch 1528/5000\n",
      " - 10s - loss: 0.0685 - acc: 0.9762 - precision_m: 0.9847 - recall_m: 0.9677 - val_loss: 0.0753 - val_acc: 0.9740 - val_precision_m: 0.9776 - val_recall_m: 0.9749\n",
      "Epoch 1529/5000\n",
      " - 10s - loss: 0.0586 - acc: 0.9809 - precision_m: 0.9872 - recall_m: 0.9747 - val_loss: 0.0927 - val_acc: 0.9701 - val_precision_m: 0.9566 - val_recall_m: 0.9901\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1530/5000\n",
      " - 10s - loss: 0.0567 - acc: 0.9814 - precision_m: 0.9876 - recall_m: 0.9751 - val_loss: 0.0689 - val_acc: 0.9790 - val_precision_m: 0.9784 - val_recall_m: 0.9831\n",
      "Epoch 1531/5000\n",
      " - 10s - loss: 0.0598 - acc: 0.9806 - precision_m: 0.9875 - recall_m: 0.9735 - val_loss: 0.1069 - val_acc: 0.9690 - val_precision_m: 0.9555 - val_recall_m: 0.9892\n",
      "Epoch 1532/5000\n",
      " - 10s - loss: 0.0700 - acc: 0.9762 - precision_m: 0.9840 - recall_m: 0.9686 - val_loss: 0.0725 - val_acc: 0.9723 - val_precision_m: 0.9883 - val_recall_m: 0.9609\n",
      "Epoch 1533/5000\n",
      " - 10s - loss: 0.0580 - acc: 0.9827 - precision_m: 0.9895 - recall_m: 0.9759 - val_loss: 0.0793 - val_acc: 0.9751 - val_precision_m: 0.9855 - val_recall_m: 0.9686\n",
      "Epoch 1534/5000\n",
      " - 10s - loss: 0.0581 - acc: 0.9811 - precision_m: 0.9879 - recall_m: 0.9740 - val_loss: 0.0795 - val_acc: 0.9695 - val_precision_m: 0.9872 - val_recall_m: 0.9567\n",
      "Epoch 1535/5000\n",
      " - 10s - loss: 0.0579 - acc: 0.9814 - precision_m: 0.9884 - recall_m: 0.9739 - val_loss: 0.0793 - val_acc: 0.9740 - val_precision_m: 0.9811 - val_recall_m: 0.9710\n",
      "Epoch 1536/5000\n",
      " - 10s - loss: 0.0553 - acc: 0.9827 - precision_m: 0.9885 - recall_m: 0.9768 - val_loss: 0.0671 - val_acc: 0.9784 - val_precision_m: 0.9815 - val_recall_m: 0.9790\n",
      "Epoch 1537/5000\n",
      " - 10s - loss: 0.0539 - acc: 0.9829 - precision_m: 0.9896 - recall_m: 0.9758 - val_loss: 0.0738 - val_acc: 0.9723 - val_precision_m: 0.9861 - val_recall_m: 0.9628\n",
      "Epoch 1538/5000\n",
      " - 10s - loss: 0.0618 - acc: 0.9793 - precision_m: 0.9853 - recall_m: 0.9731 - val_loss: 0.1290 - val_acc: 0.9441 - val_precision_m: 0.9967 - val_recall_m: 0.8997\n",
      "Epoch 1539/5000\n",
      " - 10s - loss: 0.0623 - acc: 0.9799 - precision_m: 0.9868 - recall_m: 0.9733 - val_loss: 0.0828 - val_acc: 0.9701 - val_precision_m: 0.9861 - val_recall_m: 0.9594\n",
      "Epoch 1540/5000\n",
      " - 10s - loss: 0.0594 - acc: 0.9809 - precision_m: 0.9883 - recall_m: 0.9731 - val_loss: 0.0729 - val_acc: 0.9790 - val_precision_m: 0.9785 - val_recall_m: 0.9831\n",
      "Epoch 1541/5000\n",
      " - 10s - loss: 0.0576 - acc: 0.9819 - precision_m: 0.9887 - recall_m: 0.9753 - val_loss: 0.0792 - val_acc: 0.9767 - val_precision_m: 0.9715 - val_recall_m: 0.9861\n",
      "Epoch 1542/5000\n",
      " - 10s - loss: 0.0658 - acc: 0.9779 - precision_m: 0.9839 - recall_m: 0.9723 - val_loss: 0.0964 - val_acc: 0.9607 - val_precision_m: 0.9911 - val_recall_m: 0.9354\n",
      "Epoch 1543/5000\n",
      " - 10s - loss: 0.0594 - acc: 0.9807 - precision_m: 0.9872 - recall_m: 0.9741 - val_loss: 0.0797 - val_acc: 0.9673 - val_precision_m: 0.9902 - val_recall_m: 0.9493\n",
      "Epoch 1544/5000\n",
      " - 10s - loss: 0.0578 - acc: 0.9814 - precision_m: 0.9886 - recall_m: 0.9742 - val_loss: 0.1086 - val_acc: 0.9529 - val_precision_m: 0.9956 - val_recall_m: 0.9175\n",
      "Epoch 1545/5000\n",
      " - 10s - loss: 0.0605 - acc: 0.9809 - precision_m: 0.9877 - recall_m: 0.9743 - val_loss: 0.0722 - val_acc: 0.9756 - val_precision_m: 0.9774 - val_recall_m: 0.9779\n",
      "Epoch 1546/5000\n",
      " - 10s - loss: 0.0614 - acc: 0.9795 - precision_m: 0.9857 - recall_m: 0.9732 - val_loss: 0.0703 - val_acc: 0.9740 - val_precision_m: 0.9882 - val_recall_m: 0.9640\n",
      "Epoch 1547/5000\n",
      " - 10s - loss: 0.0605 - acc: 0.9799 - precision_m: 0.9864 - recall_m: 0.9737 - val_loss: 0.0736 - val_acc: 0.9767 - val_precision_m: 0.9893 - val_recall_m: 0.9677\n",
      "Epoch 1548/5000\n",
      " - 10s - loss: 0.0545 - acc: 0.9826 - precision_m: 0.9889 - recall_m: 0.9761 - val_loss: 0.0733 - val_acc: 0.9773 - val_precision_m: 0.9726 - val_recall_m: 0.9861\n",
      "Epoch 1549/5000\n",
      " - 10s - loss: 0.0568 - acc: 0.9818 - precision_m: 0.9874 - recall_m: 0.9762 - val_loss: 0.0710 - val_acc: 0.9745 - val_precision_m: 0.9884 - val_recall_m: 0.9651\n",
      "Epoch 1550/5000\n",
      " - 10s - loss: 0.0556 - acc: 0.9827 - precision_m: 0.9892 - recall_m: 0.9760 - val_loss: 0.0733 - val_acc: 0.9801 - val_precision_m: 0.9787 - val_recall_m: 0.9852\n",
      "Epoch 1551/5000\n",
      " - 10s - loss: 0.0659 - acc: 0.9773 - precision_m: 0.9858 - recall_m: 0.9689 - val_loss: 0.0726 - val_acc: 0.9734 - val_precision_m: 0.9902 - val_recall_m: 0.9605\n",
      "Epoch 1552/5000\n",
      " - 10s - loss: 0.0589 - acc: 0.9810 - precision_m: 0.9883 - recall_m: 0.9738 - val_loss: 0.0696 - val_acc: 0.9751 - val_precision_m: 0.9873 - val_recall_m: 0.9668\n",
      "Epoch 1553/5000\n",
      " - 10s - loss: 0.0599 - acc: 0.9807 - precision_m: 0.9870 - recall_m: 0.9742 - val_loss: 0.0719 - val_acc: 0.9767 - val_precision_m: 0.9735 - val_recall_m: 0.9841\n",
      "Epoch 1554/5000\n",
      " - 10s - loss: 0.0662 - acc: 0.9772 - precision_m: 0.9851 - recall_m: 0.9695 - val_loss: 0.0627 - val_acc: 0.9795 - val_precision_m: 0.9865 - val_recall_m: 0.9761\n",
      "Epoch 1555/5000\n",
      " - 10s - loss: 0.0569 - acc: 0.9828 - precision_m: 0.9881 - recall_m: 0.9772 - val_loss: 0.0631 - val_acc: 0.9817 - val_precision_m: 0.9834 - val_recall_m: 0.9832\n",
      "Epoch 1556/5000\n",
      " - 10s - loss: 0.0553 - acc: 0.9815 - precision_m: 0.9883 - recall_m: 0.9747 - val_loss: 0.0658 - val_acc: 0.9779 - val_precision_m: 0.9832 - val_recall_m: 0.9760\n",
      "Epoch 1557/5000\n",
      " - 10s - loss: 0.0528 - acc: 0.9838 - precision_m: 0.9889 - recall_m: 0.9790 - val_loss: 0.0675 - val_acc: 0.9795 - val_precision_m: 0.9864 - val_recall_m: 0.9758\n",
      "Epoch 1558/5000\n",
      " - 10s - loss: 0.0598 - acc: 0.9808 - precision_m: 0.9880 - recall_m: 0.9738 - val_loss: 0.0724 - val_acc: 0.9767 - val_precision_m: 0.9894 - val_recall_m: 0.9678\n",
      "Epoch 1559/5000\n",
      " - 10s - loss: 0.0529 - acc: 0.9836 - precision_m: 0.9895 - recall_m: 0.9776 - val_loss: 0.0657 - val_acc: 0.9795 - val_precision_m: 0.9824 - val_recall_m: 0.9802\n",
      "Epoch 1560/5000\n",
      " - 10s - loss: 0.0570 - acc: 0.9817 - precision_m: 0.9882 - recall_m: 0.9751 - val_loss: 0.0692 - val_acc: 0.9767 - val_precision_m: 0.9773 - val_recall_m: 0.9802\n",
      "Epoch 1561/5000\n",
      " - 10s - loss: 0.0560 - acc: 0.9823 - precision_m: 0.9884 - recall_m: 0.9765 - val_loss: 0.0690 - val_acc: 0.9762 - val_precision_m: 0.9813 - val_recall_m: 0.9752\n",
      "Epoch 1562/5000\n",
      " - 10s - loss: 0.0551 - acc: 0.9831 - precision_m: 0.9890 - recall_m: 0.9771 - val_loss: 0.0785 - val_acc: 0.9707 - val_precision_m: 0.9893 - val_recall_m: 0.9566\n",
      "Epoch 1563/5000\n",
      " - 10s - loss: 0.0510 - acc: 0.9853 - precision_m: 0.9907 - recall_m: 0.9797 - val_loss: 0.0645 - val_acc: 0.9779 - val_precision_m: 0.9814 - val_recall_m: 0.9781\n",
      "Epoch 1564/5000\n",
      " - 10s - loss: 0.0606 - acc: 0.9796 - precision_m: 0.9870 - recall_m: 0.9724 - val_loss: 0.0815 - val_acc: 0.9718 - val_precision_m: 0.9859 - val_recall_m: 0.9616\n",
      "Epoch 1565/5000\n",
      " - 10s - loss: 0.0524 - acc: 0.9832 - precision_m: 0.9890 - recall_m: 0.9773 - val_loss: 0.0701 - val_acc: 0.9762 - val_precision_m: 0.9865 - val_recall_m: 0.9699\n",
      "Epoch 1566/5000\n",
      " - 10s - loss: 0.0573 - acc: 0.9814 - precision_m: 0.9882 - recall_m: 0.9747 - val_loss: 0.0668 - val_acc: 0.9812 - val_precision_m: 0.9836 - val_recall_m: 0.9822\n",
      "Epoch 1567/5000\n",
      " - 10s - loss: 0.0571 - acc: 0.9812 - precision_m: 0.9874 - recall_m: 0.9749 - val_loss: 0.0746 - val_acc: 0.9718 - val_precision_m: 0.9924 - val_recall_m: 0.9557\n",
      "Epoch 1568/5000\n",
      " - 10s - loss: 0.0564 - acc: 0.9826 - precision_m: 0.9884 - recall_m: 0.9768 - val_loss: 0.0733 - val_acc: 0.9773 - val_precision_m: 0.9763 - val_recall_m: 0.9823\n",
      "Epoch 1569/5000\n",
      " - 10s - loss: 0.0603 - acc: 0.9807 - precision_m: 0.9870 - recall_m: 0.9739 - val_loss: 0.0696 - val_acc: 0.9751 - val_precision_m: 0.9773 - val_recall_m: 0.9769\n",
      "Epoch 1570/5000\n",
      " - 10s - loss: 0.0542 - acc: 0.9825 - precision_m: 0.9880 - recall_m: 0.9768 - val_loss: 0.0744 - val_acc: 0.9740 - val_precision_m: 0.9687 - val_recall_m: 0.9841\n",
      "Epoch 1571/5000\n",
      " - 10s - loss: 0.0559 - acc: 0.9820 - precision_m: 0.9884 - recall_m: 0.9756 - val_loss: 0.0704 - val_acc: 0.9740 - val_precision_m: 0.9843 - val_recall_m: 0.9679\n",
      "Epoch 1572/5000\n",
      " - 10s - loss: 0.0547 - acc: 0.9817 - precision_m: 0.9876 - recall_m: 0.9756 - val_loss: 0.0700 - val_acc: 0.9762 - val_precision_m: 0.9824 - val_recall_m: 0.9739\n",
      "Epoch 1573/5000\n",
      " - 10s - loss: 0.0535 - acc: 0.9828 - precision_m: 0.9895 - recall_m: 0.9760 - val_loss: 0.0686 - val_acc: 0.9767 - val_precision_m: 0.9854 - val_recall_m: 0.9719\n",
      "Epoch 1574/5000\n",
      " - 10s - loss: 0.0574 - acc: 0.9817 - precision_m: 0.9880 - recall_m: 0.9754 - val_loss: 0.0659 - val_acc: 0.9801 - val_precision_m: 0.9805 - val_recall_m: 0.9831\n",
      "Epoch 1575/5000\n",
      " - 10s - loss: 0.0563 - acc: 0.9822 - precision_m: 0.9887 - recall_m: 0.9760 - val_loss: 0.0695 - val_acc: 0.9773 - val_precision_m: 0.9855 - val_recall_m: 0.9732\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1576/5000\n",
      " - 10s - loss: 0.0530 - acc: 0.9836 - precision_m: 0.9898 - recall_m: 0.9774 - val_loss: 0.0840 - val_acc: 0.9651 - val_precision_m: 0.9945 - val_recall_m: 0.9409\n",
      "Epoch 1577/5000\n",
      " - 10s - loss: 0.0562 - acc: 0.9815 - precision_m: 0.9873 - recall_m: 0.9757 - val_loss: 0.0679 - val_acc: 0.9751 - val_precision_m: 0.9852 - val_recall_m: 0.9686\n",
      "Epoch 1578/5000\n",
      " - 10s - loss: 0.0606 - acc: 0.9798 - precision_m: 0.9878 - recall_m: 0.9716 - val_loss: 0.1434 - val_acc: 0.9529 - val_precision_m: 0.9254 - val_recall_m: 0.9931\n",
      "Epoch 1579/5000\n",
      " - 10s - loss: 0.0540 - acc: 0.9831 - precision_m: 0.9887 - recall_m: 0.9776 - val_loss: 0.0696 - val_acc: 0.9745 - val_precision_m: 0.9864 - val_recall_m: 0.9667\n",
      "Epoch 1580/5000\n",
      " - 10s - loss: 0.0552 - acc: 0.9823 - precision_m: 0.9883 - recall_m: 0.9765 - val_loss: 0.0722 - val_acc: 0.9734 - val_precision_m: 0.9862 - val_recall_m: 0.9648\n",
      "Epoch 1581/5000\n",
      " - 10s - loss: 0.0523 - acc: 0.9846 - precision_m: 0.9900 - recall_m: 0.9794 - val_loss: 0.0643 - val_acc: 0.9795 - val_precision_m: 0.9884 - val_recall_m: 0.9738\n",
      "Epoch 1582/5000\n",
      " - 10s - loss: 0.0557 - acc: 0.9823 - precision_m: 0.9887 - recall_m: 0.9758 - val_loss: 0.0687 - val_acc: 0.9795 - val_precision_m: 0.9805 - val_recall_m: 0.9819\n",
      "Epoch 1583/5000\n",
      " - 10s - loss: 0.0535 - acc: 0.9828 - precision_m: 0.9889 - recall_m: 0.9767 - val_loss: 0.0994 - val_acc: 0.9551 - val_precision_m: 0.9933 - val_recall_m: 0.9234\n",
      "Epoch 1584/5000\n",
      " - 10s - loss: 0.0541 - acc: 0.9818 - precision_m: 0.9883 - recall_m: 0.9753 - val_loss: 0.0680 - val_acc: 0.9767 - val_precision_m: 0.9728 - val_recall_m: 0.9851\n",
      "Epoch 1585/5000\n",
      " - 10s - loss: 0.0550 - acc: 0.9821 - precision_m: 0.9877 - recall_m: 0.9768 - val_loss: 0.0668 - val_acc: 0.9795 - val_precision_m: 0.9803 - val_recall_m: 0.9824\n",
      "Epoch 1586/5000\n",
      " - 10s - loss: 0.0569 - acc: 0.9820 - precision_m: 0.9879 - recall_m: 0.9760 - val_loss: 0.0748 - val_acc: 0.9740 - val_precision_m: 0.9735 - val_recall_m: 0.9788\n",
      "Epoch 1587/5000\n",
      " - 10s - loss: 0.0545 - acc: 0.9812 - precision_m: 0.9880 - recall_m: 0.9745 - val_loss: 0.0696 - val_acc: 0.9790 - val_precision_m: 0.9755 - val_recall_m: 0.9861\n",
      "Epoch 1588/5000\n",
      " - 10s - loss: 0.0588 - acc: 0.9803 - precision_m: 0.9866 - recall_m: 0.9738 - val_loss: 0.0722 - val_acc: 0.9784 - val_precision_m: 0.9743 - val_recall_m: 0.9861\n",
      "Epoch 1589/5000\n",
      " - 10s - loss: 0.0643 - acc: 0.9781 - precision_m: 0.9852 - recall_m: 0.9714 - val_loss: 0.0684 - val_acc: 0.9756 - val_precision_m: 0.9802 - val_recall_m: 0.9752\n",
      "Epoch 1590/5000\n",
      " - 10s - loss: 0.0533 - acc: 0.9834 - precision_m: 0.9896 - recall_m: 0.9771 - val_loss: 0.0736 - val_acc: 0.9784 - val_precision_m: 0.9735 - val_recall_m: 0.9870\n",
      "Epoch 1591/5000\n",
      " - 10s - loss: 0.0500 - acc: 0.9847 - precision_m: 0.9902 - recall_m: 0.9792 - val_loss: 0.0652 - val_acc: 0.9790 - val_precision_m: 0.9814 - val_recall_m: 0.9801\n",
      "Epoch 1592/5000\n",
      " - 10s - loss: 0.0506 - acc: 0.9845 - precision_m: 0.9899 - recall_m: 0.9790 - val_loss: 0.1086 - val_acc: 0.9518 - val_precision_m: 0.9921 - val_recall_m: 0.9177\n",
      "Epoch 1593/5000\n",
      " - 10s - loss: 0.0521 - acc: 0.9841 - precision_m: 0.9903 - recall_m: 0.9779 - val_loss: 0.0898 - val_acc: 0.9623 - val_precision_m: 0.9957 - val_recall_m: 0.9350\n",
      "Epoch 1594/5000\n",
      " - 10s - loss: 0.0568 - acc: 0.9806 - precision_m: 0.9866 - recall_m: 0.9747 - val_loss: 0.0718 - val_acc: 0.9779 - val_precision_m: 0.9688 - val_recall_m: 0.9912\n",
      "Epoch 1595/5000\n",
      " - 10s - loss: 0.0571 - acc: 0.9831 - precision_m: 0.9872 - recall_m: 0.9787 - val_loss: 0.0637 - val_acc: 0.9779 - val_precision_m: 0.9853 - val_recall_m: 0.9741\n",
      "Epoch 1596/5000\n",
      " - 10s - loss: 0.0496 - acc: 0.9841 - precision_m: 0.9896 - recall_m: 0.9784 - val_loss: 0.0677 - val_acc: 0.9773 - val_precision_m: 0.9707 - val_recall_m: 0.9882\n",
      "Epoch 1597/5000\n",
      " - 10s - loss: 0.0526 - acc: 0.9842 - precision_m: 0.9896 - recall_m: 0.9791 - val_loss: 0.0675 - val_acc: 0.9795 - val_precision_m: 0.9806 - val_recall_m: 0.9821\n",
      "Epoch 1598/5000\n",
      " - 10s - loss: 0.0510 - acc: 0.9839 - precision_m: 0.9893 - recall_m: 0.9785 - val_loss: 0.0631 - val_acc: 0.9779 - val_precision_m: 0.9803 - val_recall_m: 0.9790\n",
      "Epoch 1599/5000\n",
      " - 10s - loss: 0.0504 - acc: 0.9839 - precision_m: 0.9883 - recall_m: 0.9794 - val_loss: 0.0715 - val_acc: 0.9801 - val_precision_m: 0.9765 - val_recall_m: 0.9871\n",
      "Epoch 1600/5000\n",
      " - 10s - loss: 0.0596 - acc: 0.9800 - precision_m: 0.9863 - recall_m: 0.9736 - val_loss: 0.0670 - val_acc: 0.9756 - val_precision_m: 0.9873 - val_recall_m: 0.9679\n",
      "Epoch 1601/5000\n",
      " - 10s - loss: 0.0597 - acc: 0.9806 - precision_m: 0.9867 - recall_m: 0.9745 - val_loss: 0.0840 - val_acc: 0.9673 - val_precision_m: 0.9957 - val_recall_m: 0.9441\n",
      "Epoch 1602/5000\n",
      " - 10s - loss: 0.0516 - acc: 0.9839 - precision_m: 0.9883 - recall_m: 0.9796 - val_loss: 0.0704 - val_acc: 0.9740 - val_precision_m: 0.9882 - val_recall_m: 0.9636\n",
      "Epoch 1603/5000\n",
      " - 10s - loss: 0.0530 - acc: 0.9834 - precision_m: 0.9891 - recall_m: 0.9774 - val_loss: 0.0625 - val_acc: 0.9817 - val_precision_m: 0.9813 - val_recall_m: 0.9851\n",
      "Epoch 1604/5000\n",
      " - 10s - loss: 0.0502 - acc: 0.9849 - precision_m: 0.9907 - recall_m: 0.9786 - val_loss: 0.0677 - val_acc: 0.9795 - val_precision_m: 0.9894 - val_recall_m: 0.9731\n",
      "Epoch 1605/5000\n",
      " - 10s - loss: 0.0513 - acc: 0.9840 - precision_m: 0.9897 - recall_m: 0.9781 - val_loss: 0.0686 - val_acc: 0.9751 - val_precision_m: 0.9873 - val_recall_m: 0.9667\n",
      "Epoch 1606/5000\n",
      " - 10s - loss: 0.0555 - acc: 0.9821 - precision_m: 0.9887 - recall_m: 0.9755 - val_loss: 0.0748 - val_acc: 0.9695 - val_precision_m: 0.9937 - val_recall_m: 0.9504\n",
      "Epoch 1607/5000\n",
      " - 10s - loss: 0.0616 - acc: 0.9801 - precision_m: 0.9868 - recall_m: 0.9731 - val_loss: 0.0633 - val_acc: 0.9790 - val_precision_m: 0.9853 - val_recall_m: 0.9760\n",
      "Epoch 1608/5000\n",
      " - 10s - loss: 0.0520 - acc: 0.9834 - precision_m: 0.9874 - recall_m: 0.9793 - val_loss: 0.0678 - val_acc: 0.9756 - val_precision_m: 0.9893 - val_recall_m: 0.9657\n",
      "Epoch 1609/5000\n",
      " - 10s - loss: 0.0498 - acc: 0.9838 - precision_m: 0.9898 - recall_m: 0.9777 - val_loss: 0.0633 - val_acc: 0.9801 - val_precision_m: 0.9843 - val_recall_m: 0.9789\n",
      "Epoch 1610/5000\n",
      " - 10s - loss: 0.0592 - acc: 0.9807 - precision_m: 0.9866 - recall_m: 0.9748 - val_loss: 0.0720 - val_acc: 0.9745 - val_precision_m: 0.9843 - val_recall_m: 0.9689\n",
      "Epoch 1611/5000\n",
      " - 10s - loss: 0.0534 - acc: 0.9838 - precision_m: 0.9893 - recall_m: 0.9782 - val_loss: 0.0651 - val_acc: 0.9806 - val_precision_m: 0.9894 - val_recall_m: 0.9748\n",
      "Epoch 1612/5000\n",
      " - 10s - loss: 0.0535 - acc: 0.9826 - precision_m: 0.9877 - recall_m: 0.9777 - val_loss: 0.0648 - val_acc: 0.9806 - val_precision_m: 0.9865 - val_recall_m: 0.9782\n",
      "Epoch 1613/5000\n",
      " - 10s - loss: 0.0533 - acc: 0.9831 - precision_m: 0.9886 - recall_m: 0.9777 - val_loss: 0.0801 - val_acc: 0.9695 - val_precision_m: 0.9923 - val_recall_m: 0.9515\n",
      "Epoch 1614/5000\n",
      " - 10s - loss: 0.0563 - acc: 0.9817 - precision_m: 0.9874 - recall_m: 0.9760 - val_loss: 0.0982 - val_acc: 0.9684 - val_precision_m: 0.9506 - val_recall_m: 0.9930\n",
      "Epoch 1615/5000\n",
      " - 10s - loss: 0.0521 - acc: 0.9851 - precision_m: 0.9900 - recall_m: 0.9803 - val_loss: 0.0615 - val_acc: 0.9812 - val_precision_m: 0.9854 - val_recall_m: 0.9800\n",
      "Epoch 1616/5000\n",
      " - 10s - loss: 0.0514 - acc: 0.9842 - precision_m: 0.9905 - recall_m: 0.9779 - val_loss: 0.0765 - val_acc: 0.9701 - val_precision_m: 0.9854 - val_recall_m: 0.9597\n",
      "Epoch 1617/5000\n",
      " - 10s - loss: 0.0492 - acc: 0.9854 - precision_m: 0.9909 - recall_m: 0.9800 - val_loss: 0.0634 - val_acc: 0.9767 - val_precision_m: 0.9895 - val_recall_m: 0.9677\n",
      "Epoch 1618/5000\n",
      " - 10s - loss: 0.0488 - acc: 0.9849 - precision_m: 0.9892 - recall_m: 0.9807 - val_loss: 0.0610 - val_acc: 0.9823 - val_precision_m: 0.9874 - val_recall_m: 0.9801\n",
      "Epoch 1619/5000\n",
      " - 10s - loss: 0.0557 - acc: 0.9818 - precision_m: 0.9876 - recall_m: 0.9762 - val_loss: 0.0642 - val_acc: 0.9779 - val_precision_m: 0.9814 - val_recall_m: 0.9779\n",
      "Epoch 1620/5000\n",
      " - 10s - loss: 0.0542 - acc: 0.9834 - precision_m: 0.9888 - recall_m: 0.9782 - val_loss: 0.0589 - val_acc: 0.9812 - val_precision_m: 0.9853 - val_recall_m: 0.9802\n",
      "Epoch 1621/5000\n",
      " - 10s - loss: 0.0493 - acc: 0.9846 - precision_m: 0.9888 - recall_m: 0.9804 - val_loss: 0.0597 - val_acc: 0.9812 - val_precision_m: 0.9804 - val_recall_m: 0.9852\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1622/5000\n",
      " - 10s - loss: 0.0511 - acc: 0.9830 - precision_m: 0.9885 - recall_m: 0.9776 - val_loss: 0.0609 - val_acc: 0.9784 - val_precision_m: 0.9874 - val_recall_m: 0.9730\n",
      "Epoch 1623/5000\n",
      " - 10s - loss: 0.0497 - acc: 0.9844 - precision_m: 0.9896 - recall_m: 0.9792 - val_loss: 0.0588 - val_acc: 0.9812 - val_precision_m: 0.9884 - val_recall_m: 0.9772\n",
      "Epoch 1624/5000\n",
      " - 10s - loss: 0.0488 - acc: 0.9852 - precision_m: 0.9914 - recall_m: 0.9790 - val_loss: 0.0604 - val_acc: 0.9767 - val_precision_m: 0.9864 - val_recall_m: 0.9710\n",
      "Epoch 1625/5000\n",
      " - 10s - loss: 0.0481 - acc: 0.9855 - precision_m: 0.9908 - recall_m: 0.9804 - val_loss: 0.0836 - val_acc: 0.9751 - val_precision_m: 0.9650 - val_recall_m: 0.9901\n",
      "Epoch 1626/5000\n",
      " - 10s - loss: 0.0553 - acc: 0.9832 - precision_m: 0.9890 - recall_m: 0.9774 - val_loss: 0.0655 - val_acc: 0.9817 - val_precision_m: 0.9787 - val_recall_m: 0.9881\n",
      "Epoch 1627/5000\n",
      " - 10s - loss: 0.0536 - acc: 0.9823 - precision_m: 0.9882 - recall_m: 0.9764 - val_loss: 0.0632 - val_acc: 0.9790 - val_precision_m: 0.9855 - val_recall_m: 0.9761\n",
      "Epoch 1628/5000\n",
      " - 10s - loss: 0.0543 - acc: 0.9826 - precision_m: 0.9868 - recall_m: 0.9784 - val_loss: 0.0663 - val_acc: 0.9784 - val_precision_m: 0.9825 - val_recall_m: 0.9779\n",
      "Epoch 1629/5000\n",
      " - 10s - loss: 0.0595 - acc: 0.9802 - precision_m: 0.9874 - recall_m: 0.9728 - val_loss: 0.0645 - val_acc: 0.9801 - val_precision_m: 0.9824 - val_recall_m: 0.9811\n",
      "Epoch 1630/5000\n",
      " - 10s - loss: 0.0514 - acc: 0.9840 - precision_m: 0.9887 - recall_m: 0.9791 - val_loss: 0.0645 - val_acc: 0.9823 - val_precision_m: 0.9804 - val_recall_m: 0.9872\n",
      "Epoch 1631/5000\n",
      " - 10s - loss: 0.0525 - acc: 0.9841 - precision_m: 0.9893 - recall_m: 0.9789 - val_loss: 0.0675 - val_acc: 0.9751 - val_precision_m: 0.9707 - val_recall_m: 0.9841\n",
      "Epoch 1632/5000\n",
      " - 10s - loss: 0.0537 - acc: 0.9828 - precision_m: 0.9884 - recall_m: 0.9772 - val_loss: 0.0834 - val_acc: 0.9679 - val_precision_m: 0.9903 - val_recall_m: 0.9504\n",
      "Epoch 1633/5000\n",
      " - 10s - loss: 0.0519 - acc: 0.9837 - precision_m: 0.9883 - recall_m: 0.9788 - val_loss: 0.0715 - val_acc: 0.9729 - val_precision_m: 0.9875 - val_recall_m: 0.9629\n",
      "Epoch 1634/5000\n",
      " - 10s - loss: 0.0498 - acc: 0.9838 - precision_m: 0.9898 - recall_m: 0.9778 - val_loss: 0.0647 - val_acc: 0.9773 - val_precision_m: 0.9893 - val_recall_m: 0.9689\n",
      "Epoch 1635/5000\n",
      " - 10s - loss: 0.0490 - acc: 0.9849 - precision_m: 0.9896 - recall_m: 0.9802 - val_loss: 0.0587 - val_acc: 0.9806 - val_precision_m: 0.9795 - val_recall_m: 0.9851\n",
      "Epoch 1636/5000\n",
      " - 10s - loss: 0.0523 - acc: 0.9832 - precision_m: 0.9888 - recall_m: 0.9776 - val_loss: 0.0693 - val_acc: 0.9779 - val_precision_m: 0.9734 - val_recall_m: 0.9860\n",
      "Epoch 1637/5000\n",
      " - 10s - loss: 0.0523 - acc: 0.9834 - precision_m: 0.9891 - recall_m: 0.9779 - val_loss: 0.0621 - val_acc: 0.9806 - val_precision_m: 0.9862 - val_recall_m: 0.9780\n",
      "Epoch 1638/5000\n",
      " - 10s - loss: 0.0554 - acc: 0.9821 - precision_m: 0.9877 - recall_m: 0.9766 - val_loss: 0.0579 - val_acc: 0.9806 - val_precision_m: 0.9816 - val_recall_m: 0.9831\n",
      "Epoch 1639/5000\n",
      " - 10s - loss: 0.0511 - acc: 0.9838 - precision_m: 0.9885 - recall_m: 0.9790 - val_loss: 0.0696 - val_acc: 0.9723 - val_precision_m: 0.9873 - val_recall_m: 0.9616\n",
      "Epoch 1640/5000\n",
      " - 10s - loss: 0.0543 - acc: 0.9830 - precision_m: 0.9893 - recall_m: 0.9767 - val_loss: 0.0818 - val_acc: 0.9662 - val_precision_m: 0.9946 - val_recall_m: 0.9429\n",
      "Epoch 1641/5000\n",
      " - 10s - loss: 0.0547 - acc: 0.9817 - precision_m: 0.9872 - recall_m: 0.9766 - val_loss: 0.0784 - val_acc: 0.9673 - val_precision_m: 0.9924 - val_recall_m: 0.9472\n",
      "Epoch 1642/5000\n",
      " - 10s - loss: 0.0505 - acc: 0.9844 - precision_m: 0.9895 - recall_m: 0.9793 - val_loss: 0.0738 - val_acc: 0.9673 - val_precision_m: 0.9957 - val_recall_m: 0.9438\n",
      "Epoch 1643/5000\n",
      " - 10s - loss: 0.0508 - acc: 0.9843 - precision_m: 0.9902 - recall_m: 0.9785 - val_loss: 0.0807 - val_acc: 0.9673 - val_precision_m: 0.9957 - val_recall_m: 0.9442\n",
      "Epoch 1644/5000\n",
      " - 10s - loss: 0.0466 - acc: 0.9853 - precision_m: 0.9895 - recall_m: 0.9810 - val_loss: 0.0613 - val_acc: 0.9779 - val_precision_m: 0.9814 - val_recall_m: 0.9778\n",
      "Epoch 1645/5000\n",
      " - 10s - loss: 0.0494 - acc: 0.9845 - precision_m: 0.9891 - recall_m: 0.9799 - val_loss: 0.0628 - val_acc: 0.9806 - val_precision_m: 0.9766 - val_recall_m: 0.9880\n",
      "Epoch 1646/5000\n",
      " - 10s - loss: 0.0556 - acc: 0.9815 - precision_m: 0.9866 - recall_m: 0.9766 - val_loss: 0.1392 - val_acc: 0.9374 - val_precision_m: 0.9955 - val_recall_m: 0.8882\n",
      "Epoch 1647/5000\n",
      " - 10s - loss: 0.0525 - acc: 0.9828 - precision_m: 0.9882 - recall_m: 0.9775 - val_loss: 0.0604 - val_acc: 0.9790 - val_precision_m: 0.9884 - val_recall_m: 0.9729\n",
      "Epoch 1648/5000\n",
      " - 10s - loss: 0.0466 - acc: 0.9857 - precision_m: 0.9905 - recall_m: 0.9809 - val_loss: 0.0709 - val_acc: 0.9790 - val_precision_m: 0.9720 - val_recall_m: 0.9901\n",
      "Epoch 1649/5000\n",
      " - 10s - loss: 0.0506 - acc: 0.9842 - precision_m: 0.9892 - recall_m: 0.9791 - val_loss: 0.0871 - val_acc: 0.9646 - val_precision_m: 0.9968 - val_recall_m: 0.9378\n",
      "Epoch 1650/5000\n",
      " - 10s - loss: 0.0509 - acc: 0.9836 - precision_m: 0.9895 - recall_m: 0.9776 - val_loss: 0.0614 - val_acc: 0.9806 - val_precision_m: 0.9815 - val_recall_m: 0.9832\n",
      "Epoch 1651/5000\n",
      " - 10s - loss: 0.0453 - acc: 0.9860 - precision_m: 0.9907 - recall_m: 0.9813 - val_loss: 0.0674 - val_acc: 0.9767 - val_precision_m: 0.9777 - val_recall_m: 0.9804\n",
      "Epoch 1652/5000\n",
      " - 10s - loss: 0.0514 - acc: 0.9839 - precision_m: 0.9895 - recall_m: 0.9783 - val_loss: 0.0795 - val_acc: 0.9751 - val_precision_m: 0.9636 - val_recall_m: 0.9921\n",
      "Epoch 1653/5000\n",
      " - 10s - loss: 0.0540 - acc: 0.9820 - precision_m: 0.9878 - recall_m: 0.9763 - val_loss: 0.0676 - val_acc: 0.9767 - val_precision_m: 0.9914 - val_recall_m: 0.9660\n",
      "Epoch 1654/5000\n",
      " - 10s - loss: 0.0538 - acc: 0.9825 - precision_m: 0.9889 - recall_m: 0.9758 - val_loss: 0.1001 - val_acc: 0.9668 - val_precision_m: 0.9511 - val_recall_m: 0.9901\n",
      "Epoch 1655/5000\n",
      " - 10s - loss: 0.0535 - acc: 0.9827 - precision_m: 0.9890 - recall_m: 0.9769 - val_loss: 0.0666 - val_acc: 0.9801 - val_precision_m: 0.9775 - val_recall_m: 0.9861\n",
      "Epoch 1656/5000\n",
      " - 10s - loss: 0.0466 - acc: 0.9852 - precision_m: 0.9901 - recall_m: 0.9806 - val_loss: 0.0743 - val_acc: 0.9767 - val_precision_m: 0.9717 - val_recall_m: 0.9861\n",
      "Epoch 1657/5000\n",
      " - 10s - loss: 0.0525 - acc: 0.9832 - precision_m: 0.9883 - recall_m: 0.9780 - val_loss: 0.0673 - val_acc: 0.9767 - val_precision_m: 0.9895 - val_recall_m: 0.9678\n",
      "Epoch 1658/5000\n",
      " - 10s - loss: 0.0443 - acc: 0.9866 - precision_m: 0.9916 - recall_m: 0.9813 - val_loss: 0.0675 - val_acc: 0.9751 - val_precision_m: 0.9925 - val_recall_m: 0.9617\n",
      "Epoch 1659/5000\n",
      " - 10s - loss: 0.0501 - acc: 0.9849 - precision_m: 0.9904 - recall_m: 0.9793 - val_loss: 0.0847 - val_acc: 0.9740 - val_precision_m: 0.9623 - val_recall_m: 0.9912\n",
      "Epoch 1660/5000\n",
      " - 10s - loss: 0.0507 - acc: 0.9838 - precision_m: 0.9878 - recall_m: 0.9798 - val_loss: 0.0613 - val_acc: 0.9779 - val_precision_m: 0.9883 - val_recall_m: 0.9710\n",
      "Epoch 1661/5000\n",
      " - 10s - loss: 0.0502 - acc: 0.9852 - precision_m: 0.9903 - recall_m: 0.9801 - val_loss: 0.0564 - val_acc: 0.9828 - val_precision_m: 0.9856 - val_recall_m: 0.9830\n",
      "Epoch 1662/5000\n",
      " - 10s - loss: 0.0514 - acc: 0.9833 - precision_m: 0.9882 - recall_m: 0.9785 - val_loss: 0.0613 - val_acc: 0.9795 - val_precision_m: 0.9834 - val_recall_m: 0.9792\n",
      "Epoch 1663/5000\n",
      " - 10s - loss: 0.0522 - acc: 0.9840 - precision_m: 0.9896 - recall_m: 0.9787 - val_loss: 0.0842 - val_acc: 0.9712 - val_precision_m: 0.9595 - val_recall_m: 0.9890\n",
      "Epoch 1664/5000\n",
      " - 10s - loss: 0.0442 - acc: 0.9865 - precision_m: 0.9905 - recall_m: 0.9827 - val_loss: 0.0552 - val_acc: 0.9817 - val_precision_m: 0.9855 - val_recall_m: 0.9813\n",
      "Epoch 1665/5000\n",
      " - 10s - loss: 0.0531 - acc: 0.9828 - precision_m: 0.9879 - recall_m: 0.9778 - val_loss: 0.0713 - val_acc: 0.9784 - val_precision_m: 0.9696 - val_recall_m: 0.9911\n",
      "Epoch 1666/5000\n",
      " - 10s - loss: 0.0538 - acc: 0.9826 - precision_m: 0.9883 - recall_m: 0.9771 - val_loss: 0.0578 - val_acc: 0.9784 - val_precision_m: 0.9863 - val_recall_m: 0.9741\n",
      "Epoch 1667/5000\n",
      " - 10s - loss: 0.0484 - acc: 0.9855 - precision_m: 0.9906 - recall_m: 0.9803 - val_loss: 0.0635 - val_acc: 0.9745 - val_precision_m: 0.9873 - val_recall_m: 0.9658\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1668/5000\n",
      " - 10s - loss: 0.0493 - acc: 0.9832 - precision_m: 0.9892 - recall_m: 0.9772 - val_loss: 0.0742 - val_acc: 0.9784 - val_precision_m: 0.9704 - val_recall_m: 0.9901\n",
      "Epoch 1669/5000\n",
      " - 10s - loss: 0.0483 - acc: 0.9842 - precision_m: 0.9890 - recall_m: 0.9793 - val_loss: 0.0776 - val_acc: 0.9740 - val_precision_m: 0.9613 - val_recall_m: 0.9920\n",
      "Epoch 1670/5000\n",
      " - 10s - loss: 0.0486 - acc: 0.9848 - precision_m: 0.9897 - recall_m: 0.9799 - val_loss: 0.0586 - val_acc: 0.9795 - val_precision_m: 0.9844 - val_recall_m: 0.9783\n",
      "Epoch 1671/5000\n",
      " - 10s - loss: 0.0481 - acc: 0.9847 - precision_m: 0.9895 - recall_m: 0.9801 - val_loss: 0.0595 - val_acc: 0.9828 - val_precision_m: 0.9816 - val_recall_m: 0.9870\n",
      "Epoch 1672/5000\n",
      " - 10s - loss: 0.0456 - acc: 0.9854 - precision_m: 0.9898 - recall_m: 0.9809 - val_loss: 0.0603 - val_acc: 0.9784 - val_precision_m: 0.9823 - val_recall_m: 0.9781\n",
      "Epoch 1673/5000\n",
      " - 10s - loss: 0.0459 - acc: 0.9857 - precision_m: 0.9903 - recall_m: 0.9812 - val_loss: 0.0573 - val_acc: 0.9828 - val_precision_m: 0.9814 - val_recall_m: 0.9871\n",
      "Epoch 1674/5000\n",
      " - 10s - loss: 0.0524 - acc: 0.9838 - precision_m: 0.9884 - recall_m: 0.9789 - val_loss: 0.0772 - val_acc: 0.9751 - val_precision_m: 0.9650 - val_recall_m: 0.9901\n",
      "Epoch 1675/5000\n",
      " - 10s - loss: 0.0516 - acc: 0.9841 - precision_m: 0.9898 - recall_m: 0.9785 - val_loss: 0.0600 - val_acc: 0.9784 - val_precision_m: 0.9823 - val_recall_m: 0.9778\n",
      "Epoch 1676/5000\n",
      " - 10s - loss: 0.0527 - acc: 0.9822 - precision_m: 0.9878 - recall_m: 0.9770 - val_loss: 0.0812 - val_acc: 0.9745 - val_precision_m: 0.9633 - val_recall_m: 0.9912\n",
      "Epoch 1677/5000\n",
      " - 10s - loss: 0.0478 - acc: 0.9858 - precision_m: 0.9900 - recall_m: 0.9816 - val_loss: 0.0551 - val_acc: 0.9828 - val_precision_m: 0.9805 - val_recall_m: 0.9881\n",
      "Epoch 1678/5000\n",
      " - 10s - loss: 0.0460 - acc: 0.9854 - precision_m: 0.9893 - recall_m: 0.9816 - val_loss: 0.0899 - val_acc: 0.9607 - val_precision_m: 0.9957 - val_recall_m: 0.9315\n",
      "Epoch 1679/5000\n",
      " - 10s - loss: 0.0546 - acc: 0.9823 - precision_m: 0.9877 - recall_m: 0.9773 - val_loss: 0.0661 - val_acc: 0.9767 - val_precision_m: 0.9904 - val_recall_m: 0.9669\n",
      "Epoch 1680/5000\n",
      " - 10s - loss: 0.0466 - acc: 0.9867 - precision_m: 0.9912 - recall_m: 0.9822 - val_loss: 0.0652 - val_acc: 0.9828 - val_precision_m: 0.9795 - val_recall_m: 0.9892\n",
      "Epoch 1681/5000\n",
      " - 10s - loss: 0.0489 - acc: 0.9847 - precision_m: 0.9898 - recall_m: 0.9795 - val_loss: 0.0761 - val_acc: 0.9806 - val_precision_m: 0.9756 - val_recall_m: 0.9891\n",
      "Epoch 1682/5000\n",
      " - 10s - loss: 0.0469 - acc: 0.9851 - precision_m: 0.9899 - recall_m: 0.9804 - val_loss: 0.0560 - val_acc: 0.9801 - val_precision_m: 0.9854 - val_recall_m: 0.9781\n",
      "Epoch 1683/5000\n",
      " - 10s - loss: 0.0464 - acc: 0.9860 - precision_m: 0.9917 - recall_m: 0.9802 - val_loss: 0.0524 - val_acc: 0.9812 - val_precision_m: 0.9904 - val_recall_m: 0.9751\n",
      "Epoch 1684/5000\n",
      " - 10s - loss: 0.0454 - acc: 0.9867 - precision_m: 0.9910 - recall_m: 0.9825 - val_loss: 0.0543 - val_acc: 0.9828 - val_precision_m: 0.9865 - val_recall_m: 0.9821\n",
      "Epoch 1685/5000\n",
      " - 10s - loss: 0.0480 - acc: 0.9843 - precision_m: 0.9893 - recall_m: 0.9794 - val_loss: 0.0604 - val_acc: 0.9817 - val_precision_m: 0.9925 - val_recall_m: 0.9742\n",
      "Epoch 1686/5000\n",
      " - 10s - loss: 0.0512 - acc: 0.9827 - precision_m: 0.9888 - recall_m: 0.9769 - val_loss: 0.0648 - val_acc: 0.9784 - val_precision_m: 0.9894 - val_recall_m: 0.9710\n",
      "Epoch 1687/5000\n",
      " - 10s - loss: 0.0444 - acc: 0.9869 - precision_m: 0.9910 - recall_m: 0.9827 - val_loss: 0.0606 - val_acc: 0.9795 - val_precision_m: 0.9914 - val_recall_m: 0.9706\n",
      "Epoch 1688/5000\n",
      " - 10s - loss: 0.0475 - acc: 0.9853 - precision_m: 0.9902 - recall_m: 0.9800 - val_loss: 0.0573 - val_acc: 0.9834 - val_precision_m: 0.9886 - val_recall_m: 0.9810\n",
      "Epoch 1689/5000\n",
      " - 10s - loss: 0.0443 - acc: 0.9865 - precision_m: 0.9912 - recall_m: 0.9816 - val_loss: 0.0650 - val_acc: 0.9834 - val_precision_m: 0.9788 - val_recall_m: 0.9912\n",
      "Epoch 1690/5000\n",
      " - 10s - loss: 0.0463 - acc: 0.9857 - precision_m: 0.9912 - recall_m: 0.9802 - val_loss: 0.0584 - val_acc: 0.9817 - val_precision_m: 0.9875 - val_recall_m: 0.9790\n",
      "Epoch 1691/5000\n",
      " - 10s - loss: 0.0525 - acc: 0.9834 - precision_m: 0.9888 - recall_m: 0.9782 - val_loss: 0.0784 - val_acc: 0.9762 - val_precision_m: 0.9662 - val_recall_m: 0.9912\n",
      "Epoch 1692/5000\n",
      " - 10s - loss: 0.0461 - acc: 0.9850 - precision_m: 0.9903 - recall_m: 0.9797 - val_loss: 0.0596 - val_acc: 0.9801 - val_precision_m: 0.9844 - val_recall_m: 0.9790\n",
      "Epoch 1693/5000\n",
      " - 10s - loss: 0.0436 - acc: 0.9865 - precision_m: 0.9905 - recall_m: 0.9826 - val_loss: 0.0561 - val_acc: 0.9801 - val_precision_m: 0.9905 - val_recall_m: 0.9730\n",
      "Epoch 1694/5000\n",
      " - 10s - loss: 0.0459 - acc: 0.9856 - precision_m: 0.9899 - recall_m: 0.9813 - val_loss: 0.0588 - val_acc: 0.9812 - val_precision_m: 0.9846 - val_recall_m: 0.9812\n",
      "Epoch 1695/5000\n",
      " - 10s - loss: 0.0473 - acc: 0.9844 - precision_m: 0.9896 - recall_m: 0.9794 - val_loss: 0.0549 - val_acc: 0.9806 - val_precision_m: 0.9823 - val_recall_m: 0.9820\n",
      "Epoch 1696/5000\n",
      " - 10s - loss: 0.0435 - acc: 0.9872 - precision_m: 0.9917 - recall_m: 0.9830 - val_loss: 0.0564 - val_acc: 0.9817 - val_precision_m: 0.9905 - val_recall_m: 0.9761\n",
      "Epoch 1697/5000\n",
      " - 10s - loss: 0.0454 - acc: 0.9866 - precision_m: 0.9908 - recall_m: 0.9823 - val_loss: 0.0568 - val_acc: 0.9801 - val_precision_m: 0.9865 - val_recall_m: 0.9770\n",
      "Epoch 1698/5000\n",
      " - 10s - loss: 0.0431 - acc: 0.9858 - precision_m: 0.9900 - recall_m: 0.9816 - val_loss: 0.0586 - val_acc: 0.9828 - val_precision_m: 0.9828 - val_recall_m: 0.9861\n",
      "Epoch 1699/5000\n",
      " - 10s - loss: 0.0515 - acc: 0.9828 - precision_m: 0.9881 - recall_m: 0.9776 - val_loss: 0.0570 - val_acc: 0.9823 - val_precision_m: 0.9846 - val_recall_m: 0.9829\n",
      "Epoch 1700/5000\n",
      " - 10s - loss: 0.0452 - acc: 0.9858 - precision_m: 0.9908 - recall_m: 0.9808 - val_loss: 0.0602 - val_acc: 0.9801 - val_precision_m: 0.9765 - val_recall_m: 0.9872\n",
      "Epoch 1701/5000\n",
      " - 10s - loss: 0.0543 - acc: 0.9825 - precision_m: 0.9889 - recall_m: 0.9771 - val_loss: 0.0712 - val_acc: 0.9812 - val_precision_m: 0.9757 - val_recall_m: 0.9899\n",
      "Epoch 1702/5000\n",
      " - 10s - loss: 0.0531 - acc: 0.9818 - precision_m: 0.9877 - recall_m: 0.9758 - val_loss: 0.0582 - val_acc: 0.9806 - val_precision_m: 0.9885 - val_recall_m: 0.9760\n",
      "Epoch 1703/5000\n",
      " - 10s - loss: 0.0485 - acc: 0.9839 - precision_m: 0.9891 - recall_m: 0.9788 - val_loss: 0.0623 - val_acc: 0.9806 - val_precision_m: 0.9767 - val_recall_m: 0.9880\n",
      "Epoch 1704/5000\n",
      " - 10s - loss: 0.0583 - acc: 0.9804 - precision_m: 0.9868 - recall_m: 0.9744 - val_loss: 0.0623 - val_acc: 0.9801 - val_precision_m: 0.9786 - val_recall_m: 0.9850\n",
      "Epoch 1705/5000\n",
      " - 10s - loss: 0.0464 - acc: 0.9854 - precision_m: 0.9898 - recall_m: 0.9809 - val_loss: 0.0561 - val_acc: 0.9812 - val_precision_m: 0.9884 - val_recall_m: 0.9771\n",
      "Epoch 1706/5000\n",
      " - 10s - loss: 0.0438 - acc: 0.9861 - precision_m: 0.9900 - recall_m: 0.9823 - val_loss: 0.0549 - val_acc: 0.9823 - val_precision_m: 0.9873 - val_recall_m: 0.9800\n",
      "Epoch 1707/5000\n",
      " - 10s - loss: 0.0439 - acc: 0.9863 - precision_m: 0.9904 - recall_m: 0.9825 - val_loss: 0.0509 - val_acc: 0.9828 - val_precision_m: 0.9876 - val_recall_m: 0.9810\n",
      "Epoch 1708/5000\n",
      " - 10s - loss: 0.0432 - acc: 0.9868 - precision_m: 0.9911 - recall_m: 0.9824 - val_loss: 0.0631 - val_acc: 0.9801 - val_precision_m: 0.9817 - val_recall_m: 0.9820\n",
      "Epoch 1709/5000\n",
      " - 10s - loss: 0.0494 - acc: 0.9852 - precision_m: 0.9903 - recall_m: 0.9799 - val_loss: 0.0580 - val_acc: 0.9812 - val_precision_m: 0.9866 - val_recall_m: 0.9789\n",
      "Epoch 1710/5000\n",
      " - 10s - loss: 0.0411 - acc: 0.9884 - precision_m: 0.9923 - recall_m: 0.9843 - val_loss: 0.0589 - val_acc: 0.9795 - val_precision_m: 0.9914 - val_recall_m: 0.9709\n",
      "Epoch 1711/5000\n",
      " - 10s - loss: 0.0443 - acc: 0.9862 - precision_m: 0.9908 - recall_m: 0.9816 - val_loss: 0.0562 - val_acc: 0.9834 - val_precision_m: 0.9815 - val_recall_m: 0.9881\n",
      "Epoch 1712/5000\n",
      " - 10s - loss: 0.0438 - acc: 0.9866 - precision_m: 0.9909 - recall_m: 0.9822 - val_loss: 0.0589 - val_acc: 0.9790 - val_precision_m: 0.9863 - val_recall_m: 0.9751\n",
      "Epoch 1713/5000\n",
      " - 10s - loss: 0.0464 - acc: 0.9843 - precision_m: 0.9890 - recall_m: 0.9793 - val_loss: 0.0762 - val_acc: 0.9773 - val_precision_m: 0.9688 - val_recall_m: 0.9900\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1714/5000\n",
      " - 10s - loss: 0.0462 - acc: 0.9861 - precision_m: 0.9910 - recall_m: 0.9813 - val_loss: 0.0560 - val_acc: 0.9823 - val_precision_m: 0.9914 - val_recall_m: 0.9760\n",
      "Epoch 1715/5000\n",
      " - 10s - loss: 0.0486 - acc: 0.9844 - precision_m: 0.9898 - recall_m: 0.9791 - val_loss: 0.0612 - val_acc: 0.9773 - val_precision_m: 0.9864 - val_recall_m: 0.9719\n",
      "Epoch 1716/5000\n",
      " - 10s - loss: 0.0472 - acc: 0.9846 - precision_m: 0.9903 - recall_m: 0.9790 - val_loss: 0.1063 - val_acc: 0.9535 - val_precision_m: 0.9977 - val_recall_m: 0.9166\n",
      "Epoch 1717/5000\n",
      " - 10s - loss: 0.0467 - acc: 0.9846 - precision_m: 0.9903 - recall_m: 0.9791 - val_loss: 0.0542 - val_acc: 0.9823 - val_precision_m: 0.9894 - val_recall_m: 0.9781\n",
      "Epoch 1718/5000\n",
      " - 10s - loss: 0.0492 - acc: 0.9846 - precision_m: 0.9896 - recall_m: 0.9800 - val_loss: 0.0580 - val_acc: 0.9795 - val_precision_m: 0.9874 - val_recall_m: 0.9750\n",
      "Epoch 1719/5000\n",
      " - 10s - loss: 0.0406 - acc: 0.9875 - precision_m: 0.9916 - recall_m: 0.9832 - val_loss: 0.0589 - val_acc: 0.9795 - val_precision_m: 0.9914 - val_recall_m: 0.9707\n",
      "Epoch 1720/5000\n",
      " - 10s - loss: 0.0457 - acc: 0.9861 - precision_m: 0.9907 - recall_m: 0.9817 - val_loss: 0.0578 - val_acc: 0.9823 - val_precision_m: 0.9936 - val_recall_m: 0.9738\n",
      "Epoch 1721/5000\n",
      " - 10s - loss: 0.0486 - acc: 0.9850 - precision_m: 0.9899 - recall_m: 0.9799 - val_loss: 0.0653 - val_acc: 0.9817 - val_precision_m: 0.9768 - val_recall_m: 0.9901\n",
      "Epoch 1722/5000\n",
      " - 10s - loss: 0.0437 - acc: 0.9875 - precision_m: 0.9918 - recall_m: 0.9833 - val_loss: 0.0594 - val_acc: 0.9790 - val_precision_m: 0.9865 - val_recall_m: 0.9751\n",
      "Epoch 1723/5000\n",
      " - 10s - loss: 0.0453 - acc: 0.9855 - precision_m: 0.9909 - recall_m: 0.9803 - val_loss: 0.0608 - val_acc: 0.9845 - val_precision_m: 0.9817 - val_recall_m: 0.9902\n",
      "Epoch 1724/5000\n",
      " - 10s - loss: 0.0423 - acc: 0.9872 - precision_m: 0.9913 - recall_m: 0.9836 - val_loss: 0.0574 - val_acc: 0.9817 - val_precision_m: 0.9833 - val_recall_m: 0.9833\n",
      "Epoch 1725/5000\n",
      " - 10s - loss: 0.0439 - acc: 0.9865 - precision_m: 0.9897 - recall_m: 0.9826 - val_loss: 0.0710 - val_acc: 0.9762 - val_precision_m: 0.9720 - val_recall_m: 0.9850\n",
      "Epoch 1726/5000\n",
      " - 10s - loss: 0.0511 - acc: 0.9830 - precision_m: 0.9877 - recall_m: 0.9781 - val_loss: 0.0893 - val_acc: 0.9623 - val_precision_m: 0.9967 - val_recall_m: 0.9335\n",
      "Epoch 1727/5000\n",
      " - 10s - loss: 0.0476 - acc: 0.9851 - precision_m: 0.9894 - recall_m: 0.9808 - val_loss: 0.0685 - val_acc: 0.9734 - val_precision_m: 0.9947 - val_recall_m: 0.9563\n",
      "Epoch 1728/5000\n",
      " - 10s - loss: 0.0425 - acc: 0.9866 - precision_m: 0.9905 - recall_m: 0.9829 - val_loss: 0.0600 - val_acc: 0.9801 - val_precision_m: 0.9804 - val_recall_m: 0.9828\n",
      "Epoch 1729/5000\n",
      " - 10s - loss: 0.0405 - acc: 0.9876 - precision_m: 0.9914 - recall_m: 0.9838 - val_loss: 0.1032 - val_acc: 0.9707 - val_precision_m: 0.9541 - val_recall_m: 0.9941\n",
      "Epoch 1730/5000\n",
      " - 10s - loss: 0.0494 - acc: 0.9834 - precision_m: 0.9871 - recall_m: 0.9800 - val_loss: 0.0918 - val_acc: 0.9596 - val_precision_m: 0.9968 - val_recall_m: 0.9288\n",
      "Epoch 1731/5000\n",
      " - 10s - loss: 0.0421 - acc: 0.9878 - precision_m: 0.9921 - recall_m: 0.9834 - val_loss: 0.0552 - val_acc: 0.9823 - val_precision_m: 0.9786 - val_recall_m: 0.9891\n",
      "Epoch 1732/5000\n",
      " - 10s - loss: 0.0408 - acc: 0.9874 - precision_m: 0.9924 - recall_m: 0.9824 - val_loss: 0.0520 - val_acc: 0.9850 - val_precision_m: 0.9817 - val_recall_m: 0.9912\n",
      "Epoch 1733/5000\n",
      " - 10s - loss: 0.0410 - acc: 0.9883 - precision_m: 0.9931 - recall_m: 0.9836 - val_loss: 0.0621 - val_acc: 0.9790 - val_precision_m: 0.9916 - val_recall_m: 0.9698\n",
      "Epoch 1734/5000\n",
      " - 10s - loss: 0.0419 - acc: 0.9873 - precision_m: 0.9901 - recall_m: 0.9843 - val_loss: 0.0720 - val_acc: 0.9695 - val_precision_m: 0.9946 - val_recall_m: 0.9494\n",
      "Epoch 1735/5000\n",
      " - 10s - loss: 0.0532 - acc: 0.9828 - precision_m: 0.9881 - recall_m: 0.9778 - val_loss: 0.0651 - val_acc: 0.9745 - val_precision_m: 0.9893 - val_recall_m: 0.9640\n",
      "Epoch 1736/5000\n",
      " - 10s - loss: 0.0444 - acc: 0.9861 - precision_m: 0.9904 - recall_m: 0.9815 - val_loss: 0.0662 - val_acc: 0.9756 - val_precision_m: 0.9896 - val_recall_m: 0.9657\n",
      "Epoch 1737/5000\n",
      " - 10s - loss: 0.0503 - acc: 0.9823 - precision_m: 0.9874 - recall_m: 0.9775 - val_loss: 0.0706 - val_acc: 0.9712 - val_precision_m: 0.9947 - val_recall_m: 0.9523\n",
      "Epoch 1738/5000\n",
      " - 10s - loss: 0.0485 - acc: 0.9849 - precision_m: 0.9900 - recall_m: 0.9797 - val_loss: 0.0829 - val_acc: 0.9679 - val_precision_m: 0.9947 - val_recall_m: 0.9465\n",
      "Epoch 1739/5000\n",
      " - 10s - loss: 0.0454 - acc: 0.9860 - precision_m: 0.9909 - recall_m: 0.9811 - val_loss: 0.0670 - val_acc: 0.9723 - val_precision_m: 0.9947 - val_recall_m: 0.9544\n",
      "Epoch 1740/5000\n",
      " - 10s - loss: 0.0427 - acc: 0.9870 - precision_m: 0.9909 - recall_m: 0.9833 - val_loss: 0.0576 - val_acc: 0.9801 - val_precision_m: 0.9787 - val_recall_m: 0.9847\n",
      "Epoch 1741/5000\n",
      " - 10s - loss: 0.0416 - acc: 0.9875 - precision_m: 0.9911 - recall_m: 0.9838 - val_loss: 0.0521 - val_acc: 0.9845 - val_precision_m: 0.9887 - val_recall_m: 0.9831\n",
      "Epoch 1742/5000\n",
      " - 10s - loss: 0.0483 - acc: 0.9841 - precision_m: 0.9892 - recall_m: 0.9795 - val_loss: 0.0517 - val_acc: 0.9828 - val_precision_m: 0.9847 - val_recall_m: 0.9842\n",
      "Epoch 1743/5000\n",
      " - 10s - loss: 0.0406 - acc: 0.9874 - precision_m: 0.9911 - recall_m: 0.9835 - val_loss: 0.0588 - val_acc: 0.9806 - val_precision_m: 0.9947 - val_recall_m: 0.9694\n",
      "Epoch 1744/5000\n",
      " - 10s - loss: 0.0402 - acc: 0.9874 - precision_m: 0.9927 - recall_m: 0.9821 - val_loss: 0.0511 - val_acc: 0.9812 - val_precision_m: 0.9814 - val_recall_m: 0.9840\n",
      "Epoch 1745/5000\n",
      " - 10s - loss: 0.0446 - acc: 0.9855 - precision_m: 0.9904 - recall_m: 0.9808 - val_loss: 0.0647 - val_acc: 0.9795 - val_precision_m: 0.9744 - val_recall_m: 0.9882\n",
      "Epoch 1746/5000\n",
      " - 10s - loss: 0.0474 - acc: 0.9855 - precision_m: 0.9901 - recall_m: 0.9811 - val_loss: 0.0559 - val_acc: 0.9823 - val_precision_m: 0.9806 - val_recall_m: 0.9870\n",
      "Epoch 1747/5000\n",
      " - 10s - loss: 0.0404 - acc: 0.9879 - precision_m: 0.9924 - recall_m: 0.9835 - val_loss: 0.0506 - val_acc: 0.9817 - val_precision_m: 0.9865 - val_recall_m: 0.9801\n",
      "Epoch 1748/5000\n",
      " - 10s - loss: 0.0450 - acc: 0.9853 - precision_m: 0.9901 - recall_m: 0.9808 - val_loss: 0.0600 - val_acc: 0.9823 - val_precision_m: 0.9775 - val_recall_m: 0.9903\n",
      "Epoch 1749/5000\n",
      " - 10s - loss: 0.0485 - acc: 0.9846 - precision_m: 0.9905 - recall_m: 0.9792 - val_loss: 0.0529 - val_acc: 0.9795 - val_precision_m: 0.9904 - val_recall_m: 0.9719\n",
      "Epoch 1750/5000\n",
      " - 10s - loss: 0.0469 - acc: 0.9848 - precision_m: 0.9898 - recall_m: 0.9797 - val_loss: 0.0551 - val_acc: 0.9812 - val_precision_m: 0.9844 - val_recall_m: 0.9809\n",
      "Epoch 1751/5000\n",
      " - 10s - loss: 0.0438 - acc: 0.9865 - precision_m: 0.9910 - recall_m: 0.9821 - val_loss: 0.0540 - val_acc: 0.9795 - val_precision_m: 0.9853 - val_recall_m: 0.9768\n",
      "Epoch 1752/5000\n",
      " - 10s - loss: 0.0419 - acc: 0.9874 - precision_m: 0.9919 - recall_m: 0.9828 - val_loss: 0.0544 - val_acc: 0.9806 - val_precision_m: 0.9903 - val_recall_m: 0.9740\n",
      "Epoch 1753/5000\n",
      " - 10s - loss: 0.0468 - acc: 0.9849 - precision_m: 0.9887 - recall_m: 0.9812 - val_loss: 0.0563 - val_acc: 0.9828 - val_precision_m: 0.9885 - val_recall_m: 0.9802\n",
      "Epoch 1754/5000\n",
      " - 10s - loss: 0.0411 - acc: 0.9878 - precision_m: 0.9920 - recall_m: 0.9834 - val_loss: 0.0535 - val_acc: 0.9828 - val_precision_m: 0.9853 - val_recall_m: 0.9830\n",
      "Epoch 1755/5000\n",
      " - 10s - loss: 0.0402 - acc: 0.9875 - precision_m: 0.9924 - recall_m: 0.9827 - val_loss: 0.0505 - val_acc: 0.9839 - val_precision_m: 0.9916 - val_recall_m: 0.9790\n",
      "Epoch 1756/5000\n",
      " - 10s - loss: 0.0418 - acc: 0.9879 - precision_m: 0.9922 - recall_m: 0.9840 - val_loss: 0.0517 - val_acc: 0.9839 - val_precision_m: 0.9856 - val_recall_m: 0.9850\n",
      "Epoch 1757/5000\n",
      " - 10s - loss: 0.0427 - acc: 0.9874 - precision_m: 0.9919 - recall_m: 0.9829 - val_loss: 0.0513 - val_acc: 0.9834 - val_precision_m: 0.9844 - val_recall_m: 0.9851\n",
      "Epoch 1758/5000\n",
      " - 10s - loss: 0.0442 - acc: 0.9865 - precision_m: 0.9907 - recall_m: 0.9823 - val_loss: 0.0499 - val_acc: 0.9856 - val_precision_m: 0.9866 - val_recall_m: 0.9870\n",
      "Epoch 1759/5000\n",
      " - 10s - loss: 0.0413 - acc: 0.9867 - precision_m: 0.9911 - recall_m: 0.9825 - val_loss: 0.0508 - val_acc: 0.9850 - val_precision_m: 0.9904 - val_recall_m: 0.9819\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1760/5000\n",
      " - 10s - loss: 0.0480 - acc: 0.9850 - precision_m: 0.9907 - recall_m: 0.9794 - val_loss: 0.0866 - val_acc: 0.9695 - val_precision_m: 0.9548 - val_recall_m: 0.9909\n",
      "Epoch 1761/5000\n",
      " - 10s - loss: 0.0464 - acc: 0.9850 - precision_m: 0.9896 - recall_m: 0.9806 - val_loss: 0.0501 - val_acc: 0.9862 - val_precision_m: 0.9885 - val_recall_m: 0.9858\n",
      "Epoch 1762/5000\n",
      " - 10s - loss: 0.0410 - acc: 0.9878 - precision_m: 0.9915 - recall_m: 0.9840 - val_loss: 0.0494 - val_acc: 0.9845 - val_precision_m: 0.9885 - val_recall_m: 0.9830\n",
      "Epoch 1763/5000\n",
      " - 10s - loss: 0.0521 - acc: 0.9822 - precision_m: 0.9874 - recall_m: 0.9773 - val_loss: 0.0627 - val_acc: 0.9767 - val_precision_m: 0.9913 - val_recall_m: 0.9656\n",
      "Epoch 1764/5000\n",
      " - 10s - loss: 0.0416 - acc: 0.9872 - precision_m: 0.9921 - recall_m: 0.9822 - val_loss: 0.0581 - val_acc: 0.9790 - val_precision_m: 0.9874 - val_recall_m: 0.9739\n",
      "Epoch 1765/5000\n",
      " - 10s - loss: 0.0425 - acc: 0.9867 - precision_m: 0.9906 - recall_m: 0.9828 - val_loss: 0.0489 - val_acc: 0.9828 - val_precision_m: 0.9894 - val_recall_m: 0.9789\n",
      "Epoch 1766/5000\n",
      " - 10s - loss: 0.0454 - acc: 0.9864 - precision_m: 0.9903 - recall_m: 0.9826 - val_loss: 0.0525 - val_acc: 0.9806 - val_precision_m: 0.9786 - val_recall_m: 0.9863\n",
      "Epoch 1767/5000\n",
      " - 10s - loss: 0.0385 - acc: 0.9878 - precision_m: 0.9920 - recall_m: 0.9836 - val_loss: 0.0501 - val_acc: 0.9867 - val_precision_m: 0.9857 - val_recall_m: 0.9901\n",
      "Epoch 1768/5000\n",
      " - 10s - loss: 0.0426 - acc: 0.9868 - precision_m: 0.9912 - recall_m: 0.9824 - val_loss: 0.0675 - val_acc: 0.9762 - val_precision_m: 0.9670 - val_recall_m: 0.9901\n",
      "Epoch 1769/5000\n",
      " - 10s - loss: 0.0447 - acc: 0.9861 - precision_m: 0.9898 - recall_m: 0.9824 - val_loss: 0.0599 - val_acc: 0.9801 - val_precision_m: 0.9915 - val_recall_m: 0.9717\n",
      "Epoch 1770/5000\n",
      " - 10s - loss: 0.0416 - acc: 0.9875 - precision_m: 0.9917 - recall_m: 0.9833 - val_loss: 0.0589 - val_acc: 0.9790 - val_precision_m: 0.9915 - val_recall_m: 0.9697\n",
      "Epoch 1771/5000\n",
      " - 10s - loss: 0.0449 - acc: 0.9849 - precision_m: 0.9890 - recall_m: 0.9811 - val_loss: 0.0510 - val_acc: 0.9850 - val_precision_m: 0.9925 - val_recall_m: 0.9804\n",
      "Epoch 1772/5000\n",
      " - 10s - loss: 0.0531 - acc: 0.9818 - precision_m: 0.9875 - recall_m: 0.9767 - val_loss: 0.0557 - val_acc: 0.9845 - val_precision_m: 0.9836 - val_recall_m: 0.9881\n",
      "Epoch 1773/5000\n",
      " - 10s - loss: 0.0442 - acc: 0.9865 - precision_m: 0.9913 - recall_m: 0.9817 - val_loss: 0.0482 - val_acc: 0.9867 - val_precision_m: 0.9905 - val_recall_m: 0.9850\n",
      "Epoch 1774/5000\n",
      " - 10s - loss: 0.0396 - acc: 0.9875 - precision_m: 0.9916 - recall_m: 0.9832 - val_loss: 0.0540 - val_acc: 0.9801 - val_precision_m: 0.9905 - val_recall_m: 0.9730\n",
      "Epoch 1775/5000\n",
      " - 10s - loss: 0.0399 - acc: 0.9881 - precision_m: 0.9919 - recall_m: 0.9844 - val_loss: 0.0623 - val_acc: 0.9795 - val_precision_m: 0.9691 - val_recall_m: 0.9941\n",
      "Epoch 1776/5000\n",
      " - 10s - loss: 0.0394 - acc: 0.9886 - precision_m: 0.9930 - recall_m: 0.9839 - val_loss: 0.0496 - val_acc: 0.9850 - val_precision_m: 0.9845 - val_recall_m: 0.9880\n",
      "Epoch 1777/5000\n",
      " - 10s - loss: 0.0400 - acc: 0.9875 - precision_m: 0.9917 - recall_m: 0.9832 - val_loss: 0.0627 - val_acc: 0.9817 - val_precision_m: 0.9757 - val_recall_m: 0.9910\n",
      "Epoch 1778/5000\n",
      " - 10s - loss: 0.0450 - acc: 0.9846 - precision_m: 0.9891 - recall_m: 0.9799 - val_loss: 0.0520 - val_acc: 0.9834 - val_precision_m: 0.9925 - val_recall_m: 0.9770\n",
      "Epoch 1779/5000\n",
      " - 10s - loss: 0.0444 - acc: 0.9867 - precision_m: 0.9910 - recall_m: 0.9822 - val_loss: 0.0562 - val_acc: 0.9795 - val_precision_m: 0.9835 - val_recall_m: 0.9790\n",
      "Epoch 1780/5000\n",
      " - 10s - loss: 0.0403 - acc: 0.9883 - precision_m: 0.9931 - recall_m: 0.9835 - val_loss: 0.0561 - val_acc: 0.9839 - val_precision_m: 0.9847 - val_recall_m: 0.9859\n",
      "Epoch 1781/5000\n",
      " - 10s - loss: 0.0425 - acc: 0.9864 - precision_m: 0.9905 - recall_m: 0.9824 - val_loss: 0.0469 - val_acc: 0.9834 - val_precision_m: 0.9896 - val_recall_m: 0.9799\n",
      "Epoch 1782/5000\n",
      " - 10s - loss: 0.0415 - acc: 0.9872 - precision_m: 0.9904 - recall_m: 0.9840 - val_loss: 0.0503 - val_acc: 0.9834 - val_precision_m: 0.9885 - val_recall_m: 0.9810\n",
      "Epoch 1783/5000\n",
      " - 10s - loss: 0.0412 - acc: 0.9876 - precision_m: 0.9909 - recall_m: 0.9842 - val_loss: 0.0538 - val_acc: 0.9823 - val_precision_m: 0.9867 - val_recall_m: 0.9809\n",
      "Epoch 1784/5000\n",
      " - 10s - loss: 0.0414 - acc: 0.9868 - precision_m: 0.9914 - recall_m: 0.9822 - val_loss: 0.0534 - val_acc: 0.9823 - val_precision_m: 0.9797 - val_recall_m: 0.9881\n",
      "Epoch 1785/5000\n",
      " - 10s - loss: 0.0395 - acc: 0.9878 - precision_m: 0.9921 - recall_m: 0.9837 - val_loss: 0.0533 - val_acc: 0.9823 - val_precision_m: 0.9905 - val_recall_m: 0.9769\n",
      "Epoch 1786/5000\n",
      " - 10s - loss: 0.0420 - acc: 0.9877 - precision_m: 0.9923 - recall_m: 0.9831 - val_loss: 0.0477 - val_acc: 0.9817 - val_precision_m: 0.9936 - val_recall_m: 0.9732\n",
      "Epoch 1787/5000\n",
      " - 10s - loss: 0.0391 - acc: 0.9887 - precision_m: 0.9925 - recall_m: 0.9852 - val_loss: 0.0636 - val_acc: 0.9773 - val_precision_m: 0.9689 - val_recall_m: 0.9901\n",
      "Epoch 1788/5000\n",
      " - 10s - loss: 0.0508 - acc: 0.9830 - precision_m: 0.9878 - recall_m: 0.9788 - val_loss: 0.0511 - val_acc: 0.9828 - val_precision_m: 0.9826 - val_recall_m: 0.9860\n",
      "Epoch 1789/5000\n",
      " - 10s - loss: 0.0435 - acc: 0.9873 - precision_m: 0.9918 - recall_m: 0.9830 - val_loss: 0.0606 - val_acc: 0.9779 - val_precision_m: 0.9738 - val_recall_m: 0.9862\n",
      "Epoch 1790/5000\n",
      " - 10s - loss: 0.0393 - acc: 0.9879 - precision_m: 0.9920 - recall_m: 0.9837 - val_loss: 0.0528 - val_acc: 0.9839 - val_precision_m: 0.9827 - val_recall_m: 0.9879\n",
      "Epoch 1791/5000\n",
      " - 10s - loss: 0.0395 - acc: 0.9887 - precision_m: 0.9919 - recall_m: 0.9855 - val_loss: 0.0514 - val_acc: 0.9812 - val_precision_m: 0.9874 - val_recall_m: 0.9779\n",
      "Epoch 1792/5000\n",
      " - 10s - loss: 0.0420 - acc: 0.9865 - precision_m: 0.9904 - recall_m: 0.9827 - val_loss: 0.0492 - val_acc: 0.9812 - val_precision_m: 0.9875 - val_recall_m: 0.9781\n",
      "Epoch 1793/5000\n",
      " - 10s - loss: 0.0443 - acc: 0.9857 - precision_m: 0.9901 - recall_m: 0.9813 - val_loss: 0.0614 - val_acc: 0.9806 - val_precision_m: 0.9914 - val_recall_m: 0.9729\n",
      "Epoch 1794/5000\n",
      " - 10s - loss: 0.0438 - acc: 0.9861 - precision_m: 0.9908 - recall_m: 0.9812 - val_loss: 0.0641 - val_acc: 0.9740 - val_precision_m: 0.9958 - val_recall_m: 0.9565\n",
      "Epoch 1795/5000\n",
      " - 10s - loss: 0.0405 - acc: 0.9884 - precision_m: 0.9924 - recall_m: 0.9843 - val_loss: 0.0500 - val_acc: 0.9812 - val_precision_m: 0.9894 - val_recall_m: 0.9763\n",
      "Epoch 1796/5000\n",
      " - 10s - loss: 0.0419 - acc: 0.9869 - precision_m: 0.9910 - recall_m: 0.9830 - val_loss: 0.0512 - val_acc: 0.9801 - val_precision_m: 0.9886 - val_recall_m: 0.9750\n",
      "Epoch 1797/5000\n",
      " - 10s - loss: 0.0409 - acc: 0.9871 - precision_m: 0.9907 - recall_m: 0.9835 - val_loss: 0.0548 - val_acc: 0.9790 - val_precision_m: 0.9894 - val_recall_m: 0.9718\n",
      "Epoch 1798/5000\n",
      " - 10s - loss: 0.0383 - acc: 0.9890 - precision_m: 0.9916 - recall_m: 0.9864 - val_loss: 0.0516 - val_acc: 0.9806 - val_precision_m: 0.9915 - val_recall_m: 0.9730\n",
      "Epoch 1799/5000\n",
      " - 10s - loss: 0.0417 - acc: 0.9870 - precision_m: 0.9904 - recall_m: 0.9835 - val_loss: 0.0520 - val_acc: 0.9828 - val_precision_m: 0.9915 - val_recall_m: 0.9773\n",
      "Epoch 1800/5000\n",
      " - 10s - loss: 0.0412 - acc: 0.9879 - precision_m: 0.9918 - recall_m: 0.9838 - val_loss: 0.0505 - val_acc: 0.9834 - val_precision_m: 0.9855 - val_recall_m: 0.9841\n",
      "Epoch 1801/5000\n",
      " - 10s - loss: 0.0441 - acc: 0.9849 - precision_m: 0.9896 - recall_m: 0.9805 - val_loss: 0.0582 - val_acc: 0.9812 - val_precision_m: 0.9736 - val_recall_m: 0.9920\n",
      "Epoch 1802/5000\n",
      " - 10s - loss: 0.0379 - acc: 0.9893 - precision_m: 0.9932 - recall_m: 0.9854 - val_loss: 0.0455 - val_acc: 0.9862 - val_precision_m: 0.9865 - val_recall_m: 0.9882\n",
      "Epoch 1803/5000\n",
      " - 10s - loss: 0.0387 - acc: 0.9886 - precision_m: 0.9923 - recall_m: 0.9853 - val_loss: 0.0503 - val_acc: 0.9845 - val_precision_m: 0.9926 - val_recall_m: 0.9792\n",
      "Epoch 1804/5000\n",
      " - 10s - loss: 0.0432 - acc: 0.9875 - precision_m: 0.9912 - recall_m: 0.9839 - val_loss: 0.0716 - val_acc: 0.9762 - val_precision_m: 0.9654 - val_recall_m: 0.9920\n",
      "Epoch 1805/5000\n",
      " - 10s - loss: 0.0455 - acc: 0.9850 - precision_m: 0.9897 - recall_m: 0.9806 - val_loss: 0.0584 - val_acc: 0.9806 - val_precision_m: 0.9758 - val_recall_m: 0.9892\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1806/5000\n",
      " - 10s - loss: 0.0370 - acc: 0.9900 - precision_m: 0.9926 - recall_m: 0.9874 - val_loss: 0.0580 - val_acc: 0.9784 - val_precision_m: 0.9718 - val_recall_m: 0.9891\n",
      "Epoch 1807/5000\n",
      " - 10s - loss: 0.0381 - acc: 0.9890 - precision_m: 0.9925 - recall_m: 0.9855 - val_loss: 0.0467 - val_acc: 0.9839 - val_precision_m: 0.9846 - val_recall_m: 0.9858\n",
      "Epoch 1808/5000\n",
      " - 10s - loss: 0.0379 - acc: 0.9889 - precision_m: 0.9927 - recall_m: 0.9848 - val_loss: 0.0458 - val_acc: 0.9850 - val_precision_m: 0.9874 - val_recall_m: 0.9851\n",
      "Epoch 1809/5000\n",
      " - 10s - loss: 0.0434 - acc: 0.9864 - precision_m: 0.9895 - recall_m: 0.9835 - val_loss: 0.0592 - val_acc: 0.9790 - val_precision_m: 0.9947 - val_recall_m: 0.9667\n",
      "Epoch 1810/5000\n",
      " - 10s - loss: 0.0428 - acc: 0.9864 - precision_m: 0.9904 - recall_m: 0.9827 - val_loss: 0.0533 - val_acc: 0.9834 - val_precision_m: 0.9816 - val_recall_m: 0.9879\n",
      "Epoch 1811/5000\n",
      " - 10s - loss: 0.0418 - acc: 0.9870 - precision_m: 0.9904 - recall_m: 0.9836 - val_loss: 0.0505 - val_acc: 0.9823 - val_precision_m: 0.9914 - val_recall_m: 0.9760\n",
      "Epoch 1812/5000\n",
      " - 10s - loss: 0.0361 - acc: 0.9888 - precision_m: 0.9936 - recall_m: 0.9838 - val_loss: 0.0491 - val_acc: 0.9839 - val_precision_m: 0.9809 - val_recall_m: 0.9900\n",
      "Epoch 1813/5000\n",
      " - 10s - loss: 0.0388 - acc: 0.9880 - precision_m: 0.9913 - recall_m: 0.9845 - val_loss: 0.0751 - val_acc: 0.9707 - val_precision_m: 0.9957 - val_recall_m: 0.9503\n",
      "Epoch 1814/5000\n",
      " - 10s - loss: 0.0403 - acc: 0.9876 - precision_m: 0.9917 - recall_m: 0.9837 - val_loss: 0.0543 - val_acc: 0.9839 - val_precision_m: 0.9817 - val_recall_m: 0.9889\n",
      "Epoch 1815/5000\n",
      " - 10s - loss: 0.0466 - acc: 0.9846 - precision_m: 0.9890 - recall_m: 0.9801 - val_loss: 0.0595 - val_acc: 0.9762 - val_precision_m: 0.9947 - val_recall_m: 0.9615\n",
      "Epoch 1816/5000\n",
      " - 10s - loss: 0.0423 - acc: 0.9868 - precision_m: 0.9913 - recall_m: 0.9825 - val_loss: 0.0483 - val_acc: 0.9839 - val_precision_m: 0.9865 - val_recall_m: 0.9841\n",
      "Epoch 1817/5000\n",
      " - 10s - loss: 0.0420 - acc: 0.9863 - precision_m: 0.9902 - recall_m: 0.9824 - val_loss: 0.0568 - val_acc: 0.9801 - val_precision_m: 0.9958 - val_recall_m: 0.9677\n",
      "Epoch 1818/5000\n",
      " - 10s - loss: 0.0375 - acc: 0.9884 - precision_m: 0.9921 - recall_m: 0.9850 - val_loss: 0.0457 - val_acc: 0.9856 - val_precision_m: 0.9866 - val_recall_m: 0.9870\n",
      "Epoch 1819/5000\n",
      " - 10s - loss: 0.0375 - acc: 0.9885 - precision_m: 0.9920 - recall_m: 0.9849 - val_loss: 0.0465 - val_acc: 0.9856 - val_precision_m: 0.9856 - val_recall_m: 0.9881\n",
      "Epoch 1820/5000\n",
      " - 10s - loss: 0.0470 - acc: 0.9850 - precision_m: 0.9895 - recall_m: 0.9813 - val_loss: 0.0703 - val_acc: 0.9790 - val_precision_m: 0.9690 - val_recall_m: 0.9931\n",
      "Epoch 1821/5000\n",
      " - 10s - loss: 0.0399 - acc: 0.9876 - precision_m: 0.9912 - recall_m: 0.9840 - val_loss: 0.0671 - val_acc: 0.9712 - val_precision_m: 0.9947 - val_recall_m: 0.9524\n",
      "Epoch 1822/5000\n",
      " - 10s - loss: 0.0466 - acc: 0.9842 - precision_m: 0.9895 - recall_m: 0.9788 - val_loss: 0.0462 - val_acc: 0.9862 - val_precision_m: 0.9877 - val_recall_m: 0.9870\n",
      "Epoch 1823/5000\n",
      " - 10s - loss: 0.0423 - acc: 0.9861 - precision_m: 0.9901 - recall_m: 0.9824 - val_loss: 0.1037 - val_acc: 0.9502 - val_precision_m: 0.9967 - val_recall_m: 0.9110\n",
      "Epoch 1824/5000\n",
      " - 10s - loss: 0.0419 - acc: 0.9870 - precision_m: 0.9919 - recall_m: 0.9819 - val_loss: 0.0474 - val_acc: 0.9862 - val_precision_m: 0.9867 - val_recall_m: 0.9881\n",
      "Epoch 1825/5000\n",
      " - 10s - loss: 0.0342 - acc: 0.9899 - precision_m: 0.9935 - recall_m: 0.9862 - val_loss: 0.0475 - val_acc: 0.9850 - val_precision_m: 0.9826 - val_recall_m: 0.9900\n",
      "Epoch 1826/5000\n",
      " - 10s - loss: 0.0443 - acc: 0.9861 - precision_m: 0.9897 - recall_m: 0.9830 - val_loss: 0.0517 - val_acc: 0.9812 - val_precision_m: 0.9767 - val_recall_m: 0.9890\n",
      "Epoch 1827/5000\n",
      " - 10s - loss: 0.0380 - acc: 0.9877 - precision_m: 0.9921 - recall_m: 0.9834 - val_loss: 0.0427 - val_acc: 0.9862 - val_precision_m: 0.9906 - val_recall_m: 0.9842\n",
      "Epoch 1828/5000\n",
      " - 10s - loss: 0.0367 - acc: 0.9893 - precision_m: 0.9936 - recall_m: 0.9851 - val_loss: 0.0470 - val_acc: 0.9850 - val_precision_m: 0.9838 - val_recall_m: 0.9892\n",
      "Epoch 1829/5000\n",
      " - 10s - loss: 0.0380 - acc: 0.9878 - precision_m: 0.9913 - recall_m: 0.9841 - val_loss: 0.0594 - val_acc: 0.9795 - val_precision_m: 0.9958 - val_recall_m: 0.9672\n",
      "Epoch 1830/5000\n",
      " - 10s - loss: 0.0397 - acc: 0.9878 - precision_m: 0.9911 - recall_m: 0.9845 - val_loss: 0.0489 - val_acc: 0.9850 - val_precision_m: 0.9928 - val_recall_m: 0.9802\n",
      "Epoch 1831/5000\n",
      " - 10s - loss: 0.0398 - acc: 0.9869 - precision_m: 0.9915 - recall_m: 0.9820 - val_loss: 0.0473 - val_acc: 0.9862 - val_precision_m: 0.9847 - val_recall_m: 0.9899\n",
      "Epoch 1832/5000\n",
      " - 10s - loss: 0.0334 - acc: 0.9912 - precision_m: 0.9943 - recall_m: 0.9881 - val_loss: 0.0463 - val_acc: 0.9856 - val_precision_m: 0.9837 - val_recall_m: 0.9900\n",
      "Epoch 1833/5000\n",
      " - 10s - loss: 0.0409 - acc: 0.9871 - precision_m: 0.9909 - recall_m: 0.9835 - val_loss: 0.0462 - val_acc: 0.9839 - val_precision_m: 0.9866 - val_recall_m: 0.9840\n",
      "Epoch 1834/5000\n",
      " - 10s - loss: 0.0414 - acc: 0.9871 - precision_m: 0.9916 - recall_m: 0.9825 - val_loss: 0.0604 - val_acc: 0.9767 - val_precision_m: 0.9927 - val_recall_m: 0.9651\n",
      "Epoch 1835/5000\n",
      " - 10s - loss: 0.0363 - acc: 0.9890 - precision_m: 0.9924 - recall_m: 0.9857 - val_loss: 0.0456 - val_acc: 0.9839 - val_precision_m: 0.9875 - val_recall_m: 0.9832\n",
      "Epoch 1836/5000\n",
      " - 10s - loss: 0.0394 - acc: 0.9868 - precision_m: 0.9905 - recall_m: 0.9831 - val_loss: 0.0526 - val_acc: 0.9839 - val_precision_m: 0.9796 - val_recall_m: 0.9910\n",
      "Epoch 1837/5000\n",
      " - 10s - loss: 0.0386 - acc: 0.9874 - precision_m: 0.9910 - recall_m: 0.9840 - val_loss: 0.0495 - val_acc: 0.9828 - val_precision_m: 0.9875 - val_recall_m: 0.9812\n",
      "Epoch 1838/5000\n",
      " - 10s - loss: 0.0368 - acc: 0.9892 - precision_m: 0.9925 - recall_m: 0.9859 - val_loss: 0.0510 - val_acc: 0.9834 - val_precision_m: 0.9787 - val_recall_m: 0.9912\n",
      "Epoch 1839/5000\n",
      " - 10s - loss: 0.0399 - acc: 0.9875 - precision_m: 0.9910 - recall_m: 0.9839 - val_loss: 0.0517 - val_acc: 0.9806 - val_precision_m: 0.9915 - val_recall_m: 0.9727\n",
      "Epoch 1840/5000\n",
      " - 10s - loss: 0.0334 - acc: 0.9901 - precision_m: 0.9941 - recall_m: 0.9862 - val_loss: 0.0544 - val_acc: 0.9801 - val_precision_m: 0.9872 - val_recall_m: 0.9761\n",
      "Epoch 1841/5000\n",
      " - 10s - loss: 0.0400 - acc: 0.9878 - precision_m: 0.9926 - recall_m: 0.9830 - val_loss: 0.0580 - val_acc: 0.9856 - val_precision_m: 0.9798 - val_recall_m: 0.9940\n",
      "Epoch 1842/5000\n",
      " - 10s - loss: 0.0405 - acc: 0.9876 - precision_m: 0.9915 - recall_m: 0.9838 - val_loss: 0.0533 - val_acc: 0.9817 - val_precision_m: 0.9807 - val_recall_m: 0.9859\n",
      "Epoch 1843/5000\n",
      " - 10s - loss: 0.0549 - acc: 0.9796 - precision_m: 0.9852 - recall_m: 0.9743 - val_loss: 0.0471 - val_acc: 0.9873 - val_precision_m: 0.9847 - val_recall_m: 0.9921\n",
      "Epoch 1844/5000\n",
      " - 10s - loss: 0.0351 - acc: 0.9905 - precision_m: 0.9944 - recall_m: 0.9865 - val_loss: 0.0467 - val_acc: 0.9834 - val_precision_m: 0.9825 - val_recall_m: 0.9870\n",
      "Epoch 1845/5000\n",
      " - 10s - loss: 0.0343 - acc: 0.9901 - precision_m: 0.9937 - recall_m: 0.9865 - val_loss: 0.0431 - val_acc: 0.9867 - val_precision_m: 0.9906 - val_recall_m: 0.9851\n",
      "Epoch 1846/5000\n",
      " - 10s - loss: 0.0331 - acc: 0.9903 - precision_m: 0.9934 - recall_m: 0.9871 - val_loss: 0.0487 - val_acc: 0.9828 - val_precision_m: 0.9925 - val_recall_m: 0.9758\n",
      "Epoch 1847/5000\n",
      " - 10s - loss: 0.0339 - acc: 0.9907 - precision_m: 0.9941 - recall_m: 0.9873 - val_loss: 0.0419 - val_acc: 0.9856 - val_precision_m: 0.9865 - val_recall_m: 0.9870\n",
      "Epoch 1848/5000\n",
      " - 10s - loss: 0.0360 - acc: 0.9894 - precision_m: 0.9927 - recall_m: 0.9860 - val_loss: 0.0469 - val_acc: 0.9867 - val_precision_m: 0.9858 - val_recall_m: 0.9900\n",
      "Epoch 1849/5000\n",
      " - 10s - loss: 0.0388 - acc: 0.9875 - precision_m: 0.9911 - recall_m: 0.9840 - val_loss: 0.0471 - val_acc: 0.9862 - val_precision_m: 0.9915 - val_recall_m: 0.9831\n",
      "Epoch 1850/5000\n",
      " - 10s - loss: 0.0456 - acc: 0.9857 - precision_m: 0.9901 - recall_m: 0.9819 - val_loss: 0.0474 - val_acc: 0.9839 - val_precision_m: 0.9844 - val_recall_m: 0.9860\n",
      "Epoch 1851/5000\n",
      " - 10s - loss: 0.0351 - acc: 0.9895 - precision_m: 0.9936 - recall_m: 0.9856 - val_loss: 0.0614 - val_acc: 0.9745 - val_precision_m: 0.9969 - val_recall_m: 0.9569\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1852/5000\n",
      " - 10s - loss: 0.0381 - acc: 0.9881 - precision_m: 0.9919 - recall_m: 0.9845 - val_loss: 0.0838 - val_acc: 0.9635 - val_precision_m: 0.9946 - val_recall_m: 0.9378\n",
      "Epoch 1853/5000\n",
      " - 10s - loss: 0.0372 - acc: 0.9895 - precision_m: 0.9928 - recall_m: 0.9862 - val_loss: 0.0498 - val_acc: 0.9834 - val_precision_m: 0.9915 - val_recall_m: 0.9781\n",
      "Epoch 1854/5000\n",
      " - 10s - loss: 0.0478 - acc: 0.9847 - precision_m: 0.9881 - recall_m: 0.9816 - val_loss: 0.0518 - val_acc: 0.9839 - val_precision_m: 0.9798 - val_recall_m: 0.9911\n",
      "Epoch 1855/5000\n",
      " - 10s - loss: 0.0374 - acc: 0.9887 - precision_m: 0.9927 - recall_m: 0.9848 - val_loss: 0.0448 - val_acc: 0.9862 - val_precision_m: 0.9878 - val_recall_m: 0.9870\n",
      "Epoch 1856/5000\n",
      " - 10s - loss: 0.0434 - acc: 0.9863 - precision_m: 0.9908 - recall_m: 0.9820 - val_loss: 0.0526 - val_acc: 0.9812 - val_precision_m: 0.9947 - val_recall_m: 0.9708\n",
      "Epoch 1857/5000\n",
      " - 10s - loss: 0.0369 - acc: 0.9891 - precision_m: 0.9920 - recall_m: 0.9861 - val_loss: 0.0456 - val_acc: 0.9839 - val_precision_m: 0.9886 - val_recall_m: 0.9822\n",
      "Epoch 1858/5000\n",
      " - 10s - loss: 0.0350 - acc: 0.9903 - precision_m: 0.9933 - recall_m: 0.9874 - val_loss: 0.0504 - val_acc: 0.9823 - val_precision_m: 0.9776 - val_recall_m: 0.9901\n",
      "Epoch 1859/5000\n",
      " - 10s - loss: 0.0349 - acc: 0.9892 - precision_m: 0.9927 - recall_m: 0.9857 - val_loss: 0.0494 - val_acc: 0.9817 - val_precision_m: 0.9926 - val_recall_m: 0.9739\n",
      "Epoch 1860/5000\n",
      " - 10s - loss: 0.0370 - acc: 0.9892 - precision_m: 0.9921 - recall_m: 0.9863 - val_loss: 0.0519 - val_acc: 0.9817 - val_precision_m: 0.9825 - val_recall_m: 0.9840\n",
      "Epoch 1861/5000\n",
      " - 10s - loss: 0.0373 - acc: 0.9887 - precision_m: 0.9924 - recall_m: 0.9852 - val_loss: 0.0441 - val_acc: 0.9856 - val_precision_m: 0.9929 - val_recall_m: 0.9810\n",
      "Epoch 1862/5000\n",
      " - 10s - loss: 0.0369 - acc: 0.9891 - precision_m: 0.9921 - recall_m: 0.9862 - val_loss: 0.0530 - val_acc: 0.9823 - val_precision_m: 0.9958 - val_recall_m: 0.9718\n",
      "Epoch 1863/5000\n",
      " - 10s - loss: 0.0377 - acc: 0.9887 - precision_m: 0.9924 - recall_m: 0.9853 - val_loss: 0.0518 - val_acc: 0.9823 - val_precision_m: 0.9937 - val_recall_m: 0.9738\n",
      "Epoch 1864/5000\n",
      " - 10s - loss: 0.0382 - acc: 0.9887 - precision_m: 0.9923 - recall_m: 0.9854 - val_loss: 0.0520 - val_acc: 0.9839 - val_precision_m: 0.9969 - val_recall_m: 0.9735\n",
      "Epoch 1865/5000\n",
      " - 10s - loss: 0.0351 - acc: 0.9896 - precision_m: 0.9924 - recall_m: 0.9868 - val_loss: 0.0432 - val_acc: 0.9850 - val_precision_m: 0.9895 - val_recall_m: 0.9828\n",
      "Epoch 1866/5000\n",
      " - 10s - loss: 0.0394 - acc: 0.9868 - precision_m: 0.9915 - recall_m: 0.9821 - val_loss: 0.0465 - val_acc: 0.9839 - val_precision_m: 0.9866 - val_recall_m: 0.9841\n",
      "Epoch 1867/5000\n",
      " - 10s - loss: 0.0376 - acc: 0.9884 - precision_m: 0.9918 - recall_m: 0.9851 - val_loss: 0.0406 - val_acc: 0.9867 - val_precision_m: 0.9926 - val_recall_m: 0.9831\n",
      "Epoch 1868/5000\n",
      " - 10s - loss: 0.0343 - acc: 0.9906 - precision_m: 0.9935 - recall_m: 0.9877 - val_loss: 0.0453 - val_acc: 0.9834 - val_precision_m: 0.9813 - val_recall_m: 0.9879\n",
      "Epoch 1869/5000\n",
      " - 10s - loss: 0.0414 - acc: 0.9868 - precision_m: 0.9905 - recall_m: 0.9831 - val_loss: 0.0720 - val_acc: 0.9707 - val_precision_m: 0.9968 - val_recall_m: 0.9496\n",
      "Epoch 1870/5000\n",
      " - 10s - loss: 0.0389 - acc: 0.9880 - precision_m: 0.9917 - recall_m: 0.9843 - val_loss: 0.0734 - val_acc: 0.9729 - val_precision_m: 0.9968 - val_recall_m: 0.9532\n",
      "Epoch 1871/5000\n",
      " - 10s - loss: 0.0451 - acc: 0.9849 - precision_m: 0.9894 - recall_m: 0.9808 - val_loss: 0.0485 - val_acc: 0.9834 - val_precision_m: 0.9806 - val_recall_m: 0.9890\n",
      "Epoch 1872/5000\n",
      " - 10s - loss: 0.0386 - acc: 0.9881 - precision_m: 0.9921 - recall_m: 0.9838 - val_loss: 0.0638 - val_acc: 0.9740 - val_precision_m: 0.9947 - val_recall_m: 0.9572\n",
      "Epoch 1873/5000\n",
      " - 10s - loss: 0.0435 - acc: 0.9866 - precision_m: 0.9898 - recall_m: 0.9835 - val_loss: 0.0450 - val_acc: 0.9862 - val_precision_m: 0.9895 - val_recall_m: 0.9850\n",
      "Epoch 1874/5000\n",
      " - 10s - loss: 0.0339 - acc: 0.9897 - precision_m: 0.9930 - recall_m: 0.9864 - val_loss: 0.0482 - val_acc: 0.9862 - val_precision_m: 0.9948 - val_recall_m: 0.9800\n",
      "Epoch 1875/5000\n",
      " - 10s - loss: 0.0403 - acc: 0.9868 - precision_m: 0.9898 - recall_m: 0.9843 - val_loss: 0.0493 - val_acc: 0.9856 - val_precision_m: 0.9948 - val_recall_m: 0.9789\n",
      "Epoch 1876/5000\n",
      " - 10s - loss: 0.0398 - acc: 0.9884 - precision_m: 0.9921 - recall_m: 0.9845 - val_loss: 0.0504 - val_acc: 0.9845 - val_precision_m: 0.9947 - val_recall_m: 0.9771\n",
      "Epoch 1877/5000\n",
      " - 10s - loss: 0.0326 - acc: 0.9901 - precision_m: 0.9929 - recall_m: 0.9873 - val_loss: 0.0401 - val_acc: 0.9856 - val_precision_m: 0.9886 - val_recall_m: 0.9850\n",
      "Epoch 1878/5000\n",
      " - 10s - loss: 0.0390 - acc: 0.9871 - precision_m: 0.9907 - recall_m: 0.9838 - val_loss: 0.0633 - val_acc: 0.9773 - val_precision_m: 0.9651 - val_recall_m: 0.9940\n",
      "Epoch 1879/5000\n",
      " - 10s - loss: 0.0494 - acc: 0.9826 - precision_m: 0.9858 - recall_m: 0.9791 - val_loss: 0.0429 - val_acc: 0.9878 - val_precision_m: 0.9916 - val_recall_m: 0.9861\n",
      "Epoch 1880/5000\n",
      " - 10s - loss: 0.0338 - acc: 0.9903 - precision_m: 0.9932 - recall_m: 0.9872 - val_loss: 0.0457 - val_acc: 0.9856 - val_precision_m: 0.9846 - val_recall_m: 0.9891\n",
      "Epoch 1881/5000\n",
      " - 10s - loss: 0.0329 - acc: 0.9908 - precision_m: 0.9944 - recall_m: 0.9871 - val_loss: 0.0593 - val_acc: 0.9762 - val_precision_m: 0.9927 - val_recall_m: 0.9634\n",
      "Epoch 1882/5000\n",
      " - 10s - loss: 0.0358 - acc: 0.9898 - precision_m: 0.9925 - recall_m: 0.9870 - val_loss: 0.0390 - val_acc: 0.9878 - val_precision_m: 0.9845 - val_recall_m: 0.9931\n",
      "Epoch 1883/5000\n",
      " - 10s - loss: 0.0333 - acc: 0.9900 - precision_m: 0.9932 - recall_m: 0.9868 - val_loss: 0.0471 - val_acc: 0.9839 - val_precision_m: 0.9925 - val_recall_m: 0.9782\n",
      "Epoch 1884/5000\n",
      " - 10s - loss: 0.0347 - acc: 0.9892 - precision_m: 0.9930 - recall_m: 0.9855 - val_loss: 0.0665 - val_acc: 0.9707 - val_precision_m: 0.9968 - val_recall_m: 0.9492\n",
      "Epoch 1885/5000\n",
      " - 10s - loss: 0.0345 - acc: 0.9896 - precision_m: 0.9931 - recall_m: 0.9864 - val_loss: 0.0511 - val_acc: 0.9823 - val_precision_m: 0.9786 - val_recall_m: 0.9890\n",
      "Epoch 1886/5000\n",
      " - 10s - loss: 0.0359 - acc: 0.9891 - precision_m: 0.9919 - recall_m: 0.9865 - val_loss: 0.0532 - val_acc: 0.9795 - val_precision_m: 0.9926 - val_recall_m: 0.9699\n",
      "Epoch 1887/5000\n",
      " - 10s - loss: 0.0407 - acc: 0.9870 - precision_m: 0.9902 - recall_m: 0.9837 - val_loss: 0.0436 - val_acc: 0.9834 - val_precision_m: 0.9894 - val_recall_m: 0.9796\n",
      "Epoch 1888/5000\n",
      " - 10s - loss: 0.0353 - acc: 0.9894 - precision_m: 0.9933 - recall_m: 0.9855 - val_loss: 0.0474 - val_acc: 0.9828 - val_precision_m: 0.9928 - val_recall_m: 0.9762\n",
      "Epoch 1889/5000\n",
      " - 10s - loss: 0.0397 - acc: 0.9872 - precision_m: 0.9910 - recall_m: 0.9834 - val_loss: 0.0537 - val_acc: 0.9812 - val_precision_m: 0.9948 - val_recall_m: 0.9709\n",
      "Epoch 1890/5000\n",
      " - 10s - loss: 0.0374 - acc: 0.9887 - precision_m: 0.9925 - recall_m: 0.9851 - val_loss: 0.0533 - val_acc: 0.9856 - val_precision_m: 0.9827 - val_recall_m: 0.9910\n",
      "Epoch 1891/5000\n",
      " - 10s - loss: 0.0355 - acc: 0.9894 - precision_m: 0.9919 - recall_m: 0.9865 - val_loss: 0.0436 - val_acc: 0.9862 - val_precision_m: 0.9895 - val_recall_m: 0.9851\n",
      "Epoch 1892/5000\n",
      " - 10s - loss: 0.0351 - acc: 0.9892 - precision_m: 0.9928 - recall_m: 0.9855 - val_loss: 0.0672 - val_acc: 0.9790 - val_precision_m: 0.9682 - val_recall_m: 0.9942\n",
      "Epoch 1893/5000\n",
      " - 10s - loss: 0.0305 - acc: 0.9919 - precision_m: 0.9942 - recall_m: 0.9898 - val_loss: 0.0599 - val_acc: 0.9779 - val_precision_m: 0.9968 - val_recall_m: 0.9626\n",
      "Epoch 1894/5000\n",
      " - 10s - loss: 0.0340 - acc: 0.9899 - precision_m: 0.9941 - recall_m: 0.9862 - val_loss: 0.0399 - val_acc: 0.9884 - val_precision_m: 0.9907 - val_recall_m: 0.9878\n",
      "Epoch 1895/5000\n",
      " - 10s - loss: 0.0321 - acc: 0.9906 - precision_m: 0.9940 - recall_m: 0.9872 - val_loss: 0.0553 - val_acc: 0.9795 - val_precision_m: 0.9746 - val_recall_m: 0.9879\n",
      "Epoch 1896/5000\n",
      " - 10s - loss: 0.0382 - acc: 0.9880 - precision_m: 0.9915 - recall_m: 0.9846 - val_loss: 0.0428 - val_acc: 0.9839 - val_precision_m: 0.9807 - val_recall_m: 0.9900\n",
      "Epoch 1897/5000\n",
      " - 10s - loss: 0.0354 - acc: 0.9890 - precision_m: 0.9912 - recall_m: 0.9868 - val_loss: 0.0427 - val_acc: 0.9889 - val_precision_m: 0.9871 - val_recall_m: 0.9931\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1898/5000\n",
      " - 10s - loss: 0.0410 - acc: 0.9862 - precision_m: 0.9905 - recall_m: 0.9814 - val_loss: 0.0526 - val_acc: 0.9834 - val_precision_m: 0.9846 - val_recall_m: 0.9848\n",
      "Epoch 1899/5000\n",
      " - 10s - loss: 0.0327 - acc: 0.9902 - precision_m: 0.9939 - recall_m: 0.9864 - val_loss: 0.0523 - val_acc: 0.9812 - val_precision_m: 0.9948 - val_recall_m: 0.9706\n",
      "Epoch 1900/5000\n",
      " - 10s - loss: 0.0360 - acc: 0.9887 - precision_m: 0.9923 - recall_m: 0.9853 - val_loss: 0.0679 - val_acc: 0.9690 - val_precision_m: 0.9958 - val_recall_m: 0.9472\n",
      "Epoch 1901/5000\n",
      " - 10s - loss: 0.0371 - acc: 0.9886 - precision_m: 0.9909 - recall_m: 0.9863 - val_loss: 0.0509 - val_acc: 0.9845 - val_precision_m: 0.9799 - val_recall_m: 0.9920\n",
      "Epoch 1902/5000\n",
      " - 10s - loss: 0.0364 - acc: 0.9892 - precision_m: 0.9921 - recall_m: 0.9861 - val_loss: 0.0486 - val_acc: 0.9856 - val_precision_m: 0.9836 - val_recall_m: 0.9900\n",
      "Epoch 1903/5000\n",
      " - 10s - loss: 0.0355 - acc: 0.9894 - precision_m: 0.9924 - recall_m: 0.9863 - val_loss: 0.0447 - val_acc: 0.9856 - val_precision_m: 0.9856 - val_recall_m: 0.9877\n",
      "Epoch 1904/5000\n",
      " - 10s - loss: 0.0417 - acc: 0.9865 - precision_m: 0.9899 - recall_m: 0.9831 - val_loss: 0.0472 - val_acc: 0.9834 - val_precision_m: 0.9896 - val_recall_m: 0.9800\n",
      "Epoch 1905/5000\n",
      " - 10s - loss: 0.0367 - acc: 0.9887 - precision_m: 0.9916 - recall_m: 0.9855 - val_loss: 0.0471 - val_acc: 0.9834 - val_precision_m: 0.9768 - val_recall_m: 0.9929\n",
      "Epoch 1906/5000\n",
      " - 10s - loss: 0.0374 - acc: 0.9884 - precision_m: 0.9918 - recall_m: 0.9847 - val_loss: 0.0611 - val_acc: 0.9823 - val_precision_m: 0.9738 - val_recall_m: 0.9941\n",
      "Epoch 1907/5000\n",
      " - 10s - loss: 0.0340 - acc: 0.9897 - precision_m: 0.9921 - recall_m: 0.9872 - val_loss: 0.0449 - val_acc: 0.9823 - val_precision_m: 0.9937 - val_recall_m: 0.9739\n",
      "Epoch 1908/5000\n",
      " - 10s - loss: 0.0392 - acc: 0.9870 - precision_m: 0.9904 - recall_m: 0.9839 - val_loss: 0.0741 - val_acc: 0.9718 - val_precision_m: 0.9979 - val_recall_m: 0.9504\n",
      "Epoch 1909/5000\n",
      " - 10s - loss: 0.0332 - acc: 0.9904 - precision_m: 0.9934 - recall_m: 0.9874 - val_loss: 0.0438 - val_acc: 0.9850 - val_precision_m: 0.9844 - val_recall_m: 0.9882\n",
      "Epoch 1910/5000\n",
      " - 10s - loss: 0.0319 - acc: 0.9911 - precision_m: 0.9942 - recall_m: 0.9881 - val_loss: 0.0530 - val_acc: 0.9828 - val_precision_m: 0.9768 - val_recall_m: 0.9921\n",
      "Epoch 1911/5000\n",
      " - 10s - loss: 0.0365 - acc: 0.9882 - precision_m: 0.9921 - recall_m: 0.9845 - val_loss: 0.0514 - val_acc: 0.9856 - val_precision_m: 0.9826 - val_recall_m: 0.9910\n",
      "Epoch 1912/5000\n",
      " - 10s - loss: 0.0322 - acc: 0.9906 - precision_m: 0.9927 - recall_m: 0.9885 - val_loss: 0.0464 - val_acc: 0.9839 - val_precision_m: 0.9916 - val_recall_m: 0.9790\n",
      "Epoch 1913/5000\n",
      " - 10s - loss: 0.0346 - acc: 0.9894 - precision_m: 0.9922 - recall_m: 0.9864 - val_loss: 0.0420 - val_acc: 0.9873 - val_precision_m: 0.9948 - val_recall_m: 0.9819\n",
      "Epoch 1914/5000\n",
      " - 10s - loss: 0.0311 - acc: 0.9913 - precision_m: 0.9944 - recall_m: 0.9883 - val_loss: 0.0427 - val_acc: 0.9867 - val_precision_m: 0.9848 - val_recall_m: 0.9911\n",
      "Epoch 1915/5000\n",
      " - 10s - loss: 0.0337 - acc: 0.9892 - precision_m: 0.9917 - recall_m: 0.9867 - val_loss: 0.0513 - val_acc: 0.9828 - val_precision_m: 0.9827 - val_recall_m: 0.9858\n",
      "Epoch 1916/5000\n",
      " - 10s - loss: 0.0327 - acc: 0.9903 - precision_m: 0.9935 - recall_m: 0.9873 - val_loss: 0.0395 - val_acc: 0.9867 - val_precision_m: 0.9926 - val_recall_m: 0.9828\n",
      "Epoch 1917/5000\n",
      " - 10s - loss: 0.0397 - acc: 0.9867 - precision_m: 0.9903 - recall_m: 0.9834 - val_loss: 0.0429 - val_acc: 0.9878 - val_precision_m: 0.9848 - val_recall_m: 0.9931\n",
      "Epoch 1918/5000\n",
      " - 10s - loss: 0.0320 - acc: 0.9914 - precision_m: 0.9944 - recall_m: 0.9884 - val_loss: 0.0415 - val_acc: 0.9856 - val_precision_m: 0.9865 - val_recall_m: 0.9870\n",
      "Epoch 1919/5000\n",
      " - 10s - loss: 0.0347 - acc: 0.9894 - precision_m: 0.9934 - recall_m: 0.9855 - val_loss: 0.0475 - val_acc: 0.9812 - val_precision_m: 0.9937 - val_recall_m: 0.9718\n",
      "Epoch 1920/5000\n",
      " - 10s - loss: 0.0330 - acc: 0.9903 - precision_m: 0.9932 - recall_m: 0.9874 - val_loss: 0.0439 - val_acc: 0.9856 - val_precision_m: 0.9899 - val_recall_m: 0.9838\n",
      "Epoch 1921/5000\n",
      " - 10s - loss: 0.0318 - acc: 0.9908 - precision_m: 0.9936 - recall_m: 0.9879 - val_loss: 0.0388 - val_acc: 0.9878 - val_precision_m: 0.9866 - val_recall_m: 0.9908\n",
      "Epoch 1922/5000\n",
      " - 10s - loss: 0.0414 - acc: 0.9866 - precision_m: 0.9901 - recall_m: 0.9834 - val_loss: 0.0468 - val_acc: 0.9806 - val_precision_m: 0.9905 - val_recall_m: 0.9737\n",
      "Epoch 1923/5000\n",
      " - 10s - loss: 0.0331 - acc: 0.9900 - precision_m: 0.9934 - recall_m: 0.9866 - val_loss: 0.0433 - val_acc: 0.9856 - val_precision_m: 0.9845 - val_recall_m: 0.9893\n",
      "Epoch 1924/5000\n",
      " - 10s - loss: 0.0313 - acc: 0.9910 - precision_m: 0.9944 - recall_m: 0.9876 - val_loss: 0.0395 - val_acc: 0.9867 - val_precision_m: 0.9856 - val_recall_m: 0.9901\n",
      "Epoch 1925/5000\n",
      " - 10s - loss: 0.0320 - acc: 0.9909 - precision_m: 0.9939 - recall_m: 0.9879 - val_loss: 0.0627 - val_acc: 0.9801 - val_precision_m: 0.9705 - val_recall_m: 0.9931\n",
      "Epoch 1926/5000\n",
      " - 10s - loss: 0.0381 - acc: 0.9884 - precision_m: 0.9917 - recall_m: 0.9851 - val_loss: 0.0427 - val_acc: 0.9850 - val_precision_m: 0.9938 - val_recall_m: 0.9790\n",
      "Epoch 1927/5000\n",
      " - 10s - loss: 0.0345 - acc: 0.9897 - precision_m: 0.9928 - recall_m: 0.9866 - val_loss: 0.0667 - val_acc: 0.9801 - val_precision_m: 0.9711 - val_recall_m: 0.9929\n",
      "Epoch 1928/5000\n",
      " - 10s - loss: 0.0360 - acc: 0.9894 - precision_m: 0.9927 - recall_m: 0.9862 - val_loss: 0.0519 - val_acc: 0.9812 - val_precision_m: 0.9969 - val_recall_m: 0.9687\n",
      "Epoch 1929/5000\n",
      " - 10s - loss: 0.0413 - acc: 0.9862 - precision_m: 0.9906 - recall_m: 0.9819 - val_loss: 0.0555 - val_acc: 0.9817 - val_precision_m: 0.9739 - val_recall_m: 0.9930\n",
      "Epoch 1930/5000\n",
      " - 10s - loss: 0.0332 - acc: 0.9906 - precision_m: 0.9928 - recall_m: 0.9886 - val_loss: 0.0392 - val_acc: 0.9873 - val_precision_m: 0.9866 - val_recall_m: 0.9901\n",
      "Epoch 1931/5000\n",
      " - 10s - loss: 0.0314 - acc: 0.9910 - precision_m: 0.9935 - recall_m: 0.9884 - val_loss: 0.0911 - val_acc: 0.9590 - val_precision_m: 0.9979 - val_recall_m: 0.9270\n",
      "Epoch 1932/5000\n",
      " - 10s - loss: 0.0354 - acc: 0.9891 - precision_m: 0.9912 - recall_m: 0.9868 - val_loss: 0.0438 - val_acc: 0.9850 - val_precision_m: 0.9925 - val_recall_m: 0.9798\n",
      "Epoch 1933/5000\n",
      " - 10s - loss: 0.0322 - acc: 0.9910 - precision_m: 0.9941 - recall_m: 0.9877 - val_loss: 0.0403 - val_acc: 0.9873 - val_precision_m: 0.9855 - val_recall_m: 0.9912\n",
      "Epoch 1934/5000\n",
      " - 10s - loss: 0.0326 - acc: 0.9901 - precision_m: 0.9932 - recall_m: 0.9873 - val_loss: 0.0443 - val_acc: 0.9839 - val_precision_m: 0.9896 - val_recall_m: 0.9809\n",
      "Epoch 1935/5000\n",
      " - 10s - loss: 0.0348 - acc: 0.9894 - precision_m: 0.9924 - recall_m: 0.9865 - val_loss: 0.0461 - val_acc: 0.9856 - val_precision_m: 0.9916 - val_recall_m: 0.9819\n",
      "Epoch 1936/5000\n",
      " - 10s - loss: 0.0383 - acc: 0.9881 - precision_m: 0.9911 - recall_m: 0.9853 - val_loss: 0.0585 - val_acc: 0.9828 - val_precision_m: 0.9737 - val_recall_m: 0.9949\n",
      "Epoch 1937/5000\n",
      " - 10s - loss: 0.0332 - acc: 0.9900 - precision_m: 0.9935 - recall_m: 0.9865 - val_loss: 0.0383 - val_acc: 0.9889 - val_precision_m: 0.9918 - val_recall_m: 0.9880\n",
      "Epoch 1938/5000\n",
      " - 10s - loss: 0.0360 - acc: 0.9884 - precision_m: 0.9918 - recall_m: 0.9850 - val_loss: 0.0465 - val_acc: 0.9839 - val_precision_m: 0.9969 - val_recall_m: 0.9739\n",
      "Epoch 1939/5000\n",
      " - 10s - loss: 0.0385 - acc: 0.9873 - precision_m: 0.9910 - recall_m: 0.9839 - val_loss: 0.0771 - val_acc: 0.9745 - val_precision_m: 0.9616 - val_recall_m: 0.9930\n",
      "Epoch 1940/5000\n",
      " - 10s - loss: 0.0319 - acc: 0.9902 - precision_m: 0.9934 - recall_m: 0.9872 - val_loss: 0.0414 - val_acc: 0.9839 - val_precision_m: 0.9914 - val_recall_m: 0.9792\n",
      "Epoch 1941/5000\n",
      " - 10s - loss: 0.0314 - acc: 0.9910 - precision_m: 0.9946 - recall_m: 0.9873 - val_loss: 0.0360 - val_acc: 0.9884 - val_precision_m: 0.9928 - val_recall_m: 0.9861\n",
      "Epoch 1942/5000\n",
      " - 10s - loss: 0.0346 - acc: 0.9902 - precision_m: 0.9933 - recall_m: 0.9873 - val_loss: 0.0507 - val_acc: 0.9823 - val_precision_m: 0.9817 - val_recall_m: 0.9860\n",
      "Epoch 1943/5000\n",
      " - 10s - loss: 0.0314 - acc: 0.9911 - precision_m: 0.9943 - recall_m: 0.9878 - val_loss: 0.0506 - val_acc: 0.9801 - val_precision_m: 0.9959 - val_recall_m: 0.9675\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1944/5000\n",
      " - 10s - loss: 0.0307 - acc: 0.9910 - precision_m: 0.9935 - recall_m: 0.9886 - val_loss: 0.0460 - val_acc: 0.9850 - val_precision_m: 0.9817 - val_recall_m: 0.9910\n",
      "Epoch 1945/5000\n",
      " - 10s - loss: 0.0290 - acc: 0.9913 - precision_m: 0.9940 - recall_m: 0.9885 - val_loss: 0.0450 - val_acc: 0.9856 - val_precision_m: 0.9826 - val_recall_m: 0.9912\n",
      "Epoch 1946/5000\n",
      " - 10s - loss: 0.0315 - acc: 0.9911 - precision_m: 0.9938 - recall_m: 0.9881 - val_loss: 0.0441 - val_acc: 0.9850 - val_precision_m: 0.9825 - val_recall_m: 0.9899\n",
      "Epoch 1947/5000\n",
      " - 10s - loss: 0.0330 - acc: 0.9903 - precision_m: 0.9936 - recall_m: 0.9872 - val_loss: 0.0447 - val_acc: 0.9845 - val_precision_m: 0.9905 - val_recall_m: 0.9811\n",
      "Epoch 1948/5000\n",
      " - 10s - loss: 0.0348 - acc: 0.9897 - precision_m: 0.9933 - recall_m: 0.9864 - val_loss: 0.0603 - val_acc: 0.9784 - val_precision_m: 0.9979 - val_recall_m: 0.9629\n",
      "Epoch 1949/5000\n",
      " - 10s - loss: 0.0295 - acc: 0.9913 - precision_m: 0.9946 - recall_m: 0.9879 - val_loss: 0.0480 - val_acc: 0.9817 - val_precision_m: 0.9948 - val_recall_m: 0.9716\n",
      "Epoch 1950/5000\n",
      " - 10s - loss: 0.0367 - acc: 0.9885 - precision_m: 0.9922 - recall_m: 0.9847 - val_loss: 0.0549 - val_acc: 0.9823 - val_precision_m: 0.9796 - val_recall_m: 0.9880\n",
      "Epoch 1951/5000\n",
      " - 10s - loss: 0.0338 - acc: 0.9900 - precision_m: 0.9930 - recall_m: 0.9872 - val_loss: 0.0402 - val_acc: 0.9867 - val_precision_m: 0.9915 - val_recall_m: 0.9841\n",
      "Epoch 1952/5000\n",
      " - 10s - loss: 0.0273 - acc: 0.9926 - precision_m: 0.9953 - recall_m: 0.9899 - val_loss: 0.0377 - val_acc: 0.9878 - val_precision_m: 0.9910 - val_recall_m: 0.9872\n",
      "Epoch 1953/5000\n",
      " - 10s - loss: 0.0330 - acc: 0.9898 - precision_m: 0.9917 - recall_m: 0.9880 - val_loss: 0.0405 - val_acc: 0.9878 - val_precision_m: 0.9917 - val_recall_m: 0.9862\n",
      "Epoch 1954/5000\n",
      " - 10s - loss: 0.0369 - acc: 0.9885 - precision_m: 0.9924 - recall_m: 0.9848 - val_loss: 0.0562 - val_acc: 0.9790 - val_precision_m: 0.9959 - val_recall_m: 0.9654\n",
      "Epoch 1955/5000\n",
      " - 10s - loss: 0.0329 - acc: 0.9904 - precision_m: 0.9939 - recall_m: 0.9869 - val_loss: 0.0439 - val_acc: 0.9862 - val_precision_m: 0.9817 - val_recall_m: 0.9930\n",
      "Epoch 1956/5000\n",
      " - 10s - loss: 0.0289 - acc: 0.9926 - precision_m: 0.9946 - recall_m: 0.9906 - val_loss: 0.0481 - val_acc: 0.9828 - val_precision_m: 0.9777 - val_recall_m: 0.9912\n",
      "Epoch 1957/5000\n",
      " - 10s - loss: 0.0366 - acc: 0.9881 - precision_m: 0.9920 - recall_m: 0.9844 - val_loss: 0.0410 - val_acc: 0.9873 - val_precision_m: 0.9856 - val_recall_m: 0.9910\n",
      "Epoch 1958/5000\n",
      " - 10s - loss: 0.0347 - acc: 0.9895 - precision_m: 0.9931 - recall_m: 0.9857 - val_loss: 0.0596 - val_acc: 0.9828 - val_precision_m: 0.9757 - val_recall_m: 0.9929\n",
      "Epoch 1959/5000\n",
      " - 10s - loss: 0.0354 - acc: 0.9889 - precision_m: 0.9919 - recall_m: 0.9857 - val_loss: 0.0731 - val_acc: 0.9701 - val_precision_m: 0.9968 - val_recall_m: 0.9480\n",
      "Epoch 1960/5000\n",
      " - 10s - loss: 0.0298 - acc: 0.9908 - precision_m: 0.9940 - recall_m: 0.9876 - val_loss: 0.0496 - val_acc: 0.9845 - val_precision_m: 0.9786 - val_recall_m: 0.9931\n",
      "Epoch 1961/5000\n",
      " - 10s - loss: 0.0343 - acc: 0.9892 - precision_m: 0.9923 - recall_m: 0.9862 - val_loss: 0.0408 - val_acc: 0.9828 - val_precision_m: 0.9834 - val_recall_m: 0.9850\n",
      "Epoch 1962/5000\n",
      " - 10s - loss: 0.0363 - acc: 0.9881 - precision_m: 0.9916 - recall_m: 0.9848 - val_loss: 0.0404 - val_acc: 0.9856 - val_precision_m: 0.9948 - val_recall_m: 0.9791\n",
      "Epoch 1963/5000\n",
      " - 10s - loss: 0.0280 - acc: 0.9919 - precision_m: 0.9951 - recall_m: 0.9884 - val_loss: 0.0365 - val_acc: 0.9884 - val_precision_m: 0.9915 - val_recall_m: 0.9869\n",
      "Epoch 1964/5000\n",
      " - 10s - loss: 0.0427 - acc: 0.9860 - precision_m: 0.9907 - recall_m: 0.9812 - val_loss: 0.0425 - val_acc: 0.9856 - val_precision_m: 0.9876 - val_recall_m: 0.9857\n",
      "Epoch 1965/5000\n",
      " - 10s - loss: 0.0356 - acc: 0.9887 - precision_m: 0.9913 - recall_m: 0.9863 - val_loss: 0.0397 - val_acc: 0.9873 - val_precision_m: 0.9949 - val_recall_m: 0.9819\n",
      "Epoch 1966/5000\n",
      " - 10s - loss: 0.0313 - acc: 0.9908 - precision_m: 0.9941 - recall_m: 0.9873 - val_loss: 0.0405 - val_acc: 0.9856 - val_precision_m: 0.9846 - val_recall_m: 0.9891\n",
      "Epoch 1967/5000\n",
      " - 10s - loss: 0.0378 - acc: 0.9874 - precision_m: 0.9908 - recall_m: 0.9841 - val_loss: 0.0387 - val_acc: 0.9850 - val_precision_m: 0.9877 - val_recall_m: 0.9849\n",
      "Epoch 1968/5000\n",
      " - 10s - loss: 0.0301 - acc: 0.9913 - precision_m: 0.9940 - recall_m: 0.9885 - val_loss: 0.0374 - val_acc: 0.9878 - val_precision_m: 0.9867 - val_recall_m: 0.9912\n",
      "Epoch 1969/5000\n",
      " - 10s - loss: 0.0280 - acc: 0.9923 - precision_m: 0.9947 - recall_m: 0.9899 - val_loss: 0.0373 - val_acc: 0.9884 - val_precision_m: 0.9939 - val_recall_m: 0.9851\n",
      "Epoch 1970/5000\n",
      " - 10s - loss: 0.0324 - acc: 0.9900 - precision_m: 0.9928 - recall_m: 0.9870 - val_loss: 0.0439 - val_acc: 0.9834 - val_precision_m: 0.9927 - val_recall_m: 0.9769\n",
      "Epoch 1971/5000\n",
      " - 10s - loss: 0.0292 - acc: 0.9917 - precision_m: 0.9939 - recall_m: 0.9895 - val_loss: 0.0407 - val_acc: 0.9884 - val_precision_m: 0.9878 - val_recall_m: 0.9911\n",
      "Epoch 1972/5000\n",
      " - 10s - loss: 0.0297 - acc: 0.9911 - precision_m: 0.9935 - recall_m: 0.9887 - val_loss: 0.0899 - val_acc: 0.9707 - val_precision_m: 0.9514 - val_recall_m: 0.9970\n",
      "Epoch 1973/5000\n",
      " - 10s - loss: 0.0383 - acc: 0.9879 - precision_m: 0.9914 - recall_m: 0.9845 - val_loss: 0.0400 - val_acc: 0.9878 - val_precision_m: 0.9857 - val_recall_m: 0.9921\n",
      "Epoch 1974/5000\n",
      " - 10s - loss: 0.0313 - acc: 0.9900 - precision_m: 0.9932 - recall_m: 0.9869 - val_loss: 0.0393 - val_acc: 0.9878 - val_precision_m: 0.9888 - val_recall_m: 0.9889\n",
      "Epoch 1975/5000\n",
      " - 10s - loss: 0.0288 - acc: 0.9919 - precision_m: 0.9952 - recall_m: 0.9887 - val_loss: 0.0366 - val_acc: 0.9889 - val_precision_m: 0.9876 - val_recall_m: 0.9921\n",
      "Epoch 1976/5000\n",
      " - 10s - loss: 0.0325 - acc: 0.9900 - precision_m: 0.9932 - recall_m: 0.9868 - val_loss: 0.0719 - val_acc: 0.9718 - val_precision_m: 0.9958 - val_recall_m: 0.9523\n",
      "Epoch 1977/5000\n",
      " - 10s - loss: 0.0289 - acc: 0.9924 - precision_m: 0.9955 - recall_m: 0.9893 - val_loss: 0.0401 - val_acc: 0.9862 - val_precision_m: 0.9927 - val_recall_m: 0.9818\n",
      "Epoch 1978/5000\n",
      " - 10s - loss: 0.0313 - acc: 0.9903 - precision_m: 0.9934 - recall_m: 0.9875 - val_loss: 0.0437 - val_acc: 0.9862 - val_precision_m: 0.9826 - val_recall_m: 0.9921\n",
      "Epoch 1979/5000\n",
      " - 10s - loss: 0.0289 - acc: 0.9914 - precision_m: 0.9945 - recall_m: 0.9882 - val_loss: 0.0339 - val_acc: 0.9884 - val_precision_m: 0.9885 - val_recall_m: 0.9898\n",
      "Epoch 1980/5000\n",
      " - 10s - loss: 0.0290 - acc: 0.9921 - precision_m: 0.9949 - recall_m: 0.9894 - val_loss: 0.0430 - val_acc: 0.9873 - val_precision_m: 0.9807 - val_recall_m: 0.9959\n",
      "Epoch 1981/5000\n",
      " - 10s - loss: 0.0367 - acc: 0.9881 - precision_m: 0.9913 - recall_m: 0.9852 - val_loss: 0.0610 - val_acc: 0.9828 - val_precision_m: 0.9750 - val_recall_m: 0.9940\n",
      "Epoch 1982/5000\n",
      " - 10s - loss: 0.0465 - acc: 0.9842 - precision_m: 0.9887 - recall_m: 0.9802 - val_loss: 0.0354 - val_acc: 0.9878 - val_precision_m: 0.9905 - val_recall_m: 0.9867\n",
      "Epoch 1983/5000\n",
      " - 10s - loss: 0.0262 - acc: 0.9930 - precision_m: 0.9955 - recall_m: 0.9905 - val_loss: 0.0378 - val_acc: 0.9878 - val_precision_m: 0.9865 - val_recall_m: 0.9908\n",
      "Epoch 1984/5000\n",
      " - 10s - loss: 0.0299 - acc: 0.9918 - precision_m: 0.9948 - recall_m: 0.9886 - val_loss: 0.0474 - val_acc: 0.9790 - val_precision_m: 0.9948 - val_recall_m: 0.9665\n",
      "Epoch 1985/5000\n",
      " - 10s - loss: 0.0320 - acc: 0.9903 - precision_m: 0.9934 - recall_m: 0.9872 - val_loss: 0.0491 - val_acc: 0.9845 - val_precision_m: 0.9837 - val_recall_m: 0.9880\n",
      "Epoch 1986/5000\n",
      " - 10s - loss: 0.0393 - acc: 0.9865 - precision_m: 0.9896 - recall_m: 0.9835 - val_loss: 0.0716 - val_acc: 0.9662 - val_precision_m: 0.9968 - val_recall_m: 0.9411\n",
      "Epoch 1987/5000\n",
      " - 10s - loss: 0.0402 - acc: 0.9873 - precision_m: 0.9904 - recall_m: 0.9838 - val_loss: 0.0479 - val_acc: 0.9812 - val_precision_m: 0.9884 - val_recall_m: 0.9769\n",
      "Epoch 1988/5000\n",
      " - 10s - loss: 0.0300 - acc: 0.9905 - precision_m: 0.9937 - recall_m: 0.9871 - val_loss: 0.0471 - val_acc: 0.9828 - val_precision_m: 0.9948 - val_recall_m: 0.9739\n",
      "Epoch 1989/5000\n",
      " - 10s - loss: 0.0295 - acc: 0.9913 - precision_m: 0.9949 - recall_m: 0.9878 - val_loss: 0.0359 - val_acc: 0.9873 - val_precision_m: 0.9919 - val_recall_m: 0.9853\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1990/5000\n",
      " - 10s - loss: 0.0293 - acc: 0.9910 - precision_m: 0.9928 - recall_m: 0.9893 - val_loss: 0.0425 - val_acc: 0.9873 - val_precision_m: 0.9959 - val_recall_m: 0.9810\n",
      "Epoch 1991/5000\n",
      " - 10s - loss: 0.0282 - acc: 0.9921 - precision_m: 0.9948 - recall_m: 0.9895 - val_loss: 0.0358 - val_acc: 0.9895 - val_precision_m: 0.9937 - val_recall_m: 0.9868\n",
      "Epoch 1992/5000\n",
      " - 10s - loss: 0.0292 - acc: 0.9913 - precision_m: 0.9944 - recall_m: 0.9881 - val_loss: 0.0451 - val_acc: 0.9850 - val_precision_m: 0.9979 - val_recall_m: 0.9748\n",
      "Epoch 1993/5000\n",
      " - 10s - loss: 0.0263 - acc: 0.9926 - precision_m: 0.9950 - recall_m: 0.9903 - val_loss: 0.0351 - val_acc: 0.9878 - val_precision_m: 0.9918 - val_recall_m: 0.9858\n",
      "Epoch 1994/5000\n",
      " - 10s - loss: 0.0287 - acc: 0.9914 - precision_m: 0.9937 - recall_m: 0.9888 - val_loss: 0.0403 - val_acc: 0.9867 - val_precision_m: 0.9918 - val_recall_m: 0.9838\n",
      "Epoch 1995/5000\n",
      " - 10s - loss: 0.0306 - acc: 0.9905 - precision_m: 0.9926 - recall_m: 0.9884 - val_loss: 0.0397 - val_acc: 0.9873 - val_precision_m: 0.9905 - val_recall_m: 0.9861\n",
      "Epoch 1996/5000\n",
      " - 10s - loss: 0.0289 - acc: 0.9913 - precision_m: 0.9941 - recall_m: 0.9883 - val_loss: 0.0401 - val_acc: 0.9856 - val_precision_m: 0.9837 - val_recall_m: 0.9898\n",
      "Epoch 1997/5000\n",
      " - 10s - loss: 0.0260 - acc: 0.9930 - precision_m: 0.9953 - recall_m: 0.9905 - val_loss: 0.0377 - val_acc: 0.9884 - val_precision_m: 0.9877 - val_recall_m: 0.9910\n",
      "Epoch 1998/5000\n",
      " - 10s - loss: 0.0298 - acc: 0.9911 - precision_m: 0.9948 - recall_m: 0.9874 - val_loss: 0.0433 - val_acc: 0.9856 - val_precision_m: 0.9796 - val_recall_m: 0.9940\n",
      "Epoch 1999/5000\n",
      " - 10s - loss: 0.0311 - acc: 0.9904 - precision_m: 0.9934 - recall_m: 0.9876 - val_loss: 0.0453 - val_acc: 0.9828 - val_precision_m: 0.9959 - val_recall_m: 0.9728\n",
      "Epoch 2000/5000\n",
      " - 10s - loss: 0.0293 - acc: 0.9907 - precision_m: 0.9939 - recall_m: 0.9874 - val_loss: 0.0382 - val_acc: 0.9878 - val_precision_m: 0.9959 - val_recall_m: 0.9820\n",
      "Epoch 2001/5000\n",
      " - 10s - loss: 0.0309 - acc: 0.9905 - precision_m: 0.9939 - recall_m: 0.9873 - val_loss: 0.0367 - val_acc: 0.9867 - val_precision_m: 0.9948 - val_recall_m: 0.9809\n",
      "Epoch 2002/5000\n",
      " - 10s - loss: 0.0324 - acc: 0.9902 - precision_m: 0.9930 - recall_m: 0.9873 - val_loss: 0.0387 - val_acc: 0.9862 - val_precision_m: 0.9900 - val_recall_m: 0.9850\n",
      "Epoch 2003/5000\n",
      " - 10s - loss: 0.0348 - acc: 0.9885 - precision_m: 0.9923 - recall_m: 0.9846 - val_loss: 0.0411 - val_acc: 0.9862 - val_precision_m: 0.9816 - val_recall_m: 0.9930\n",
      "Epoch 2004/5000\n",
      " - 10s - loss: 0.0353 - acc: 0.9876 - precision_m: 0.9911 - recall_m: 0.9841 - val_loss: 0.0381 - val_acc: 0.9862 - val_precision_m: 0.9916 - val_recall_m: 0.9832\n",
      "Epoch 2005/5000\n",
      " - 10s - loss: 0.0296 - acc: 0.9910 - precision_m: 0.9937 - recall_m: 0.9884 - val_loss: 0.0449 - val_acc: 0.9784 - val_precision_m: 0.9948 - val_recall_m: 0.9654\n",
      "Epoch 2006/5000\n",
      " - 10s - loss: 0.0311 - acc: 0.9903 - precision_m: 0.9933 - recall_m: 0.9877 - val_loss: 0.0408 - val_acc: 0.9856 - val_precision_m: 0.9959 - val_recall_m: 0.9777\n",
      "Epoch 2007/5000\n",
      " - 10s - loss: 0.0276 - acc: 0.9918 - precision_m: 0.9946 - recall_m: 0.9889 - val_loss: 0.0401 - val_acc: 0.9850 - val_precision_m: 0.9959 - val_recall_m: 0.9769\n",
      "Epoch 2008/5000\n",
      " - 10s - loss: 0.0352 - acc: 0.9887 - precision_m: 0.9912 - recall_m: 0.9862 - val_loss: 0.0433 - val_acc: 0.9878 - val_precision_m: 0.9837 - val_recall_m: 0.9941\n",
      "Epoch 2009/5000\n",
      " - 10s - loss: 0.0305 - acc: 0.9895 - precision_m: 0.9927 - recall_m: 0.9860 - val_loss: 0.0404 - val_acc: 0.9878 - val_precision_m: 0.9878 - val_recall_m: 0.9899\n",
      "Epoch 2010/5000\n",
      " - 10s - loss: 0.0414 - acc: 0.9871 - precision_m: 0.9907 - recall_m: 0.9834 - val_loss: 0.0591 - val_acc: 0.9823 - val_precision_m: 0.9740 - val_recall_m: 0.9940\n",
      "Epoch 2011/5000\n",
      " - 10s - loss: 0.0435 - acc: 0.9860 - precision_m: 0.9902 - recall_m: 0.9824 - val_loss: 0.0400 - val_acc: 0.9867 - val_precision_m: 0.9937 - val_recall_m: 0.9818\n",
      "Epoch 2012/5000\n",
      " - 10s - loss: 0.0343 - acc: 0.9892 - precision_m: 0.9924 - recall_m: 0.9862 - val_loss: 0.0464 - val_acc: 0.9806 - val_precision_m: 0.9948 - val_recall_m: 0.9695\n",
      "Epoch 2013/5000\n",
      " - 10s - loss: 0.0267 - acc: 0.9924 - precision_m: 0.9945 - recall_m: 0.9903 - val_loss: 0.0355 - val_acc: 0.9889 - val_precision_m: 0.9887 - val_recall_m: 0.9910\n",
      "Epoch 2014/5000\n",
      " - 10s - loss: 0.0283 - acc: 0.9916 - precision_m: 0.9939 - recall_m: 0.9892 - val_loss: 0.0379 - val_acc: 0.9889 - val_precision_m: 0.9937 - val_recall_m: 0.9858\n",
      "Epoch 2015/5000\n",
      " - 10s - loss: 0.0374 - acc: 0.9889 - precision_m: 0.9922 - recall_m: 0.9861 - val_loss: 0.0458 - val_acc: 0.9828 - val_precision_m: 0.9924 - val_recall_m: 0.9759\n",
      "Epoch 2016/5000\n",
      " - 10s - loss: 0.0299 - acc: 0.9913 - precision_m: 0.9938 - recall_m: 0.9886 - val_loss: 0.0446 - val_acc: 0.9845 - val_precision_m: 0.9959 - val_recall_m: 0.9758\n",
      "Epoch 2017/5000\n",
      " - 10s - loss: 0.0272 - acc: 0.9924 - precision_m: 0.9947 - recall_m: 0.9901 - val_loss: 0.0519 - val_acc: 0.9790 - val_precision_m: 0.9958 - val_recall_m: 0.9657\n",
      "Epoch 2018/5000\n",
      " - 10s - loss: 0.0288 - acc: 0.9919 - precision_m: 0.9940 - recall_m: 0.9899 - val_loss: 0.0363 - val_acc: 0.9862 - val_precision_m: 0.9887 - val_recall_m: 0.9859\n",
      "Epoch 2019/5000\n",
      " - 10s - loss: 0.0306 - acc: 0.9905 - precision_m: 0.9935 - recall_m: 0.9875 - val_loss: 0.0367 - val_acc: 0.9884 - val_precision_m: 0.9847 - val_recall_m: 0.9938\n",
      "Epoch 2020/5000\n",
      " - 10s - loss: 0.0292 - acc: 0.9914 - precision_m: 0.9943 - recall_m: 0.9886 - val_loss: 0.0578 - val_acc: 0.9790 - val_precision_m: 0.9968 - val_recall_m: 0.9644\n",
      "Epoch 2021/5000\n",
      " - 10s - loss: 0.0262 - acc: 0.9932 - precision_m: 0.9955 - recall_m: 0.9908 - val_loss: 0.0325 - val_acc: 0.9889 - val_precision_m: 0.9876 - val_recall_m: 0.9919\n",
      "Epoch 2022/5000\n",
      " - 10s - loss: 0.0268 - acc: 0.9922 - precision_m: 0.9947 - recall_m: 0.9898 - val_loss: 0.0351 - val_acc: 0.9884 - val_precision_m: 0.9949 - val_recall_m: 0.9840\n",
      "Epoch 2023/5000\n",
      " - 10s - loss: 0.0274 - acc: 0.9922 - precision_m: 0.9948 - recall_m: 0.9897 - val_loss: 0.0392 - val_acc: 0.9862 - val_precision_m: 0.9948 - val_recall_m: 0.9796\n",
      "Epoch 2024/5000\n",
      " - 10s - loss: 0.0311 - acc: 0.9899 - precision_m: 0.9924 - recall_m: 0.9876 - val_loss: 0.0368 - val_acc: 0.9911 - val_precision_m: 0.9927 - val_recall_m: 0.9910\n",
      "Epoch 2025/5000\n",
      " - 10s - loss: 0.0299 - acc: 0.9910 - precision_m: 0.9932 - recall_m: 0.9889 - val_loss: 0.0339 - val_acc: 0.9895 - val_precision_m: 0.9917 - val_recall_m: 0.9888\n",
      "Epoch 2026/5000\n",
      " - 10s - loss: 0.0289 - acc: 0.9919 - precision_m: 0.9950 - recall_m: 0.9888 - val_loss: 0.0347 - val_acc: 0.9895 - val_precision_m: 0.9887 - val_recall_m: 0.9920\n",
      "Epoch 2027/5000\n",
      " - 10s - loss: 0.0255 - acc: 0.9926 - precision_m: 0.9950 - recall_m: 0.9904 - val_loss: 0.0818 - val_acc: 0.9651 - val_precision_m: 1.0000 - val_recall_m: 0.9359\n",
      "Epoch 2028/5000\n",
      " - 10s - loss: 0.0290 - acc: 0.9913 - precision_m: 0.9945 - recall_m: 0.9883 - val_loss: 0.0331 - val_acc: 0.9889 - val_precision_m: 0.9949 - val_recall_m: 0.9850\n",
      "Epoch 2029/5000\n",
      " - 10s - loss: 0.0352 - acc: 0.9890 - precision_m: 0.9925 - recall_m: 0.9855 - val_loss: 0.0556 - val_acc: 0.9828 - val_precision_m: 0.9743 - val_recall_m: 0.9949\n",
      "Epoch 2030/5000\n",
      " - 10s - loss: 0.0406 - acc: 0.9874 - precision_m: 0.9912 - recall_m: 0.9840 - val_loss: 0.0566 - val_acc: 0.9751 - val_precision_m: 0.9969 - val_recall_m: 0.9569\n",
      "Epoch 2031/5000\n",
      " - 10s - loss: 0.0321 - acc: 0.9898 - precision_m: 0.9935 - recall_m: 0.9863 - val_loss: 0.0338 - val_acc: 0.9895 - val_precision_m: 0.9926 - val_recall_m: 0.9880\n",
      "Epoch 2032/5000\n",
      " - 10s - loss: 0.0258 - acc: 0.9927 - precision_m: 0.9941 - recall_m: 0.9913 - val_loss: 0.0357 - val_acc: 0.9867 - val_precision_m: 0.9948 - val_recall_m: 0.9807\n",
      "Epoch 2033/5000\n",
      " - 10s - loss: 0.0249 - acc: 0.9929 - precision_m: 0.9942 - recall_m: 0.9914 - val_loss: 0.0310 - val_acc: 0.9906 - val_precision_m: 0.9939 - val_recall_m: 0.9887\n",
      "\n",
      "Reached perfect accuracy so cancelling training!\n"
     ]
    }
   ],
   "source": [
    "# train the model\n",
    "bsize = 128\n",
    "history = model.fit(X_train, y_train, validation_data=(X_valid, y_valid), epochs=5000, batch_size = bsize, verbose=2, shuffle = True, callbacks = [epoch_schedule])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Save the training history"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# convert the history.history dict to a pandas DataFrame:     \n",
    "hist_df = pd.DataFrame(history.history) \n",
    "\n",
    "# or save to csv: \n",
    "hist_csv_file = 'historyDetect2class_1dcnn.csv'\n",
    "with open(hist_csv_file, mode='w', newline='') as f:\n",
    "    hist_df.to_csv(f)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "ibf8nlWdBNgH"
   },
   "source": [
    "## Plots"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Binary Cross-entropy Loss curve"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 474
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 60809,
     "status": "ok",
     "timestamp": 1571627063845,
     "user": {
      "displayName": "Mahindra Singh Rautela",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mBTSnpeHr1j42u0w7ABMmG3BYbWyCtFQVbOvr6sAA=s64",
      "userId": "15859880813264051870"
     },
     "user_tz": -330
    },
    "id": "kySYDbsQ7uON",
    "outputId": "4e673bae-c760-4b52-a4a1-0460bb47ab57"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#---Summarize history for loss\n",
    "plt.figure(figsize=(20,5))\n",
    "plt.plot(history.history['loss'],'-o')\n",
    "plt.plot(history.history['val_loss'],'-s')\n",
    "plt.title('Loss curve',fontsize=20)\n",
    "plt.ylabel('Binary crossentropy loss',fontsize=18)\n",
    "#plt.grid()\n",
    "plt.xticks(fontsize=18)\n",
    "plt.yticks(fontsize=18)\n",
    "plt.xlabel('Number of epochs',fontsize=18)\n",
    "plt.legend(['train', 'test'], loc='upper right',fontsize=18)\n",
    "#plt.axis([0,1000,0,500])\n",
    "plt.show()\n",
    "#plt.savefig('LossCurve_2class_1dcnn.eps', format='eps')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Accuracy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#---Summarize history for accuracy\n",
    "plt.figure(figsize=(20,5))\n",
    "plt.plot(history.history['acc'],'-o')\n",
    "plt.plot(history.history['val_acc'],'-s')\n",
    "plt.title('Accuracy Curve',fontsize=20)\n",
    "plt.ylabel('Accuracy',fontsize=18)\n",
    "#plt.grid()\n",
    "plt.xticks(fontsize=18)\n",
    "plt.yticks(fontsize=18)\n",
    "plt.xlabel('Number of epochs',fontsize=18)\n",
    "plt.legend(['train', 'test'], loc='lower right',fontsize=18)\n",
    "#plt.axis([0,1000,0,500])\n",
    "plt.show()\n",
    "#plt.savefig('AccCurve_2class_1dcnn.eps', format='eps')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Recall"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#---Summarize history for Recall\n",
    "plt.figure(figsize=(20,5))\n",
    "plt.plot(history.history['recall_m'],'-o')\n",
    "plt.plot(history.history['val_recall_m'],'-s')\n",
    "plt.title('Recall Curve',fontsize=18)\n",
    "plt.ylabel('Recall',fontsize=18)\n",
    "#plt.grid()\n",
    "plt.xticks(fontsize=14)\n",
    "plt.yticks(fontsize=14)\n",
    "plt.xlabel('Number of epochs',fontsize=18)\n",
    "plt.legend(['train', 'test'], loc='upper right',fontsize=18)\n",
    "#plt.axis([0,1000,0,500])\n",
    "plt.show()\n",
    "#plt.savefig('RecallCurve_2class_1dcnn.eps', format='eps')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Precision"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#---Summarize history for Precision\n",
    "plt.figure(figsize=(20,5))\n",
    "plt.plot(history.history['recall_m'],'-o')\n",
    "plt.plot(history.history['val_recall_m'],'-s')\n",
    "plt.title('Precision Curve',fontsize=18)\n",
    "plt.ylabel('Precision',fontsize=18)\n",
    "#plt.grid()\n",
    "plt.xticks(fontsize=14)\n",
    "plt.yticks(fontsize=14)\n",
    "plt.xlabel('Number of epochs',fontsize=18)\n",
    "plt.legend(['train', 'test'], loc='lower right',fontsize=18)\n",
    "#plt.axis([0,1000,0,500])\n",
    "plt.show()\n",
    "#plt.savefig('PrecisionCurve_2class_1dcnn.eps', format='eps')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Network's values"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Output of final layer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# trained model weights\n",
    "layer_outputs = model.layers[6].output\n",
    "print(layer_outputs.shape)\n",
    "print(layer_outputs)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Weights and biases "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "layer_weights = model.layers[0].get_weights()[0]\n",
    "print(layer_weights.shape)\n",
    "print(layer_weights)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "layer_biases  = model.layers[0].get_weights()[1]\n",
    "print(layer_biases.shape)\n",
    "print(layer_biases)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Predictions"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Predict"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 70
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 55878,
     "status": "ok",
     "timestamp": 1571627063846,
     "user": {
      "displayName": "Mahindra Singh Rautela",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mBTSnpeHr1j42u0w7ABMmG3BYbWyCtFQVbOvr6sAA=s64",
      "userId": "15859880813264051870"
     },
     "user_tz": -330
    },
    "id": "o5FiLL2UBArA",
    "outputId": "bad0a6e8-f8cf-489f-fea2-3b14a28df061"
   },
   "outputs": [],
   "source": [
    "y_predicted = model.predict(X_valid, verbose = 2)\n",
    "y_actual = y_valid"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[1.9742695e-04]\n",
      " [9.4511372e-01]\n",
      " [3.0578154e-03]\n",
      " ...\n",
      " [9.2609393e-05]\n",
      " [9.9991024e-01]\n",
      " [9.9977952e-01]]\n"
     ]
    }
   ],
   "source": [
    "print(y_predicted)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [],
   "source": [
    "# rounding the predictions\n",
    "y_pred = np.round(y_predicted)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "BYXCrOVrBJe2"
   },
   "source": [
    "### Confusion Matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[884   6]\n",
      " [ 11 905]]\n"
     ]
    }
   ],
   "source": [
    "cm = confusion_matrix(y_actual, y_pred)\n",
    "print(cm)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 55655,
     "status": "ok",
     "timestamp": 1571627064529,
     "user": {
      "displayName": "Mahindra Singh Rautela",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mBTSnpeHr1j42u0w7ABMmG3BYbWyCtFQVbOvr6sAA=s64",
      "userId": "15859880813264051870"
     },
     "user_tz": -330
    },
    "id": "jlfuW7lBBBcf",
    "outputId": "a000e66f-3457-4b95-f249-57037ce07bb2"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Figure size 432x288 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# This is a function to plot confusion matrix --> Do not disturb this\n",
    "# Confusion matrix gives a representation of True positive, True Negatives, False Positives and False Negatives\n",
    "from sklearn.utils.multiclass import unique_labels\n",
    "import itertools\n",
    "import matplotlib.pyplot as plt\n",
    "from keras.utils import to_categorical\n",
    "\n",
    "fig = plt.gcf()\n",
    "\n",
    "def plot_confusion_matrix(cm, classes,\n",
    "                          normalize=False,\n",
    "                          title=None,\n",
    "                          cmap=plt.cm.Blues):\n",
    "    \"\"\"\n",
    "    This function prints and plots the confusion matrix.\n",
    "    Normalization can be applied by setting `normalize=True`.\n",
    "    \"\"\"\n",
    "    if not title:\n",
    "        if normalize:\n",
    "            title = 'Normalized confusion matrix'\n",
    "        else:\n",
    "            title = 'Confusion matrix, without normalization'\n",
    "\n",
    "    plt.imshow(cm, interpolation = 'nearest', cmap = cmap)\n",
    "    plt.title(title)\n",
    "    plt.colorbar()\n",
    "    tick_marks = np.arange(len(classes))\n",
    "    plt.xticks(tick_marks, classes, rotation = 45)\n",
    "    plt.yticks(tick_marks, classes)\n",
    "    \n",
    "    if normalize:\n",
    "        cm = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis]\n",
    "        cm = np.round(cm, 2)\n",
    "        print(\"Normalized confusion matrix\")\n",
    "    else:\n",
    "        print('Confusion matrix, without normalization')\n",
    "\n",
    "    print(cm)\n",
    "\n",
    "    thresh = cm.max() / 2.\n",
    "    for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):\n",
    "      plt.text(j, i, cm[i,j],\n",
    "      horizontalalignment = 'center',\n",
    "      color = \"white\" if cm[i,j] > thresh else \"black\")\n",
    "\n",
    "    fig.tight_layout()\n",
    "    plt.ylabel('True Label')\n",
    "    plt.xlabel('Predicted label')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 387
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 54656,
     "status": "ok",
     "timestamp": 1571627064533,
     "user": {
      "displayName": "Mahindra Singh Rautela",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mBTSnpeHr1j42u0w7ABMmG3BYbWyCtFQVbOvr6sAA=s64",
      "userId": "15859880813264051870"
     },
     "user_tz": -330
    },
    "id": "VMJrmxVcBE1r",
    "outputId": "200c7578-b144-4494-a020-99bebeae9312"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Confusion matrix, without normalization\n",
      "[[884   6]\n",
      " [ 11 905]]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot Non-Normalized confusion matrix\n",
    "cm_labels = ['Undamaged','Damaged']\n",
    "plot_confusion_matrix(cm, classes=cm_labels, title='Confusion matrix')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 387
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 53311,
     "status": "ok",
     "timestamp": 1571627064536,
     "user": {
      "displayName": "Mahindra Singh Rautela",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mBTSnpeHr1j42u0w7ABMmG3BYbWyCtFQVbOvr6sAA=s64",
      "userId": "15859880813264051870"
     },
     "user_tz": -330
    },
    "id": "SQe4uuaxBHMw",
    "outputId": "d6b51bcb-bbfd-4379-a126-af41b8ba7064"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Normalized confusion matrix\n",
      "[[0.99 0.01]\n",
      " [0.01 0.99]]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot Normalized confusion matrix\n",
    "cm_labels = ['Undamaged','Damaged']\n",
    "plot_confusion_matrix(cm, classes=cm_labels, normalize = True, title='Confusion matrix')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Classification Report"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 194
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 51946,
     "status": "ok",
     "timestamp": 1571627064538,
     "user": {
      "displayName": "Mahindra Singh Rautela",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mBTSnpeHr1j42u0w7ABMmG3BYbWyCtFQVbOvr6sAA=s64",
      "userId": "15859880813264051870"
     },
     "user_tz": -330
    },
    "id": "njgwXFFTHljJ",
    "outputId": "0dd938f2-872c-4b4d-94b9-c3d3950f0dd3"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Classification Report\n",
      "                 precision    recall  f1-score   support\n",
      "\n",
      "Case1_undamaged       0.99      0.99      0.99       890\n",
      "  Case2_damaged       0.99      0.99      0.99       916\n",
      "\n",
      "       accuracy                           0.99      1806\n",
      "      macro avg       0.99      0.99      0.99      1806\n",
      "   weighted avg       0.99      0.99      0.99      1806\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Classification Report\n",
    "print('Classification Report')\n",
    "target_names = ['Case1_undamaged', 'Case2_damaged']\n",
    "print(classification_report(y_actual, y_pred, target_names = target_names))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### AUC - ROC curve "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# It is a performance measurement for classification problem at various thresholds settings. \n",
    "# ROC is a probability curve and AUC represents degree or measure of separability. \n",
    "# AUC stands for \"Area under the ROC Curve.\" \n",
    "# AUC measures the entire two-dimensional area underneath the entire ROC curve.\n",
    "# The ROC curve is plotted with TPR against the FPR where TPR is on y-axis and FPR is on the x-axis."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import roc_curve\n",
    "fpr, tpr, thresholds = roc_curve(y_actual, y_pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import auc\n",
    "auc_value = auc(fpr, tpr)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(8,8))\n",
    "plt.plot([0, 1], [0, 1], 'k--')\n",
    "plt.plot(fpr, tpr,'-o', label='AUC (area = {:.3f})'.format(auc_value))\n",
    "plt.xlabel('False positive rate',fontsize=18)\n",
    "plt.ylabel('True positive rate',fontsize=18)\n",
    "plt.title('ROC curve',fontsize=18)\n",
    "plt.legend(loc='best',fontsize=18)\n",
    "plt.xticks(fontsize=18)\n",
    "plt.yticks(fontsize=18)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "collapsed_sections": [],
   "machine_shape": "hm",
   "name": "1DCNN_OGW_Binaryclassification.ipynb",
   "provenance": [],
   "toc_visible": true
  },
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}