CNNLSTM-tunning.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "%load_ext autoreload\n",
    "%autoreload 2\n",
    "%config InlineBackend.figure_format ='retina'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import tensorflow as tf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "gpus = tf.config.experimental.list_physical_devices('GPU')\n",
    "if gpus:\n",
    "    gpu_num = 1\n",
    "    try:\n",
    "        tf.config.experimental.set_visible_devices(gpus[gpu_num], 'GPU')\n",
    "        #tf.config.experimental.set_memory_growth(gpus[gpu_num], True)\n",
    "        tf.config.experimental.set_virtual_device_configuration(\n",
    "            gpus[gpu_num],\n",
    "            [tf.config.experimental.VirtualDeviceConfiguration(memory_limit=4000)])\n",
    "    except RuntimeError as e:\n",
    "        print(e)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "default_HP = {\n",
    "    'CNN_1_kernal' : 16,\n",
    "    'CNN_1_RNN' : 64,\n",
    "    'CNN_2_kernal' : 32,\n",
    "    'CNN_2_RNN' : 128,\n",
    "    'CNN_3_kernal' : 64,\n",
    "    'CNN_3_RNN' : 256,\n",
    "    'CNN_4_kernal' : 128,\n",
    "    'CNN_4_RNN' : 256,\n",
    "    'CNN_5_kernal' : 128,\n",
    "    'CNN_5_RNN' : 256,\n",
    "\n",
    "    'DNN_1' : 100,\n",
    "    'DNN_2' : 100,\n",
    "    'DNN_rate' : 100,\n",
    "\n",
    "    'CNN_dropout' : 0.5,\n",
    "    'DNN_dropout' : 0.5,\n",
    "\n",
    "    'learning_rate' : 0.0001\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "from Model import *"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "CNNLSTM_SpCas9_model = CNNLSTM_SpCas9(HP=default_HP, random_seed=1234)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model\"\n",
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "input_1 (InputLayer)            [(None, 30, 4)]      0                                            \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_15 (Conv1D)              (None, 30, 128)      640         input_1[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_16 (Conv1D)              (None, 30, 128)      1152        input_1[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_17 (Conv1D)              (None, 30, 128)      1664        input_1[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_18 (Conv1D)              (None, 30, 128)      2176        input_1[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_19 (Conv1D)              (None, 30, 128)      2688        input_1[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_20 (Conv1D)              (None, 30, 128)      640         input_1[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_21 (Conv1D)              (None, 30, 128)      1152        input_1[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_22 (Conv1D)              (None, 30, 128)      1664        input_1[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_23 (Conv1D)              (None, 30, 128)      2176        input_1[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_24 (Conv1D)              (None, 30, 128)      2688        input_1[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "dropout_18 (Dropout)            (None, 30, 128)      0           conv1d_15[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dropout_19 (Dropout)            (None, 30, 128)      0           conv1d_16[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dropout_20 (Dropout)            (None, 30, 128)      0           conv1d_17[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dropout_21 (Dropout)            (None, 30, 128)      0           conv1d_18[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dropout_22 (Dropout)            (None, 30, 128)      0           conv1d_19[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dropout_24 (Dropout)            (None, 30, 128)      0           conv1d_20[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dropout_25 (Dropout)            (None, 30, 128)      0           conv1d_21[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dropout_26 (Dropout)            (None, 30, 128)      0           conv1d_22[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dropout_27 (Dropout)            (None, 30, 128)      0           conv1d_23[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dropout_28 (Dropout)            (None, 30, 128)      0           conv1d_24[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_5 (Conv1D)               (None, 30, 32)       160         input_1[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_6 (Conv1D)               (None, 30, 32)       288         input_1[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_7 (Conv1D)               (None, 30, 32)       416         input_1[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_8 (Conv1D)               (None, 30, 32)       544         input_1[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_9 (Conv1D)               (None, 30, 32)       672         input_1[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_10 (Conv1D)              (None, 30, 64)       320         input_1[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_11 (Conv1D)              (None, 30, 64)       576         input_1[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_12 (Conv1D)              (None, 30, 64)       832         input_1[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_13 (Conv1D)              (None, 30, 64)       1088        input_1[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_14 (Conv1D)              (None, 30, 64)       1344        input_1[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_6 (Concatenate)     (None, 30, 640)      0           dropout_18[0][0]                 \n",
      "                                                                 dropout_19[0][0]                 \n",
      "                                                                 dropout_20[0][0]                 \n",
      "                                                                 dropout_21[0][0]                 \n",
      "                                                                 dropout_22[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_8 (Concatenate)     (None, 30, 640)      0           dropout_24[0][0]                 \n",
      "                                                                 dropout_25[0][0]                 \n",
      "                                                                 dropout_26[0][0]                 \n",
      "                                                                 dropout_27[0][0]                 \n",
      "                                                                 dropout_28[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "dropout_6 (Dropout)             (None, 30, 32)       0           conv1d_5[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "dropout_7 (Dropout)             (None, 30, 32)       0           conv1d_6[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "dropout_8 (Dropout)             (None, 30, 32)       0           conv1d_7[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "dropout_9 (Dropout)             (None, 30, 32)       0           conv1d_8[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "dropout_10 (Dropout)            (None, 30, 32)       0           conv1d_9[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "dropout_12 (Dropout)            (None, 30, 64)       0           conv1d_10[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dropout_13 (Dropout)            (None, 30, 64)       0           conv1d_11[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dropout_14 (Dropout)            (None, 30, 64)       0           conv1d_12[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dropout_15 (Dropout)            (None, 30, 64)       0           conv1d_13[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dropout_16 (Dropout)            (None, 30, 64)       0           conv1d_14[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lstm_3 (LSTM)                   (None, 30, 256)      918528      concatenate_6[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "lstm_4 (LSTM)                   (None, 30, 256)      918528      concatenate_8[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_2 (Concatenate)     (None, 30, 160)      0           dropout_6[0][0]                  \n",
      "                                                                 dropout_7[0][0]                  \n",
      "                                                                 dropout_8[0][0]                  \n",
      "                                                                 dropout_9[0][0]                  \n",
      "                                                                 dropout_10[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_4 (Concatenate)     (None, 30, 320)      0           dropout_12[0][0]                 \n",
      "                                                                 dropout_13[0][0]                 \n",
      "                                                                 dropout_14[0][0]                 \n",
      "                                                                 dropout_15[0][0]                 \n",
      "                                                                 dropout_16[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "dropout_23 (Dropout)            (None, 30, 256)      0           lstm_3[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "dropout_29 (Dropout)            (None, 30, 256)      0           lstm_4[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "lstm_1 (LSTM)                   (None, 30, 128)      147968      concatenate_2[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "lstm_2 (LSTM)                   (None, 30, 256)      590848      concatenate_4[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_9 (Concatenate)     (None, 30, 512)      0           dropout_23[0][0]                 \n",
      "                                                                 dropout_29[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "dropout_11 (Dropout)            (None, 30, 128)      0           lstm_1[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "dropout_17 (Dropout)            (None, 30, 256)      0           lstm_2[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "lambda_2 (Lambda)               (None, 512)          0           concatenate_9[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_5 (Concatenate)     (None, 30, 384)      0           dropout_11[0][0]                 \n",
      "                                                                 dropout_17[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_7 (Concatenate)     (None, 30, 512)      0           dropout_17[0][0]                 \n",
      "                                                                 dropout_23[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "dense_4 (Dense)                 (None, 100)          51300       lambda_2[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "lambda (Lambda)                 (None, 384)          0           concatenate_5[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "lambda_1 (Lambda)               (None, 512)          0           concatenate_7[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "dropout_34 (Dropout)            (None, 100)          0           dense_4[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "dense (Dense)                   (None, 100)          38500       lambda[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "dense_2 (Dense)                 (None, 100)          51300       lambda_1[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "dense_5 (Dense)                 (None, 100)          10100       dropout_34[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "dropout_30 (Dropout)            (None, 100)          0           dense[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "dropout_32 (Dropout)            (None, 100)          0           dense_2[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "dropout_35 (Dropout)            (None, 100)          0           dense_5[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "dense_1 (Dense)                 (None, 100)          10100       dropout_30[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "dense_3 (Dense)                 (None, 100)          10100       dropout_32[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "dense_6 (Dense)                 (None, 100)          10100       dropout_35[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "dropout_31 (Dropout)            (None, 100)          0           dense_1[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "dropout_33 (Dropout)            (None, 100)          0           dense_3[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "dropout_36 (Dropout)            (None, 100)          0           dense_6[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "class_1 (Dense)                 (None, 11)           1111        dropout_31[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "class_2 (Dense)                 (None, 11)           1111        dropout_33[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "class_final (Dense)             (None, 11)           1111        dropout_35[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "rate (Dense)                    (None, 1)            101         dropout_36[0][0]                 \n",
      "==================================================================================================\n",
      "Total params: 2,783,686\n",
      "Trainable params: 2,783,686\n",
      "Non-trainable params: 0\n",
      "__________________________________________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "CNNLSTM_SpCas9_model.CNNLSTM_MTL_model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 11548 samples, validate on 1284 samples\n",
      "Epoch 1/200\n",
      "11548/11548 [==============================] - 15s 1ms/sample - loss: 6.0570 - class_1_loss: 2.3590 - class_2_loss: 2.3575 - class_final_loss: 2.3898 - rate_loss: 0.1454 - class_1_accuracy: 0.1182 - class_2_accuracy: 0.1216 - class_final_accuracy: 0.1137 - val_loss: 5.8695 - val_class_1_loss: 2.2782 - val_class_2_loss: 2.2909 - val_class_final_loss: 2.3768 - val_rate_loss: 0.1226 - val_class_1_accuracy: 0.1355 - val_class_2_accuracy: 0.1340 - val_class_final_accuracy: 0.1192\n",
      "Epoch 2/200\n",
      "11548/11548 [==============================] - 8s 673us/sample - loss: 5.8717 - class_1_loss: 2.2985 - class_2_loss: 2.3057 - class_final_loss: 2.3533 - rate_loss: 0.0907 - class_1_accuracy: 0.1286 - class_2_accuracy: 0.1191 - class_final_accuracy: 0.1205 - val_loss: 5.7620 - val_class_1_loss: 2.2532 - val_class_2_loss: 2.2518 - val_class_final_loss: 2.3564 - val_rate_loss: 0.0883 - val_class_1_accuracy: 0.1363 - val_class_2_accuracy: 0.1449 - val_class_final_accuracy: 0.1394\n",
      "Epoch 3/200\n",
      "11548/11548 [==============================] - 8s 675us/sample - loss: 5.8008 - class_1_loss: 2.2778 - class_2_loss: 2.2788 - class_final_loss: 2.3293 - rate_loss: 0.0795 - class_1_accuracy: 0.1291 - class_2_accuracy: 0.1300 - class_final_accuracy: 0.1285 - val_loss: 5.7081 - val_class_1_loss: 2.2391 - val_class_2_loss: 2.2362 - val_class_final_loss: 2.3372 - val_rate_loss: 0.0748 - val_class_1_accuracy: 0.1316 - val_class_2_accuracy: 0.1542 - val_class_final_accuracy: 0.1340\n",
      "Epoch 4/200\n",
      "11548/11548 [==============================] - 8s 695us/sample - loss: 5.7430 - class_1_loss: 2.2548 - class_2_loss: 2.2577 - class_final_loss: 2.3099 - rate_loss: 0.0759 - class_1_accuracy: 0.1332 - class_2_accuracy: 0.1375 - class_final_accuracy: 0.1307 - val_loss: 5.6232 - val_class_1_loss: 2.2155 - val_class_2_loss: 2.1995 - val_class_final_loss: 2.3082 - val_rate_loss: 0.0657 - val_class_1_accuracy: 0.1651 - val_class_2_accuracy: 0.1573 - val_class_final_accuracy: 0.1394\n",
      "Epoch 5/200\n",
      "11548/11548 [==============================] - 8s 708us/sample - loss: 5.6772 - class_1_loss: 2.2316 - class_2_loss: 2.2275 - class_final_loss: 2.2871 - rate_loss: 0.0741 - class_1_accuracy: 0.1443 - class_2_accuracy: 0.1513 - class_final_accuracy: 0.1345 - val_loss: 5.5544 - val_class_1_loss: 2.1805 - val_class_2_loss: 2.1777 - val_class_final_loss: 2.2852 - val_rate_loss: 0.0649 - val_class_1_accuracy: 0.1799 - val_class_2_accuracy: 0.1597 - val_class_final_accuracy: 0.1332\n",
      "Epoch 6/200\n",
      "11548/11548 [==============================] - 8s 698us/sample - loss: 5.6160 - class_1_loss: 2.2084 - class_2_loss: 2.2002 - class_final_loss: 2.2728 - rate_loss: 0.0716 - class_1_accuracy: 0.1492 - class_2_accuracy: 0.1505 - class_final_accuracy: 0.1363 - val_loss: 5.4830 - val_class_1_loss: 2.1515 - val_class_2_loss: 2.1485 - val_class_final_loss: 2.2636 - val_rate_loss: 0.0660 - val_class_1_accuracy: 0.1667 - val_class_2_accuracy: 0.1776 - val_class_final_accuracy: 0.1363\n",
      "Epoch 7/200\n",
      "11548/11548 [==============================] - 8s 699us/sample - loss: 5.5519 - class_1_loss: 2.1814 - class_2_loss: 2.1760 - class_final_loss: 2.2494 - rate_loss: 0.0694 - class_1_accuracy: 0.1646 - class_2_accuracy: 0.1645 - class_final_accuracy: 0.1381 - val_loss: 5.4195 - val_class_1_loss: 2.1283 - val_class_2_loss: 2.1278 - val_class_final_loss: 2.2312 - val_rate_loss: 0.0659 - val_class_1_accuracy: 0.1791 - val_class_2_accuracy: 0.1706 - val_class_final_accuracy: 0.1620\n",
      "Epoch 8/200\n",
      "11548/11548 [==============================] - 8s 700us/sample - loss: 5.4994 - class_1_loss: 2.1644 - class_2_loss: 2.1506 - class_final_loss: 2.2315 - rate_loss: 0.0682 - class_1_accuracy: 0.1662 - class_2_accuracy: 0.1777 - class_final_accuracy: 0.1483 - val_loss: 5.3893 - val_class_1_loss: 2.1153 - val_class_2_loss: 2.1144 - val_class_final_loss: 2.2131 - val_rate_loss: 0.0738 - val_class_1_accuracy: 0.1776 - val_class_2_accuracy: 0.1893 - val_class_final_accuracy: 0.1729\n",
      "Epoch 9/200\n",
      "11548/11548 [==============================] - 8s 699us/sample - loss: 5.4604 - class_1_loss: 2.1495 - class_2_loss: 2.1409 - class_final_loss: 2.2091 - rate_loss: 0.0645 - class_1_accuracy: 0.1720 - class_2_accuracy: 0.1795 - class_final_accuracy: 0.1574 - val_loss: 5.3329 - val_class_1_loss: 2.1026 - val_class_2_loss: 2.0986 - val_class_final_loss: 2.1795 - val_rate_loss: 0.0663 - val_class_1_accuracy: 0.1885 - val_class_2_accuracy: 0.2009 - val_class_final_accuracy: 0.1768\n",
      "Epoch 10/200\n",
      "11548/11548 [==============================] - 8s 699us/sample - loss: 5.4218 - class_1_loss: 2.1401 - class_2_loss: 2.1260 - class_final_loss: 2.1833 - rate_loss: 0.0634 - class_1_accuracy: 0.1762 - class_2_accuracy: 0.1765 - class_final_accuracy: 0.1675 - val_loss: 5.3520 - val_class_1_loss: 2.1079 - val_class_2_loss: 2.1146 - val_class_final_loss: 2.1547 - val_rate_loss: 0.0764 - val_class_1_accuracy: 0.1893 - val_class_2_accuracy: 0.1900 - val_class_final_accuracy: 0.1791\n",
      "Epoch 11/200\n",
      "11548/11548 [==============================] - 8s 698us/sample - loss: 5.3879 - class_1_loss: 2.1292 - class_2_loss: 2.1127 - class_final_loss: 2.1661 - rate_loss: 0.0621 - class_1_accuracy: 0.1799 - class_2_accuracy: 0.1831 - class_final_accuracy: 0.1706 - val_loss: 5.2797 - val_class_1_loss: 2.0838 - val_class_2_loss: 2.0864 - val_class_final_loss: 2.1400 - val_rate_loss: 0.0693 - val_class_1_accuracy: 0.2025 - val_class_2_accuracy: 0.1893 - val_class_final_accuracy: 0.1815\n",
      "Epoch 12/200\n",
      "11548/11548 [==============================] - 8s 693us/sample - loss: 5.3422 - class_1_loss: 2.1170 - class_2_loss: 2.0914 - class_final_loss: 2.1481 - rate_loss: 0.0601 - class_1_accuracy: 0.1791 - class_2_accuracy: 0.1940 - class_final_accuracy: 0.1754 - val_loss: 5.2440 - val_class_1_loss: 2.0731 - val_class_2_loss: 2.0668 - val_class_final_loss: 2.1152 - val_rate_loss: 0.0753 - val_class_1_accuracy: 0.2157 - val_class_2_accuracy: 0.2017 - val_class_final_accuracy: 0.1869\n",
      "Epoch 13/200\n",
      "11548/11548 [==============================] - 8s 702us/sample - loss: 5.3137 - class_1_loss: 2.1005 - class_2_loss: 2.0898 - class_final_loss: 2.1323 - rate_loss: 0.0571 - class_1_accuracy: 0.1860 - class_2_accuracy: 0.1929 - class_final_accuracy: 0.1827 - val_loss: 5.1734 - val_class_1_loss: 2.0483 - val_class_2_loss: 2.0407 - val_class_final_loss: 2.1020 - val_rate_loss: 0.0682 - val_class_1_accuracy: 0.2087 - val_class_2_accuracy: 0.2165 - val_class_final_accuracy: 0.1869\n",
      "Epoch 14/200\n",
      "11548/11548 [==============================] - 8s 699us/sample - loss: 5.2834 - class_1_loss: 2.0922 - class_2_loss: 2.0740 - class_final_loss: 2.1210 - rate_loss: 0.0563 - class_1_accuracy: 0.1859 - class_2_accuracy: 0.1950 - class_final_accuracy: 0.1786 - val_loss: 5.1600 - val_class_1_loss: 2.0455 - val_class_2_loss: 2.0477 - val_class_final_loss: 2.0814 - val_rate_loss: 0.0600 - val_class_1_accuracy: 0.2095 - val_class_2_accuracy: 0.2056 - val_class_final_accuracy: 0.1963\n",
      "Epoch 15/200\n",
      "11548/11548 [==============================] - 8s 701us/sample - loss: 5.2557 - class_1_loss: 2.0828 - class_2_loss: 2.0657 - class_final_loss: 2.1069 - rate_loss: 0.0541 - class_1_accuracy: 0.1941 - class_2_accuracy: 0.1960 - class_final_accuracy: 0.1917 - val_loss: 5.1363 - val_class_1_loss: 2.0305 - val_class_2_loss: 2.0274 - val_class_final_loss: 2.0842 - val_rate_loss: 0.0671 - val_class_1_accuracy: 0.2064 - val_class_2_accuracy: 0.2188 - val_class_final_accuracy: 0.1869\n",
      "Epoch 16/200\n",
      "11548/11548 [==============================] - 8s 699us/sample - loss: 5.2154 - class_1_loss: 2.0616 - class_2_loss: 2.0544 - class_final_loss: 2.0942 - rate_loss: 0.0516 - class_1_accuracy: 0.2000 - class_2_accuracy: 0.2060 - class_final_accuracy: 0.1871 - val_loss: 5.0939 - val_class_1_loss: 2.0236 - val_class_2_loss: 2.0218 - val_class_final_loss: 2.0625 - val_rate_loss: 0.0577 - val_class_1_accuracy: 0.2095 - val_class_2_accuracy: 0.2157 - val_class_final_accuracy: 0.2048\n",
      "Epoch 17/200\n",
      "11548/11548 [==============================] - 8s 695us/sample - loss: 5.2056 - class_1_loss: 2.0655 - class_2_loss: 2.0474 - class_final_loss: 2.0794 - rate_loss: 0.0512 - class_1_accuracy: 0.1984 - class_2_accuracy: 0.2086 - class_final_accuracy: 0.1927 - val_loss: 5.0914 - val_class_1_loss: 2.0244 - val_class_2_loss: 2.0262 - val_class_final_loss: 2.0494 - val_rate_loss: 0.0533 - val_class_1_accuracy: 0.2087 - val_class_2_accuracy: 0.2118 - val_class_final_accuracy: 0.1978\n",
      "Epoch 18/200\n",
      "11548/11548 [==============================] - 8s 697us/sample - loss: 5.1680 - class_1_loss: 2.0482 - class_2_loss: 2.0326 - class_final_loss: 2.0742 - rate_loss: 0.0492 - class_1_accuracy: 0.2030 - class_2_accuracy: 0.2116 - class_final_accuracy: 0.1950 - val_loss: 5.0332 - val_class_1_loss: 2.0005 - val_class_2_loss: 1.9918 - val_class_final_loss: 2.0463 - val_rate_loss: 0.0593 - val_class_1_accuracy: 0.2181 - val_class_2_accuracy: 0.2134 - val_class_final_accuracy: 0.1994\n",
      "Epoch 19/200\n",
      "11548/11548 [==============================] - 8s 692us/sample - loss: 5.1534 - class_1_loss: 2.0436 - class_2_loss: 2.0284 - class_final_loss: 2.0665 - rate_loss: 0.0482 - class_1_accuracy: 0.2059 - class_2_accuracy: 0.2034 - class_final_accuracy: 0.1977 - val_loss: 5.0119 - val_class_1_loss: 1.9918 - val_class_2_loss: 1.9891 - val_class_final_loss: 2.0415 - val_rate_loss: 0.0482 - val_class_1_accuracy: 0.2165 - val_class_2_accuracy: 0.2126 - val_class_final_accuracy: 0.2079\n",
      "Epoch 20/200\n",
      "11548/11548 [==============================] - 8s 692us/sample - loss: 5.1262 - class_1_loss: 2.0345 - class_2_loss: 2.0167 - class_final_loss: 2.0553 - rate_loss: 0.0473 - class_1_accuracy: 0.2035 - class_2_accuracy: 0.2125 - class_final_accuracy: 0.2018 - val_loss: 5.0366 - val_class_1_loss: 2.0144 - val_class_2_loss: 1.9998 - val_class_final_loss: 2.0227 - val_rate_loss: 0.0504 - val_class_1_accuracy: 0.2111 - val_class_2_accuracy: 0.2220 - val_class_final_accuracy: 0.2134\n",
      "Epoch 21/200\n",
      "11548/11548 [==============================] - 8s 706us/sample - loss: 5.0982 - class_1_loss: 2.0229 - class_2_loss: 2.0051 - class_final_loss: 2.0477 - rate_loss: 0.0458 - class_1_accuracy: 0.2154 - class_2_accuracy: 0.2162 - class_final_accuracy: 0.1993 - val_loss: 4.9781 - val_class_1_loss: 1.9897 - val_class_2_loss: 1.9733 - val_class_final_loss: 2.0171 - val_rate_loss: 0.0497 - val_class_1_accuracy: 0.2227 - val_class_2_accuracy: 0.2220 - val_class_final_accuracy: 0.2040\n",
      "Epoch 22/200\n",
      "11548/11548 [==============================] - 8s 698us/sample - loss: 5.0626 - class_1_loss: 2.0044 - class_2_loss: 1.9934 - class_final_loss: 2.0408 - rate_loss: 0.0445 - class_1_accuracy: 0.2129 - class_2_accuracy: 0.2155 - class_final_accuracy: 0.2014 - val_loss: 4.9377 - val_class_1_loss: 1.9732 - val_class_2_loss: 1.9587 - val_class_final_loss: 2.0046 - val_rate_loss: 0.0464 - val_class_1_accuracy: 0.2274 - val_class_2_accuracy: 0.2251 - val_class_final_accuracy: 0.2142\n",
      "Epoch 23/200\n",
      "11548/11548 [==============================] - 8s 701us/sample - loss: 5.0640 - class_1_loss: 2.0088 - class_2_loss: 1.9976 - class_final_loss: 2.0301 - rate_loss: 0.0433 - class_1_accuracy: 0.2143 - class_2_accuracy: 0.2145 - class_final_accuracy: 0.2100 - val_loss: 4.9132 - val_class_1_loss: 1.9629 - val_class_2_loss: 1.9531 - val_class_final_loss: 1.9985 - val_rate_loss: 0.0455 - val_class_1_accuracy: 0.2220 - val_class_2_accuracy: 0.2212 - val_class_final_accuracy: 0.2204\n",
      "Epoch 24/200\n",
      "11548/11548 [==============================] - 8s 701us/sample - loss: 5.0274 - class_1_loss: 1.9955 - class_2_loss: 1.9811 - class_final_loss: 2.0168 - rate_loss: 0.0428 - class_1_accuracy: 0.2200 - class_2_accuracy: 0.2205 - class_final_accuracy: 0.2135 - val_loss: 4.9441 - val_class_1_loss: 1.9786 - val_class_2_loss: 1.9556 - val_class_final_loss: 2.0011 - val_rate_loss: 0.0476 - val_class_1_accuracy: 0.2064 - val_class_2_accuracy: 0.2266 - val_class_final_accuracy: 0.2188\n",
      "Epoch 25/200\n",
      "11548/11548 [==============================] - 8s 702us/sample - loss: 5.0096 - class_1_loss: 1.9900 - class_2_loss: 1.9701 - class_final_loss: 2.0163 - rate_loss: 0.0420 - class_1_accuracy: 0.2222 - class_2_accuracy: 0.2191 - class_final_accuracy: 0.2135 - val_loss: 4.8857 - val_class_1_loss: 1.9508 - val_class_2_loss: 1.9380 - val_class_final_loss: 1.9805 - val_rate_loss: 0.0439 - val_class_1_accuracy: 0.2266 - val_class_2_accuracy: 0.2352 - val_class_final_accuracy: 0.2305\n",
      "Epoch 26/200\n",
      "11548/11548 [==============================] - 8s 702us/sample - loss: 4.9971 - class_1_loss: 1.9849 - class_2_loss: 1.9653 - class_final_loss: 2.0108 - rate_loss: 0.0413 - class_1_accuracy: 0.2240 - class_2_accuracy: 0.2201 - class_final_accuracy: 0.2182 - val_loss: 4.8411 - val_class_1_loss: 1.9335 - val_class_2_loss: 1.9199 - val_class_final_loss: 1.9675 - val_rate_loss: 0.0444 - val_class_1_accuracy: 0.2227 - val_class_2_accuracy: 0.2414 - val_class_final_accuracy: 0.2298\n",
      "Epoch 27/200\n",
      "11548/11548 [==============================] - 8s 703us/sample - loss: 4.9690 - class_1_loss: 1.9751 - class_2_loss: 1.9529 - class_final_loss: 2.0018 - rate_loss: 0.0408 - class_1_accuracy: 0.2265 - class_2_accuracy: 0.2294 - class_final_accuracy: 0.2144 - val_loss: 4.8557 - val_class_1_loss: 1.9422 - val_class_2_loss: 1.9274 - val_class_final_loss: 1.9689 - val_rate_loss: 0.0453 - val_class_1_accuracy: 0.2235 - val_class_2_accuracy: 0.2445 - val_class_final_accuracy: 0.2344\n",
      "Epoch 28/200\n",
      "11548/11548 [==============================] - 8s 703us/sample - loss: 4.9589 - class_1_loss: 1.9706 - class_2_loss: 1.9457 - class_final_loss: 2.0028 - rate_loss: 0.0397 - class_1_accuracy: 0.2255 - class_2_accuracy: 0.2321 - class_final_accuracy: 0.2170 - val_loss: 4.8850 - val_class_1_loss: 1.9645 - val_class_2_loss: 1.9432 - val_class_final_loss: 1.9649 - val_rate_loss: 0.0394 - val_class_1_accuracy: 0.2399 - val_class_2_accuracy: 0.2399 - val_class_final_accuracy: 0.2313\n",
      "Epoch 29/200\n",
      "11548/11548 [==============================] - 8s 705us/sample - loss: 4.9486 - class_1_loss: 1.9686 - class_2_loss: 1.9476 - class_final_loss: 1.9876 - rate_loss: 0.0392 - class_1_accuracy: 0.2297 - class_2_accuracy: 0.2298 - class_final_accuracy: 0.2206 - val_loss: 4.7916 - val_class_1_loss: 1.9224 - val_class_2_loss: 1.9072 - val_class_final_loss: 1.9473 - val_rate_loss: 0.0396 - val_class_1_accuracy: 0.2368 - val_class_2_accuracy: 0.2477 - val_class_final_accuracy: 0.2352\n",
      "Epoch 30/200\n",
      "11548/11548 [==============================] - 8s 698us/sample - loss: 4.9339 - class_1_loss: 1.9642 - class_2_loss: 1.9433 - class_final_loss: 1.9766 - rate_loss: 0.0383 - class_1_accuracy: 0.2217 - class_2_accuracy: 0.2272 - class_final_accuracy: 0.2205 - val_loss: 4.7983 - val_class_1_loss: 1.9244 - val_class_2_loss: 1.9024 - val_class_final_loss: 1.9339 - val_rate_loss: 0.0408 - val_class_1_accuracy: 0.2321 - val_class_2_accuracy: 0.2399 - val_class_final_accuracy: 0.2461\n",
      "Epoch 31/200\n",
      "11548/11548 [==============================] - 8s 706us/sample - loss: 4.9098 - class_1_loss: 1.9580 - class_2_loss: 1.9282 - class_final_loss: 1.9716 - rate_loss: 0.0378 - class_1_accuracy: 0.2281 - class_2_accuracy: 0.2423 - class_final_accuracy: 0.2216 - val_loss: 4.8060 - val_class_1_loss: 1.9299 - val_class_2_loss: 1.9176 - val_class_final_loss: 1.9437 - val_rate_loss: 0.0366 - val_class_1_accuracy: 0.2227 - val_class_2_accuracy: 0.2344 - val_class_final_accuracy: 0.2329\n",
      "Epoch 32/200\n",
      "11548/11548 [==============================] - 8s 705us/sample - loss: 4.8962 - class_1_loss: 1.9390 - class_2_loss: 1.9353 - class_final_loss: 1.9704 - rate_loss: 0.0370 - class_1_accuracy: 0.2340 - class_2_accuracy: 0.2318 - class_final_accuracy: 0.2247 - val_loss: 4.7655 - val_class_1_loss: 1.9101 - val_class_2_loss: 1.8990 - val_class_final_loss: 1.9324 - val_rate_loss: 0.0371 - val_class_1_accuracy: 0.2321 - val_class_2_accuracy: 0.2445 - val_class_final_accuracy: 0.2329\n",
      "Epoch 33/200\n",
      "11548/11548 [==============================] - 8s 706us/sample - loss: 4.8941 - class_1_loss: 1.9462 - class_2_loss: 1.9277 - class_final_loss: 1.9648 - rate_loss: 0.0369 - class_1_accuracy: 0.2280 - class_2_accuracy: 0.2296 - class_final_accuracy: 0.2233 - val_loss: 4.7587 - val_class_1_loss: 1.9064 - val_class_2_loss: 1.8927 - val_class_final_loss: 1.9263 - val_rate_loss: 0.0366 - val_class_1_accuracy: 0.2477 - val_class_2_accuracy: 0.2407 - val_class_final_accuracy: 0.2453\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 34/200\n",
      "11548/11548 [==============================] - 8s 702us/sample - loss: 4.8727 - class_1_loss: 1.9371 - class_2_loss: 1.9244 - class_final_loss: 1.9531 - rate_loss: 0.0359 - class_1_accuracy: 0.2305 - class_2_accuracy: 0.2335 - class_final_accuracy: 0.2311 - val_loss: 4.7928 - val_class_1_loss: 1.9233 - val_class_2_loss: 1.9163 - val_class_final_loss: 1.9325 - val_rate_loss: 0.0339 - val_class_1_accuracy: 0.2305 - val_class_2_accuracy: 0.2368 - val_class_final_accuracy: 0.2368\n",
      "Epoch 35/200\n",
      "11548/11548 [==============================] - 8s 703us/sample - loss: 4.8552 - class_1_loss: 1.9352 - class_2_loss: 1.9102 - class_final_loss: 1.9509 - rate_loss: 0.0356 - class_1_accuracy: 0.2350 - class_2_accuracy: 0.2389 - class_final_accuracy: 0.2265 - val_loss: 4.7522 - val_class_1_loss: 1.9013 - val_class_2_loss: 1.8938 - val_class_final_loss: 1.9202 - val_rate_loss: 0.0348 - val_class_1_accuracy: 0.2336 - val_class_2_accuracy: 0.2375 - val_class_final_accuracy: 0.2407\n",
      "Epoch 36/200\n",
      "11548/11548 [==============================] - 8s 705us/sample - loss: 4.8530 - class_1_loss: 1.9351 - class_2_loss: 1.9080 - class_final_loss: 1.9493 - rate_loss: 0.0356 - class_1_accuracy: 0.2387 - class_2_accuracy: 0.2426 - class_final_accuracy: 0.2302 - val_loss: 4.7746 - val_class_1_loss: 1.9165 - val_class_2_loss: 1.8961 - val_class_final_loss: 1.9168 - val_rate_loss: 0.0381 - val_class_1_accuracy: 0.2375 - val_class_2_accuracy: 0.2484 - val_class_final_accuracy: 0.2422\n",
      "Epoch 37/200\n",
      "11548/11548 [==============================] - 8s 703us/sample - loss: 4.8328 - class_1_loss: 1.9294 - class_2_loss: 1.9020 - class_final_loss: 1.9317 - rate_loss: 0.0345 - class_1_accuracy: 0.2377 - class_2_accuracy: 0.2437 - class_final_accuracy: 0.2348 - val_loss: 4.7800 - val_class_1_loss: 1.9122 - val_class_2_loss: 1.9062 - val_class_final_loss: 1.9249 - val_rate_loss: 0.0338 - val_class_1_accuracy: 0.2360 - val_class_2_accuracy: 0.2407 - val_class_final_accuracy: 0.2352\n",
      "Epoch 38/200\n",
      "11548/11548 [==============================] - 8s 702us/sample - loss: 4.8216 - class_1_loss: 1.9182 - class_2_loss: 1.8989 - class_final_loss: 1.9398 - rate_loss: 0.0344 - class_1_accuracy: 0.2350 - class_2_accuracy: 0.2384 - class_final_accuracy: 0.2324 - val_loss: 4.7790 - val_class_1_loss: 1.9203 - val_class_2_loss: 1.9098 - val_class_final_loss: 1.9191 - val_rate_loss: 0.0326 - val_class_1_accuracy: 0.2422 - val_class_2_accuracy: 0.2344 - val_class_final_accuracy: 0.2438\n",
      "Epoch 39/200\n",
      "11548/11548 [==============================] - 8s 699us/sample - loss: 4.8229 - class_1_loss: 1.9240 - class_2_loss: 1.8982 - class_final_loss: 1.9303 - rate_loss: 0.0345 - class_1_accuracy: 0.2422 - class_2_accuracy: 0.2481 - class_final_accuracy: 0.2405 - val_loss: 4.7259 - val_class_1_loss: 1.8962 - val_class_2_loss: 1.8783 - val_class_final_loss: 1.8961 - val_rate_loss: 0.0363 - val_class_1_accuracy: 0.2383 - val_class_2_accuracy: 0.2500 - val_class_final_accuracy: 0.2305\n",
      "Epoch 40/200\n",
      "11548/11548 [==============================] - 8s 693us/sample - loss: 4.8087 - class_1_loss: 1.9189 - class_2_loss: 1.8912 - class_final_loss: 1.9282 - rate_loss: 0.0339 - class_1_accuracy: 0.2353 - class_2_accuracy: 0.2438 - class_final_accuracy: 0.2316 - val_loss: 4.7262 - val_class_1_loss: 1.8972 - val_class_2_loss: 1.8827 - val_class_final_loss: 1.8960 - val_rate_loss: 0.0335 - val_class_1_accuracy: 0.2375 - val_class_2_accuracy: 0.2399 - val_class_final_accuracy: 0.2477\n",
      "Epoch 41/200\n",
      "11548/11548 [==============================] - 8s 702us/sample - loss: 4.7921 - class_1_loss: 1.9113 - class_2_loss: 1.8908 - class_final_loss: 1.9149 - rate_loss: 0.0333 - class_1_accuracy: 0.2380 - class_2_accuracy: 0.2465 - class_final_accuracy: 0.2357 - val_loss: 4.8131 - val_class_1_loss: 1.9311 - val_class_2_loss: 1.9160 - val_class_final_loss: 1.9178 - val_rate_loss: 0.0340 - val_class_1_accuracy: 0.2305 - val_class_2_accuracy: 0.2453 - val_class_final_accuracy: 0.2461\n",
      "Epoch 42/200\n",
      "11548/11548 [==============================] - 8s 702us/sample - loss: 4.7833 - class_1_loss: 1.9081 - class_2_loss: 1.8817 - class_final_loss: 1.9165 - rate_loss: 0.0334 - class_1_accuracy: 0.2391 - class_2_accuracy: 0.2484 - class_final_accuracy: 0.2396 - val_loss: 4.7293 - val_class_1_loss: 1.8988 - val_class_2_loss: 1.8817 - val_class_final_loss: 1.8937 - val_rate_loss: 0.0328 - val_class_1_accuracy: 0.2508 - val_class_2_accuracy: 0.2523 - val_class_final_accuracy: 0.2578\n",
      "Epoch 43/200\n",
      "11548/11548 [==============================] - 8s 706us/sample - loss: 4.7909 - class_1_loss: 1.9112 - class_2_loss: 1.8902 - class_final_loss: 1.9120 - rate_loss: 0.0329 - class_1_accuracy: 0.2383 - class_2_accuracy: 0.2471 - class_final_accuracy: 0.2439 - val_loss: 4.7122 - val_class_1_loss: 1.8909 - val_class_2_loss: 1.8753 - val_class_final_loss: 1.8826 - val_rate_loss: 0.0313 - val_class_1_accuracy: 0.2531 - val_class_2_accuracy: 0.2531 - val_class_final_accuracy: 0.2500\n",
      "Epoch 44/200\n",
      "11548/11548 [==============================] - 8s 701us/sample - loss: 4.7739 - class_1_loss: 1.9097 - class_2_loss: 1.8794 - class_final_loss: 1.9043 - rate_loss: 0.0330 - class_1_accuracy: 0.2468 - class_2_accuracy: 0.2500 - class_final_accuracy: 0.2326 - val_loss: 4.7276 - val_class_1_loss: 1.9026 - val_class_2_loss: 1.8841 - val_class_final_loss: 1.8785 - val_rate_loss: 0.0315 - val_class_1_accuracy: 0.2484 - val_class_2_accuracy: 0.2578 - val_class_final_accuracy: 0.2562\n",
      "Epoch 45/200\n",
      "11548/11548 [==============================] - 8s 707us/sample - loss: 4.7638 - class_1_loss: 1.8986 - class_2_loss: 1.8810 - class_final_loss: 1.9033 - rate_loss: 0.0324 - class_1_accuracy: 0.2452 - class_2_accuracy: 0.2477 - class_final_accuracy: 0.2438 - val_loss: 4.6960 - val_class_1_loss: 1.8807 - val_class_2_loss: 1.8721 - val_class_final_loss: 1.8805 - val_rate_loss: 0.0312 - val_class_1_accuracy: 0.2586 - val_class_2_accuracy: 0.2531 - val_class_final_accuracy: 0.2586\n",
      "Epoch 46/200\n",
      "11548/11548 [==============================] - 8s 704us/sample - loss: 4.7398 - class_1_loss: 1.8937 - class_2_loss: 1.8668 - class_final_loss: 1.8945 - rate_loss: 0.0316 - class_1_accuracy: 0.2404 - class_2_accuracy: 0.2539 - class_final_accuracy: 0.2458 - val_loss: 4.6990 - val_class_1_loss: 1.8825 - val_class_2_loss: 1.8695 - val_class_final_loss: 1.8859 - val_rate_loss: 0.0317 - val_class_1_accuracy: 0.2578 - val_class_2_accuracy: 0.2422 - val_class_final_accuracy: 0.2578\n",
      "Epoch 47/200\n",
      "11548/11548 [==============================] - 8s 705us/sample - loss: 4.7370 - class_1_loss: 1.8927 - class_2_loss: 1.8665 - class_final_loss: 1.8932 - rate_loss: 0.0317 - class_1_accuracy: 0.2537 - class_2_accuracy: 0.2574 - class_final_accuracy: 0.2485 - val_loss: 4.6658 - val_class_1_loss: 1.8751 - val_class_2_loss: 1.8579 - val_class_final_loss: 1.8637 - val_rate_loss: 0.0317 - val_class_1_accuracy: 0.2539 - val_class_2_accuracy: 0.2407 - val_class_final_accuracy: 0.2547\n",
      "Epoch 48/200\n",
      "11548/11548 [==============================] - 8s 708us/sample - loss: 4.7391 - class_1_loss: 1.8966 - class_2_loss: 1.8647 - class_final_loss: 1.8921 - rate_loss: 0.0315 - class_1_accuracy: 0.2398 - class_2_accuracy: 0.2526 - class_final_accuracy: 0.2427 - val_loss: 4.6754 - val_class_1_loss: 1.8780 - val_class_2_loss: 1.8634 - val_class_final_loss: 1.8682 - val_rate_loss: 0.0298 - val_class_1_accuracy: 0.2539 - val_class_2_accuracy: 0.2586 - val_class_final_accuracy: 0.2640\n",
      "Epoch 49/200\n",
      "11548/11548 [==============================] - 8s 711us/sample - loss: 4.7409 - class_1_loss: 1.8922 - class_2_loss: 1.8721 - class_final_loss: 1.8900 - rate_loss: 0.0318 - class_1_accuracy: 0.2484 - class_2_accuracy: 0.2565 - class_final_accuracy: 0.2497 - val_loss: 4.6667 - val_class_1_loss: 1.8753 - val_class_2_loss: 1.8537 - val_class_final_loss: 1.8583 - val_rate_loss: 0.0318 - val_class_1_accuracy: 0.2484 - val_class_2_accuracy: 0.2399 - val_class_final_accuracy: 0.2547\n",
      "Epoch 50/200\n",
      "11548/11548 [==============================] - 8s 700us/sample - loss: 4.7255 - class_1_loss: 1.8883 - class_2_loss: 1.8596 - class_final_loss: 1.8878 - rate_loss: 0.0314 - class_1_accuracy: 0.2548 - class_2_accuracy: 0.2593 - class_final_accuracy: 0.2483 - val_loss: 4.6494 - val_class_1_loss: 1.8709 - val_class_2_loss: 1.8523 - val_class_final_loss: 1.8582 - val_rate_loss: 0.0303 - val_class_1_accuracy: 0.2523 - val_class_2_accuracy: 0.2570 - val_class_final_accuracy: 0.2523\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 51/200\n",
      "11548/11548 [==============================] - 8s 707us/sample - loss: 4.7220 - class_1_loss: 1.8887 - class_2_loss: 1.8630 - class_final_loss: 1.8782 - rate_loss: 0.0310 - class_1_accuracy: 0.2496 - class_2_accuracy: 0.2579 - class_final_accuracy: 0.2567 - val_loss: 4.6832 - val_class_1_loss: 1.8781 - val_class_2_loss: 1.8617 - val_class_final_loss: 1.8850 - val_rate_loss: 0.0295 - val_class_1_accuracy: 0.2687 - val_class_2_accuracy: 0.2617 - val_class_final_accuracy: 0.2617\n",
      "Epoch 52/200\n",
      "11548/11548 [==============================] - 8s 705us/sample - loss: 4.7097 - class_1_loss: 1.8830 - class_2_loss: 1.8532 - class_final_loss: 1.8843 - rate_loss: 0.0309 - class_1_accuracy: 0.2473 - class_2_accuracy: 0.2547 - class_final_accuracy: 0.2518 - val_loss: 4.6698 - val_class_1_loss: 1.8842 - val_class_2_loss: 1.8570 - val_class_final_loss: 1.8590 - val_rate_loss: 0.0299 - val_class_1_accuracy: 0.2578 - val_class_2_accuracy: 0.2617 - val_class_final_accuracy: 0.2578\n",
      "Epoch 53/200\n",
      "11548/11548 [==============================] - 8s 699us/sample - loss: 4.6995 - class_1_loss: 1.8767 - class_2_loss: 1.8523 - class_final_loss: 1.8775 - rate_loss: 0.0307 - class_1_accuracy: 0.2509 - class_2_accuracy: 0.2563 - class_final_accuracy: 0.2416 - val_loss: 4.6514 - val_class_1_loss: 1.8714 - val_class_2_loss: 1.8485 - val_class_final_loss: 1.8509 - val_rate_loss: 0.0311 - val_class_1_accuracy: 0.2578 - val_class_2_accuracy: 0.2438 - val_class_final_accuracy: 0.2679\n",
      "Epoch 54/200\n",
      "11548/11548 [==============================] - 8s 701us/sample - loss: 4.7005 - class_1_loss: 1.8801 - class_2_loss: 1.8522 - class_final_loss: 1.8742 - rate_loss: 0.0307 - class_1_accuracy: 0.2558 - class_2_accuracy: 0.2561 - class_final_accuracy: 0.2472 - val_loss: 4.6303 - val_class_1_loss: 1.8666 - val_class_2_loss: 1.8447 - val_class_final_loss: 1.8481 - val_rate_loss: 0.0293 - val_class_1_accuracy: 0.2523 - val_class_2_accuracy: 0.2601 - val_class_final_accuracy: 0.2695\n",
      "Epoch 55/200\n",
      "11548/11548 [==============================] - 8s 703us/sample - loss: 4.6861 - class_1_loss: 1.8728 - class_2_loss: 1.8470 - class_final_loss: 1.8698 - rate_loss: 0.0304 - class_1_accuracy: 0.2564 - class_2_accuracy: 0.2616 - class_final_accuracy: 0.2524 - val_loss: 4.6546 - val_class_1_loss: 1.8697 - val_class_2_loss: 1.8497 - val_class_final_loss: 1.8512 - val_rate_loss: 0.0297 - val_class_1_accuracy: 0.2555 - val_class_2_accuracy: 0.2508 - val_class_final_accuracy: 0.2648\n",
      "Epoch 56/200\n",
      "11548/11548 [==============================] - 8s 707us/sample - loss: 4.6904 - class_1_loss: 1.8790 - class_2_loss: 1.8468 - class_final_loss: 1.8691 - rate_loss: 0.0304 - class_1_accuracy: 0.2484 - class_2_accuracy: 0.2572 - class_final_accuracy: 0.2536 - val_loss: 4.6308 - val_class_1_loss: 1.8593 - val_class_2_loss: 1.8452 - val_class_final_loss: 1.8496 - val_rate_loss: 0.0302 - val_class_1_accuracy: 0.2555 - val_class_2_accuracy: 0.2648 - val_class_final_accuracy: 0.2679\n",
      "Epoch 57/200\n",
      "11548/11548 [==============================] - 8s 703us/sample - loss: 4.6703 - class_1_loss: 1.8670 - class_2_loss: 1.8441 - class_final_loss: 1.8617 - rate_loss: 0.0296 - class_1_accuracy: 0.2567 - class_2_accuracy: 0.2626 - class_final_accuracy: 0.2529 - val_loss: 4.6263 - val_class_1_loss: 1.8562 - val_class_2_loss: 1.8421 - val_class_final_loss: 1.8468 - val_rate_loss: 0.0287 - val_class_1_accuracy: 0.2648 - val_class_2_accuracy: 0.2695 - val_class_final_accuracy: 0.2726\n",
      "Epoch 58/200\n",
      "11548/11548 [==============================] - 8s 707us/sample - loss: 4.6559 - class_1_loss: 1.8621 - class_2_loss: 1.8329 - class_final_loss: 1.8610 - rate_loss: 0.0297 - class_1_accuracy: 0.2510 - class_2_accuracy: 0.2624 - class_final_accuracy: 0.2567 - val_loss: 4.6070 - val_class_1_loss: 1.8579 - val_class_2_loss: 1.8385 - val_class_final_loss: 1.8416 - val_rate_loss: 0.0293 - val_class_1_accuracy: 0.2710 - val_class_2_accuracy: 0.2695 - val_class_final_accuracy: 0.2734\n",
      "Epoch 59/200\n",
      "11548/11548 [==============================] - 8s 700us/sample - loss: 4.6503 - class_1_loss: 1.8642 - class_2_loss: 1.8304 - class_final_loss: 1.8555 - rate_loss: 0.0295 - class_1_accuracy: 0.2508 - class_2_accuracy: 0.2663 - class_final_accuracy: 0.2548 - val_loss: 4.6173 - val_class_1_loss: 1.8626 - val_class_2_loss: 1.8387 - val_class_final_loss: 1.8368 - val_rate_loss: 0.0289 - val_class_1_accuracy: 0.2687 - val_class_2_accuracy: 0.2695 - val_class_final_accuracy: 0.2773\n",
      "Epoch 60/200\n",
      "11548/11548 [==============================] - 8s 701us/sample - loss: 4.6452 - class_1_loss: 1.8640 - class_2_loss: 1.8271 - class_final_loss: 1.8509 - rate_loss: 0.0293 - class_1_accuracy: 0.2574 - class_2_accuracy: 0.2691 - class_final_accuracy: 0.2580 - val_loss: 4.5962 - val_class_1_loss: 1.8526 - val_class_2_loss: 1.8351 - val_class_final_loss: 1.8390 - val_rate_loss: 0.0296 - val_class_1_accuracy: 0.2695 - val_class_2_accuracy: 0.2687 - val_class_final_accuracy: 0.2726\n",
      "Epoch 61/200\n",
      "11548/11548 [==============================] - 8s 705us/sample - loss: 4.6394 - class_1_loss: 1.8592 - class_2_loss: 1.8263 - class_final_loss: 1.8477 - rate_loss: 0.0290 - class_1_accuracy: 0.2560 - class_2_accuracy: 0.2636 - class_final_accuracy: 0.2574 - val_loss: 4.6148 - val_class_1_loss: 1.8568 - val_class_2_loss: 1.8401 - val_class_final_loss: 1.8411 - val_rate_loss: 0.0284 - val_class_1_accuracy: 0.2695 - val_class_2_accuracy: 0.2695 - val_class_final_accuracy: 0.2702\n",
      "Epoch 62/200\n",
      "11548/11548 [==============================] - 8s 700us/sample - loss: 4.6452 - class_1_loss: 1.8624 - class_2_loss: 1.8286 - class_final_loss: 1.8509 - rate_loss: 0.0296 - class_1_accuracy: 0.2577 - class_2_accuracy: 0.2683 - class_final_accuracy: 0.2573 - val_loss: 4.5831 - val_class_1_loss: 1.8469 - val_class_2_loss: 1.8269 - val_class_final_loss: 1.8335 - val_rate_loss: 0.0279 - val_class_1_accuracy: 0.2625 - val_class_2_accuracy: 0.2625 - val_class_final_accuracy: 0.2741\n",
      "Epoch 63/200\n",
      "11548/11548 [==============================] - 8s 701us/sample - loss: 4.6259 - class_1_loss: 1.8536 - class_2_loss: 1.8209 - class_final_loss: 1.8443 - rate_loss: 0.0290 - class_1_accuracy: 0.2607 - class_2_accuracy: 0.2742 - class_final_accuracy: 0.2607 - val_loss: 4.6500 - val_class_1_loss: 1.8748 - val_class_2_loss: 1.8590 - val_class_final_loss: 1.8592 - val_rate_loss: 0.0294 - val_class_1_accuracy: 0.2710 - val_class_2_accuracy: 0.2664 - val_class_final_accuracy: 0.2648\n",
      "Epoch 64/200\n",
      "11548/11548 [==============================] - 8s 705us/sample - loss: 4.6141 - class_1_loss: 1.8513 - class_2_loss: 1.8129 - class_final_loss: 1.8395 - rate_loss: 0.0288 - class_1_accuracy: 0.2622 - class_2_accuracy: 0.2710 - class_final_accuracy: 0.2653 - val_loss: 4.6044 - val_class_1_loss: 1.8576 - val_class_2_loss: 1.8419 - val_class_final_loss: 1.8441 - val_rate_loss: 0.0286 - val_class_1_accuracy: 0.2640 - val_class_2_accuracy: 0.2625 - val_class_final_accuracy: 0.2695\n",
      "Epoch 65/200\n",
      "11548/11548 [==============================] - 8s 700us/sample - loss: 4.6058 - class_1_loss: 1.8480 - class_2_loss: 1.8103 - class_final_loss: 1.8368 - rate_loss: 0.0285 - class_1_accuracy: 0.2553 - class_2_accuracy: 0.2728 - class_final_accuracy: 0.2651 - val_loss: 4.6174 - val_class_1_loss: 1.8567 - val_class_2_loss: 1.8464 - val_class_final_loss: 1.8480 - val_rate_loss: 0.0272 - val_class_1_accuracy: 0.2586 - val_class_2_accuracy: 0.2695 - val_class_final_accuracy: 0.2773\n",
      "Epoch 66/200\n",
      "11548/11548 [==============================] - 8s 705us/sample - loss: 4.6048 - class_1_loss: 1.8487 - class_2_loss: 1.8119 - class_final_loss: 1.8302 - rate_loss: 0.0284 - class_1_accuracy: 0.2537 - class_2_accuracy: 0.2742 - class_final_accuracy: 0.2694 - val_loss: 4.5834 - val_class_1_loss: 1.8560 - val_class_2_loss: 1.8283 - val_class_final_loss: 1.8257 - val_rate_loss: 0.0280 - val_class_1_accuracy: 0.2625 - val_class_2_accuracy: 0.2687 - val_class_final_accuracy: 0.2695\n",
      "Epoch 67/200\n",
      "11548/11548 [==============================] - 8s 707us/sample - loss: 4.6101 - class_1_loss: 1.8512 - class_2_loss: 1.8136 - class_final_loss: 1.8338 - rate_loss: 0.0287 - class_1_accuracy: 0.2538 - class_2_accuracy: 0.2691 - class_final_accuracy: 0.2637 - val_loss: 4.5898 - val_class_1_loss: 1.8550 - val_class_2_loss: 1.8322 - val_class_final_loss: 1.8299 - val_rate_loss: 0.0272 - val_class_1_accuracy: 0.2664 - val_class_2_accuracy: 0.2765 - val_class_final_accuracy: 0.2788\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 68/200\n",
      "11548/11548 [==============================] - 8s 703us/sample - loss: 4.5829 - class_1_loss: 1.8397 - class_2_loss: 1.8030 - class_final_loss: 1.8234 - rate_loss: 0.0283 - class_1_accuracy: 0.2610 - class_2_accuracy: 0.2749 - class_final_accuracy: 0.2718 - val_loss: 4.5725 - val_class_1_loss: 1.8441 - val_class_2_loss: 1.8209 - val_class_final_loss: 1.8279 - val_rate_loss: 0.0270 - val_class_1_accuracy: 0.2609 - val_class_2_accuracy: 0.2687 - val_class_final_accuracy: 0.2734\n",
      "Epoch 69/200\n",
      "11548/11548 [==============================] - 8s 708us/sample - loss: 4.5868 - class_1_loss: 1.8393 - class_2_loss: 1.8052 - class_final_loss: 1.8272 - rate_loss: 0.0286 - class_1_accuracy: 0.2642 - class_2_accuracy: 0.2749 - class_final_accuracy: 0.2641 - val_loss: 4.5951 - val_class_1_loss: 1.8614 - val_class_2_loss: 1.8395 - val_class_final_loss: 1.8384 - val_rate_loss: 0.0272 - val_class_1_accuracy: 0.2562 - val_class_2_accuracy: 0.2679 - val_class_final_accuracy: 0.2819\n",
      "Epoch 70/200\n",
      "11548/11548 [==============================] - 8s 704us/sample - loss: 4.5727 - class_1_loss: 1.8329 - class_2_loss: 1.7985 - class_final_loss: 1.8237 - rate_loss: 0.0282 - class_1_accuracy: 0.2624 - class_2_accuracy: 0.2765 - class_final_accuracy: 0.2664 - val_loss: 4.5730 - val_class_1_loss: 1.8549 - val_class_2_loss: 1.8319 - val_class_final_loss: 1.8253 - val_rate_loss: 0.0283 - val_class_1_accuracy: 0.2671 - val_class_2_accuracy: 0.2586 - val_class_final_accuracy: 0.2734\n",
      "Epoch 71/200\n",
      "11548/11548 [==============================] - 8s 709us/sample - loss: 4.5701 - class_1_loss: 1.8405 - class_2_loss: 1.7931 - class_final_loss: 1.8198 - rate_loss: 0.0280 - class_1_accuracy: 0.2568 - class_2_accuracy: 0.2794 - class_final_accuracy: 0.2729 - val_loss: 4.5733 - val_class_1_loss: 1.8547 - val_class_2_loss: 1.8233 - val_class_final_loss: 1.8241 - val_rate_loss: 0.0269 - val_class_1_accuracy: 0.2741 - val_class_2_accuracy: 0.2710 - val_class_final_accuracy: 0.2882\n",
      "Epoch 72/200\n",
      "11548/11548 [==============================] - 8s 705us/sample - loss: 4.5673 - class_1_loss: 1.8355 - class_2_loss: 1.7959 - class_final_loss: 1.8148 - rate_loss: 0.0279 - class_1_accuracy: 0.2635 - class_2_accuracy: 0.2822 - class_final_accuracy: 0.2659 - val_loss: 4.5920 - val_class_1_loss: 1.8622 - val_class_2_loss: 1.8351 - val_class_final_loss: 1.8339 - val_rate_loss: 0.0275 - val_class_1_accuracy: 0.2601 - val_class_2_accuracy: 0.2702 - val_class_final_accuracy: 0.2741\n",
      "Epoch 73/200\n",
      "11548/11548 [==============================] - 8s 703us/sample - loss: 4.5551 - class_1_loss: 1.8295 - class_2_loss: 1.7907 - class_final_loss: 1.8128 - rate_loss: 0.0278 - class_1_accuracy: 0.2652 - class_2_accuracy: 0.2729 - class_final_accuracy: 0.2642 - val_loss: 4.5922 - val_class_1_loss: 1.8580 - val_class_2_loss: 1.8273 - val_class_final_loss: 1.8308 - val_rate_loss: 0.0267 - val_class_1_accuracy: 0.2578 - val_class_2_accuracy: 0.2749 - val_class_final_accuracy: 0.2702\n",
      "Epoch 74/200\n",
      "11548/11548 [==============================] - 8s 707us/sample - loss: 4.5451 - class_1_loss: 1.8287 - class_2_loss: 1.7848 - class_final_loss: 1.8106 - rate_loss: 0.0276 - class_1_accuracy: 0.2609 - class_2_accuracy: 0.2837 - class_final_accuracy: 0.2677 - val_loss: 4.5817 - val_class_1_loss: 1.8538 - val_class_2_loss: 1.8249 - val_class_final_loss: 1.8284 - val_rate_loss: 0.0269 - val_class_1_accuracy: 0.2702 - val_class_2_accuracy: 0.2796 - val_class_final_accuracy: 0.2866\n",
      "Epoch 75/200\n",
      "11548/11548 [==============================] - 8s 713us/sample - loss: 4.5478 - class_1_loss: 1.8329 - class_2_loss: 1.7843 - class_final_loss: 1.8032 - rate_loss: 0.0273 - class_1_accuracy: 0.2637 - class_2_accuracy: 0.2840 - class_final_accuracy: 0.2776 - val_loss: 4.5488 - val_class_1_loss: 1.8430 - val_class_2_loss: 1.8175 - val_class_final_loss: 1.8197 - val_rate_loss: 0.0269 - val_class_1_accuracy: 0.2617 - val_class_2_accuracy: 0.2843 - val_class_final_accuracy: 0.2812\n",
      "Epoch 76/200\n",
      "11548/11548 [==============================] - 8s 699us/sample - loss: 4.5309 - class_1_loss: 1.8246 - class_2_loss: 1.7792 - class_final_loss: 1.8011 - rate_loss: 0.0274 - class_1_accuracy: 0.2684 - class_2_accuracy: 0.2780 - class_final_accuracy: 0.2743 - val_loss: 4.5508 - val_class_1_loss: 1.8435 - val_class_2_loss: 1.8151 - val_class_final_loss: 1.8191 - val_rate_loss: 0.0274 - val_class_1_accuracy: 0.2757 - val_class_2_accuracy: 0.2835 - val_class_final_accuracy: 0.2765\n",
      "Epoch 77/200\n",
      "11548/11548 [==============================] - 8s 703us/sample - loss: 4.5387 - class_1_loss: 1.8284 - class_2_loss: 1.7800 - class_final_loss: 1.8015 - rate_loss: 0.0270 - class_1_accuracy: 0.2692 - class_2_accuracy: 0.2853 - class_final_accuracy: 0.2742 - val_loss: 4.5558 - val_class_1_loss: 1.8554 - val_class_2_loss: 1.8326 - val_class_final_loss: 1.8318 - val_rate_loss: 0.0277 - val_class_1_accuracy: 0.2555 - val_class_2_accuracy: 0.2632 - val_class_final_accuracy: 0.2749\n",
      "Epoch 78/200\n",
      "11548/11548 [==============================] - 8s 705us/sample - loss: 4.5145 - class_1_loss: 1.8211 - class_2_loss: 1.7697 - class_final_loss: 1.7917 - rate_loss: 0.0266 - class_1_accuracy: 0.2704 - class_2_accuracy: 0.2821 - class_final_accuracy: 0.2797 - val_loss: 4.5851 - val_class_1_loss: 1.8512 - val_class_2_loss: 1.8300 - val_class_final_loss: 1.8409 - val_rate_loss: 0.0269 - val_class_1_accuracy: 0.2578 - val_class_2_accuracy: 0.2757 - val_class_final_accuracy: 0.2819\n",
      "Epoch 79/200\n",
      "11548/11548 [==============================] - 8s 701us/sample - loss: 4.5232 - class_1_loss: 1.8249 - class_2_loss: 1.7721 - class_final_loss: 1.7976 - rate_loss: 0.0266 - class_1_accuracy: 0.2659 - class_2_accuracy: 0.2841 - class_final_accuracy: 0.2753 - val_loss: 4.5832 - val_class_1_loss: 1.8515 - val_class_2_loss: 1.8360 - val_class_final_loss: 1.8356 - val_rate_loss: 0.0272 - val_class_1_accuracy: 0.2562 - val_class_2_accuracy: 0.2702 - val_class_final_accuracy: 0.2702\n",
      "Epoch 80/200\n",
      "11548/11548 [==============================] - 8s 705us/sample - loss: 4.4955 - class_1_loss: 1.8118 - class_2_loss: 1.7633 - class_final_loss: 1.7867 - rate_loss: 0.0267 - class_1_accuracy: 0.2743 - class_2_accuracy: 0.2836 - class_final_accuracy: 0.2771 - val_loss: 4.5451 - val_class_1_loss: 1.8353 - val_class_2_loss: 1.8225 - val_class_final_loss: 1.8320 - val_rate_loss: 0.0283 - val_class_1_accuracy: 0.2687 - val_class_2_accuracy: 0.2765 - val_class_final_accuracy: 0.2726\n",
      "Epoch 81/200\n",
      "11548/11548 [==============================] - 8s 703us/sample - loss: 4.4944 - class_1_loss: 1.8141 - class_2_loss: 1.7622 - class_final_loss: 1.7824 - rate_loss: 0.0265 - class_1_accuracy: 0.2727 - class_2_accuracy: 0.2930 - class_final_accuracy: 0.2786 - val_loss: 4.5496 - val_class_1_loss: 1.8353 - val_class_2_loss: 1.8249 - val_class_final_loss: 1.8447 - val_rate_loss: 0.0281 - val_class_1_accuracy: 0.2804 - val_class_2_accuracy: 0.2780 - val_class_final_accuracy: 0.2835\n",
      "Epoch 82/200\n",
      "11548/11548 [==============================] - 8s 704us/sample - loss: 4.5017 - class_1_loss: 1.8185 - class_2_loss: 1.7653 - class_final_loss: 1.7832 - rate_loss: 0.0264 - class_1_accuracy: 0.2680 - class_2_accuracy: 0.2850 - class_final_accuracy: 0.2796 - val_loss: 4.5783 - val_class_1_loss: 1.8452 - val_class_2_loss: 1.8304 - val_class_final_loss: 1.8397 - val_rate_loss: 0.0264 - val_class_1_accuracy: 0.2726 - val_class_2_accuracy: 0.2734 - val_class_final_accuracy: 0.2679\n",
      "Epoch 83/200\n",
      "11548/11548 [==============================] - 8s 700us/sample - loss: 4.4836 - class_1_loss: 1.8136 - class_2_loss: 1.7552 - class_final_loss: 1.7779 - rate_loss: 0.0260 - class_1_accuracy: 0.2674 - class_2_accuracy: 0.2864 - class_final_accuracy: 0.2761 - val_loss: 4.5797 - val_class_1_loss: 1.8481 - val_class_2_loss: 1.8405 - val_class_final_loss: 1.8453 - val_rate_loss: 0.0283 - val_class_1_accuracy: 0.2679 - val_class_2_accuracy: 0.2671 - val_class_final_accuracy: 0.2664\n",
      "Epoch 84/200\n",
      "11548/11548 [==============================] - 8s 703us/sample - loss: 4.4708 - class_1_loss: 1.8040 - class_2_loss: 1.7525 - class_final_loss: 1.7777 - rate_loss: 0.0262 - class_1_accuracy: 0.2738 - class_2_accuracy: 0.2900 - class_final_accuracy: 0.2822 - val_loss: 4.5474 - val_class_1_loss: 1.8387 - val_class_2_loss: 1.8236 - val_class_final_loss: 1.8329 - val_rate_loss: 0.0280 - val_class_1_accuracy: 0.2640 - val_class_2_accuracy: 0.2812 - val_class_final_accuracy: 0.2640\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 85/200\n",
      "11548/11548 [==============================] - 8s 707us/sample - loss: 4.4679 - class_1_loss: 1.8050 - class_2_loss: 1.7529 - class_final_loss: 1.7707 - rate_loss: 0.0259 - class_1_accuracy: 0.2735 - class_2_accuracy: 0.2968 - class_final_accuracy: 0.2840 - val_loss: 4.5846 - val_class_1_loss: 1.8527 - val_class_2_loss: 1.8420 - val_class_final_loss: 1.8450 - val_rate_loss: 0.0270 - val_class_1_accuracy: 0.2664 - val_class_2_accuracy: 0.2835 - val_class_final_accuracy: 0.2819\n",
      "Epoch 86/200\n",
      "11548/11548 [==============================] - 8s 701us/sample - loss: 4.4578 - class_1_loss: 1.8017 - class_2_loss: 1.7434 - class_final_loss: 1.7711 - rate_loss: 0.0259 - class_1_accuracy: 0.2754 - class_2_accuracy: 0.2905 - class_final_accuracy: 0.2841 - val_loss: 4.5675 - val_class_1_loss: 1.8494 - val_class_2_loss: 1.8359 - val_class_final_loss: 1.8366 - val_rate_loss: 0.0279 - val_class_1_accuracy: 0.2601 - val_class_2_accuracy: 0.2827 - val_class_final_accuracy: 0.2749\n",
      "Epoch 87/200\n",
      "11548/11548 [==============================] - 8s 703us/sample - loss: 4.4463 - class_1_loss: 1.7989 - class_2_loss: 1.7396 - class_final_loss: 1.7640 - rate_loss: 0.0259 - class_1_accuracy: 0.2720 - class_2_accuracy: 0.2969 - class_final_accuracy: 0.2931 - val_loss: 4.5753 - val_class_1_loss: 1.8465 - val_class_2_loss: 1.8324 - val_class_final_loss: 1.8409 - val_rate_loss: 0.0267 - val_class_1_accuracy: 0.2609 - val_class_2_accuracy: 0.2866 - val_class_final_accuracy: 0.2804\n",
      "Epoch 88/200\n",
      "11548/11548 [==============================] - 8s 708us/sample - loss: 4.4412 - class_1_loss: 1.8026 - class_2_loss: 1.7359 - class_final_loss: 1.7562 - rate_loss: 0.0253 - class_1_accuracy: 0.2787 - class_2_accuracy: 0.2970 - class_final_accuracy: 0.2889 - val_loss: 4.6087 - val_class_1_loss: 1.8488 - val_class_2_loss: 1.8613 - val_class_final_loss: 1.8759 - val_rate_loss: 0.0280 - val_class_1_accuracy: 0.2687 - val_class_2_accuracy: 0.2835 - val_class_final_accuracy: 0.2695\n",
      "Epoch 89/200\n",
      "11548/11548 [==============================] - 8s 705us/sample - loss: 4.4363 - class_1_loss: 1.7996 - class_2_loss: 1.7320 - class_final_loss: 1.7579 - rate_loss: 0.0253 - class_1_accuracy: 0.2726 - class_2_accuracy: 0.3011 - class_final_accuracy: 0.2896 - val_loss: 4.5737 - val_class_1_loss: 1.8407 - val_class_2_loss: 1.8386 - val_class_final_loss: 1.8449 - val_rate_loss: 0.0267 - val_class_1_accuracy: 0.2617 - val_class_2_accuracy: 0.2757 - val_class_final_accuracy: 0.2780\n",
      "Epoch 90/200\n",
      "11548/11548 [==============================] - 8s 708us/sample - loss: 4.4259 - class_1_loss: 1.7908 - class_2_loss: 1.7297 - class_final_loss: 1.7595 - rate_loss: 0.0254 - class_1_accuracy: 0.2719 - class_2_accuracy: 0.2962 - class_final_accuracy: 0.2852 - val_loss: 4.5800 - val_class_1_loss: 1.8546 - val_class_2_loss: 1.8455 - val_class_final_loss: 1.8420 - val_rate_loss: 0.0270 - val_class_1_accuracy: 0.2632 - val_class_2_accuracy: 0.2757 - val_class_final_accuracy: 0.2765\n",
      "Epoch 91/200\n",
      "11548/11548 [==============================] - 8s 703us/sample - loss: 4.4107 - class_1_loss: 1.7855 - class_2_loss: 1.7226 - class_final_loss: 1.7497 - rate_loss: 0.0255 - class_1_accuracy: 0.2807 - class_2_accuracy: 0.3026 - class_final_accuracy: 0.2894 - val_loss: 4.5845 - val_class_1_loss: 1.8538 - val_class_2_loss: 1.8508 - val_class_final_loss: 1.8551 - val_rate_loss: 0.0277 - val_class_1_accuracy: 0.2648 - val_class_2_accuracy: 0.2741 - val_class_final_accuracy: 0.2640\n",
      "Epoch 92/200\n",
      "11548/11548 [==============================] - 8s 706us/sample - loss: 4.4136 - class_1_loss: 1.7899 - class_2_loss: 1.7252 - class_final_loss: 1.7500 - rate_loss: 0.0250 - class_1_accuracy: 0.2766 - class_2_accuracy: 0.3027 - class_final_accuracy: 0.2896 - val_loss: 4.5661 - val_class_1_loss: 1.8421 - val_class_2_loss: 1.8356 - val_class_final_loss: 1.8396 - val_rate_loss: 0.0267 - val_class_1_accuracy: 0.2671 - val_class_2_accuracy: 0.2905 - val_class_final_accuracy: 0.2741\n",
      "Epoch 93/200\n",
      "11548/11548 [==============================] - 8s 703us/sample - loss: 4.4009 - class_1_loss: 1.7885 - class_2_loss: 1.7175 - class_final_loss: 1.7416 - rate_loss: 0.0250 - class_1_accuracy: 0.2797 - class_2_accuracy: 0.3001 - class_final_accuracy: 0.2922 - val_loss: 4.5829 - val_class_1_loss: 1.8519 - val_class_2_loss: 1.8377 - val_class_final_loss: 1.8365 - val_rate_loss: 0.0268 - val_class_1_accuracy: 0.2710 - val_class_2_accuracy: 0.2687 - val_class_final_accuracy: 0.2757\n",
      "Epoch 94/200\n",
      "11548/11548 [==============================] - 8s 697us/sample - loss: 4.3995 - class_1_loss: 1.7879 - class_2_loss: 1.7153 - class_final_loss: 1.7438 - rate_loss: 0.0251 - class_1_accuracy: 0.2849 - class_2_accuracy: 0.3018 - class_final_accuracy: 0.2900 - val_loss: 4.5688 - val_class_1_loss: 1.8460 - val_class_2_loss: 1.8401 - val_class_final_loss: 1.8417 - val_rate_loss: 0.0270 - val_class_1_accuracy: 0.2578 - val_class_2_accuracy: 0.2687 - val_class_final_accuracy: 0.2695\n",
      "Epoch 95/200\n",
      "11548/11548 [==============================] - 8s 705us/sample - loss: 4.3967 - class_1_loss: 1.7871 - class_2_loss: 1.7171 - class_final_loss: 1.7393 - rate_loss: 0.0246 - class_1_accuracy: 0.2761 - class_2_accuracy: 0.3019 - class_final_accuracy: 0.2946 - val_loss: 4.5619 - val_class_1_loss: 1.8390 - val_class_2_loss: 1.8424 - val_class_final_loss: 1.8464 - val_rate_loss: 0.0272 - val_class_1_accuracy: 0.2664 - val_class_2_accuracy: 0.2843 - val_class_final_accuracy: 0.2827\n",
      "Epoch 96/200\n",
      "11548/11548 [==============================] - 8s 702us/sample - loss: 4.3854 - class_1_loss: 1.7790 - class_2_loss: 1.7149 - class_final_loss: 1.7345 - rate_loss: 0.0248 - class_1_accuracy: 0.2887 - class_2_accuracy: 0.3067 - class_final_accuracy: 0.2967 - val_loss: 4.5932 - val_class_1_loss: 1.8619 - val_class_2_loss: 1.8350 - val_class_final_loss: 1.8344 - val_rate_loss: 0.0264 - val_class_1_accuracy: 0.2687 - val_class_2_accuracy: 0.2695 - val_class_final_accuracy: 0.2734\n",
      "Epoch 97/200\n",
      "11548/11548 [==============================] - 8s 702us/sample - loss: 4.3744 - class_1_loss: 1.7794 - class_2_loss: 1.7054 - class_final_loss: 1.7316 - rate_loss: 0.0246 - class_1_accuracy: 0.2839 - class_2_accuracy: 0.3088 - class_final_accuracy: 0.2949 - val_loss: 4.5613 - val_class_1_loss: 1.8336 - val_class_2_loss: 1.8364 - val_class_final_loss: 1.8487 - val_rate_loss: 0.0266 - val_class_1_accuracy: 0.2757 - val_class_2_accuracy: 0.2749 - val_class_final_accuracy: 0.2757\n",
      "Epoch 98/200\n",
      "11548/11548 [==============================] - 8s 707us/sample - loss: 4.3742 - class_1_loss: 1.7759 - class_2_loss: 1.7072 - class_final_loss: 1.7322 - rate_loss: 0.0245 - class_1_accuracy: 0.2764 - class_2_accuracy: 0.3055 - class_final_accuracy: 0.2941 - val_loss: 4.5523 - val_class_1_loss: 1.8359 - val_class_2_loss: 1.8281 - val_class_final_loss: 1.8295 - val_rate_loss: 0.0272 - val_class_1_accuracy: 0.2734 - val_class_2_accuracy: 0.2656 - val_class_final_accuracy: 0.2710\n",
      "Epoch 99/200\n",
      "11548/11548 [==============================] - 8s 703us/sample - loss: 4.3627 - class_1_loss: 1.7832 - class_2_loss: 1.6936 - class_final_loss: 1.7233 - rate_loss: 0.0240 - class_1_accuracy: 0.2836 - class_2_accuracy: 0.3073 - class_final_accuracy: 0.3001 - val_loss: 4.5964 - val_class_1_loss: 1.8414 - val_class_2_loss: 1.8536 - val_class_final_loss: 1.8648 - val_rate_loss: 0.0268 - val_class_1_accuracy: 0.2718 - val_class_2_accuracy: 0.2866 - val_class_final_accuracy: 0.2819\n",
      "Epoch 100/200\n",
      "11548/11548 [==============================] - 8s 701us/sample - loss: 4.3569 - class_1_loss: 1.7764 - class_2_loss: 1.6977 - class_final_loss: 1.7198 - rate_loss: 0.0239 - class_1_accuracy: 0.2800 - class_2_accuracy: 0.3088 - class_final_accuracy: 0.2961 - val_loss: 4.6207 - val_class_1_loss: 1.8472 - val_class_2_loss: 1.8669 - val_class_final_loss: 1.8771 - val_rate_loss: 0.0271 - val_class_1_accuracy: 0.2640 - val_class_2_accuracy: 0.2632 - val_class_final_accuracy: 0.2718\n",
      "Epoch 101/200\n",
      "11548/11548 [==============================] - 8s 706us/sample - loss: 4.3463 - class_1_loss: 1.7654 - class_2_loss: 1.6982 - class_final_loss: 1.7170 - rate_loss: 0.0239 - class_1_accuracy: 0.2878 - class_2_accuracy: 0.3069 - class_final_accuracy: 0.3117 - val_loss: 4.5966 - val_class_1_loss: 1.8564 - val_class_2_loss: 1.8510 - val_class_final_loss: 1.8522 - val_rate_loss: 0.0269 - val_class_1_accuracy: 0.2648 - val_class_2_accuracy: 0.2671 - val_class_final_accuracy: 0.2718\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 102/200\n",
      "11548/11548 [==============================] - 8s 706us/sample - loss: 4.3413 - class_1_loss: 1.7718 - class_2_loss: 1.6894 - class_final_loss: 1.7107 - rate_loss: 0.0236 - class_1_accuracy: 0.2879 - class_2_accuracy: 0.3156 - class_final_accuracy: 0.3099 - val_loss: 4.5812 - val_class_1_loss: 1.8408 - val_class_2_loss: 1.8451 - val_class_final_loss: 1.8479 - val_rate_loss: 0.0270 - val_class_1_accuracy: 0.2648 - val_class_2_accuracy: 0.2812 - val_class_final_accuracy: 0.2819\n",
      "Epoch 103/200\n",
      "11548/11548 [==============================] - 8s 703us/sample - loss: 4.3164 - class_1_loss: 1.7617 - class_2_loss: 1.6801 - class_final_loss: 1.6998 - rate_loss: 0.0232 - class_1_accuracy: 0.2927 - class_2_accuracy: 0.3158 - class_final_accuracy: 0.3061 - val_loss: 4.5764 - val_class_1_loss: 1.8397 - val_class_2_loss: 1.8506 - val_class_final_loss: 1.8656 - val_rate_loss: 0.0270 - val_class_1_accuracy: 0.2765 - val_class_2_accuracy: 0.2765 - val_class_final_accuracy: 0.2936\n",
      "Epoch 104/200\n",
      "11548/11548 [==============================] - 8s 707us/sample - loss: 4.3177 - class_1_loss: 1.7554 - class_2_loss: 1.6855 - class_final_loss: 1.7046 - rate_loss: 0.0233 - class_1_accuracy: 0.2926 - class_2_accuracy: 0.3153 - class_final_accuracy: 0.3039 - val_loss: 4.6061 - val_class_1_loss: 1.8488 - val_class_2_loss: 1.8579 - val_class_final_loss: 1.8614 - val_rate_loss: 0.0268 - val_class_1_accuracy: 0.2679 - val_class_2_accuracy: 0.2718 - val_class_final_accuracy: 0.2788\n",
      "Epoch 105/200\n",
      "11548/11548 [==============================] - 8s 703us/sample - loss: 4.3055 - class_1_loss: 1.7571 - class_2_loss: 1.6762 - class_final_loss: 1.7006 - rate_loss: 0.0233 - class_1_accuracy: 0.2878 - class_2_accuracy: 0.3109 - class_final_accuracy: 0.3042 - val_loss: 4.5713 - val_class_1_loss: 1.8440 - val_class_2_loss: 1.8517 - val_class_final_loss: 1.8519 - val_rate_loss: 0.0274 - val_class_1_accuracy: 0.2734 - val_class_2_accuracy: 0.2741 - val_class_final_accuracy: 0.2609\n",
      "Epoch 106/200\n",
      "11548/11548 [==============================] - 8s 704us/sample - loss: 4.3053 - class_1_loss: 1.7568 - class_2_loss: 1.6735 - class_final_loss: 1.7023 - rate_loss: 0.0233 - class_1_accuracy: 0.2908 - class_2_accuracy: 0.3166 - class_final_accuracy: 0.3014 - val_loss: 4.6032 - val_class_1_loss: 1.8492 - val_class_2_loss: 1.8760 - val_class_final_loss: 1.8748 - val_rate_loss: 0.0280 - val_class_1_accuracy: 0.2625 - val_class_2_accuracy: 0.2648 - val_class_final_accuracy: 0.2780\n",
      "Epoch 107/200\n",
      "11548/11548 [==============================] - 8s 699us/sample - loss: 4.2940 - class_1_loss: 1.7504 - class_2_loss: 1.6704 - class_final_loss: 1.6991 - rate_loss: 0.0234 - class_1_accuracy: 0.2932 - class_2_accuracy: 0.3152 - class_final_accuracy: 0.3119 - val_loss: 4.5832 - val_class_1_loss: 1.8360 - val_class_2_loss: 1.8426 - val_class_final_loss: 1.8446 - val_rate_loss: 0.0269 - val_class_1_accuracy: 0.2656 - val_class_2_accuracy: 0.2702 - val_class_final_accuracy: 0.2695\n",
      "Epoch 108/200\n",
      "11548/11548 [==============================] - 8s 703us/sample - loss: 4.2758 - class_1_loss: 1.7538 - class_2_loss: 1.6561 - class_final_loss: 1.6869 - rate_loss: 0.0230 - class_1_accuracy: 0.2898 - class_2_accuracy: 0.3222 - class_final_accuracy: 0.3137 - val_loss: 4.6199 - val_class_1_loss: 1.8522 - val_class_2_loss: 1.8756 - val_class_final_loss: 1.8720 - val_rate_loss: 0.0280 - val_class_1_accuracy: 0.2570 - val_class_2_accuracy: 0.2718 - val_class_final_accuracy: 0.2765\n",
      "Epoch 109/200\n",
      "11548/11548 [==============================] - 8s 704us/sample - loss: 4.2735 - class_1_loss: 1.7463 - class_2_loss: 1.6621 - class_final_loss: 1.6848 - rate_loss: 0.0230 - class_1_accuracy: 0.2972 - class_2_accuracy: 0.3188 - class_final_accuracy: 0.3134 - val_loss: 4.6159 - val_class_1_loss: 1.8382 - val_class_2_loss: 1.8766 - val_class_final_loss: 1.8968 - val_rate_loss: 0.0270 - val_class_1_accuracy: 0.2664 - val_class_2_accuracy: 0.2858 - val_class_final_accuracy: 0.2796\n",
      "Epoch 110/200\n",
      "11548/11548 [==============================] - 8s 695us/sample - loss: 4.2604 - class_1_loss: 1.7416 - class_2_loss: 1.6567 - class_final_loss: 1.6794 - rate_loss: 0.0228 - class_1_accuracy: 0.2908 - class_2_accuracy: 0.3166 - class_final_accuracy: 0.3127 - val_loss: 4.6270 - val_class_1_loss: 1.8470 - val_class_2_loss: 1.8820 - val_class_final_loss: 1.8844 - val_rate_loss: 0.0272 - val_class_1_accuracy: 0.2687 - val_class_2_accuracy: 0.2819 - val_class_final_accuracy: 0.2850\n"
     ]
    }
   ],
   "source": [
    "CNNLSTM_SpCas9_model.model_train()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0.7130656689159826, 2.5229186083729905e-85)"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "CNNLSTM_SpCas9_model.model_evaluate(1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_prediction = CNNLSTM_SpCas9_model.CNNLSTM_regression_model.predict(CNNLSTM_SpCas9_model.X_test)\n",
    "test_prediction = test_prediction.reshape(-1,)\n",
    "test_true = CNNLSTM_SpCas9_model.rate_test.reshape(-1,)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(542,)"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "image/png": {
       "height": 505,
       "width": 510
      },
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = CNNLSTM_SpCas9_model.plot_scatter(test_prediction,test_true,'Test data')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "image = CNNLSTM_SpCas9_model.log_plot(epoch=1,\n",
    "                                      model=CNNLSTM_SpCas9_model.CNNLSTM_regression_model,\n",
    "                                     input_data=CNNLSTM_SpCas9_model.X_test,\n",
    "                                     true_data=CNNLSTM_SpCas9_model.rate_test,\n",
    "                                     plot=CNNLSTM_SpCas9_model.plot_scatter,\n",
    "                                     )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}