deep_mpc_trainer_v2.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "instrumental-college",
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import random\n",
    "\n",
    "import tensorflow as tf\n",
    "from tensorflow import keras\n",
    "from keras.models import Model, load_model\n",
    "from keras.layers import Dense, LSTM, Dropout, Input, Concatenate, Normalization, Conv2D, Flatten\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "spare-swaziland",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "================================\n",
      "current data size: \n",
      "\n",
      "(406340, 2) (406341, 110) (406341, 110) (406340, 2) (406340, 20, 20)\n",
      "reduced to: \n",
      "\n",
      "(406340, 2) (406340, 110) (406340, 110) (406340, 2) (406340, 20, 20)\n"
     ]
    }
   ],
   "source": [
    "FOLDER_INDEX = 8\n",
    "PLAN_LENGTH = 50\n",
    "DATASET_ADDRESS = f'../../bag/log_{FOLDER_INDEX}'\n",
    "\n",
    "cmdVel = pd.read_csv(\n",
    "    f'{DATASET_ADDRESS}/cmd_vel_log.csv', delimiter=',').values\n",
    "heading = pd.read_csv(f'{DATASET_ADDRESS}/heading.csv', delimiter=',').values\n",
    "costmap = pd.read_csv(f'{DATASET_ADDRESS}/costmap.csv',\n",
    "                      delimiter=',').values.reshape((-1, 20, 20))\n",
    "planX = pd.read_csv(f'{DATASET_ADDRESS}/global_plan_x_log.csv',\n",
    "                    delimiter=',', names=list(range(110)), header=None).values\n",
    "planY = pd.read_csv(f'{DATASET_ADDRESS}/global_plan_y_log.csv',\n",
    "                    delimiter=',', names=list(range(110)), header=None).values\n",
    "\n",
    "print(\"================================\")\n",
    "print(\"current data size: \\n\")\n",
    "print(cmdVel.shape, planX.shape, planY.shape, heading.shape, costmap.shape)\n",
    "len = np.amin([cmdVel.shape[0], planX.shape[0], planY.shape[0],\n",
    "              heading.shape[0], costmap.shape[0]])\n",
    "cmdVel = cmdVel[:len]\n",
    "heading = heading[:len]\n",
    "planX = planX[:len]\n",
    "planY = planY[:len]\n",
    "costmap = costmap[:len]\n",
    "print(\"reduced to: \\n\")\n",
    "print(cmdVel.shape, planX.shape, planY.shape, heading.shape, costmap.shape)\n",
    "\n",
    "planX = planX[:, :PLAN_LENGTH]\n",
    "planY = planY[:, :PLAN_LENGTH]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "proof-binary",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "================================\n",
      "deleting NaNs; size reduced to:\n",
      " 306746\n"
     ]
    }
   ],
   "source": [
    "def emptyRows(arr):\n",
    "    delList = []\n",
    "    for r, row in enumerate(arr):\n",
    "        if np.any(np.isnan(row)):\n",
    "            delList += [r]\n",
    "    return delList\n",
    "\n",
    "\n",
    "delList1 = emptyRows(planX)\n",
    "delList2 = emptyRows(planY)\n",
    "assert delList1 == delList2\n",
    "\n",
    "planX = np.delete(planX, delList1, axis=0)\n",
    "planY = np.delete(planY, delList1, axis=0)\n",
    "cmdVel = np.delete(cmdVel, delList1, axis=0)\n",
    "heading = np.delete(heading, delList1, axis=0)\n",
    "costmap = np.delete(costmap, delList1, axis=0)\n",
    "\n",
    "print(\"================================\")\n",
    "print(f'deleting NaNs; size reduced to:\\n {planX.shape[0]}')\n",
    "\n",
    "assert not np.any(np.isnan(planX))\n",
    "assert not np.any(np.isnan(planY))\n",
    "assert not np.any(np.isnan(cmdVel))\n",
    "assert not np.any(np.isnan(heading))\n",
    "assert not np.any(np.isnan(costmap))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "atomic-damages",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "================================\n",
      "data range:\n",
      "\n",
      "-1.37155 1.31044\n",
      "-1.34073 1.32024\n",
      "0.0 99.88889\n",
      "-0.22 0.22\n",
      "-2.77979 2.81163\n",
      "-0.99999 0.99999\n",
      "-0.99999 0.99999\n"
     ]
    }
   ],
   "source": [
    "print(\"================================\")\n",
    "print(\"data range:\\n\")\n",
    "print(planX.min().round(5), planX.max().round(5))\n",
    "print(planY.min().round(5), planY.max().round(5))\n",
    "print(costmap.min().round(5), costmap.max().round(5))\n",
    "for i in range(cmdVel.shape[1]):\n",
    "    print(cmdVel[:, i].min().round(5), cmdVel[:, i].max().round(5))\n",
    "for i in range(heading.shape[1]):\n",
    "    print(heading[:, i].min().round(5), heading[:, i].max().round(5))\n",
    "\n",
    "plan = np.stack([planX, planY], axis=2)\n",
    "\n",
    "heading = tf.convert_to_tensor(heading)\n",
    "cmdVel = tf.convert_to_tensor(cmdVel)\n",
    "plan = tf.convert_to_tensor(plan)\n",
    "costmap = tf.convert_to_tensor(costmap)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "capital-spine",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "================================\n",
      "input shape:\n",
      " (TensorShape([306746, 50, 2]), TensorShape([306746, 2]), TensorShape([306746, 20, 20]))\n",
      "output shape:\n",
      " (306746, 2)\n",
      "validation length: 15337\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(\"================================\")\n",
    "print(f\"input shape:\\n {plan.shape, heading.shape, costmap.shape}\\noutput shape:\\n {cmdVel.shape}\")\n",
    "\n",
    "validationLength = int(plan.shape[0]*0.05)\n",
    "print(f'validation length: {validationLength}\\n')\n",
    "\n",
    "planNormalizer = Normalization()\n",
    "headingNormalizer = Normalization()\n",
    "costmapNormalizer = Normalization()\n",
    "\n",
    "planNormalizer.adapt(plan)\n",
    "headingNormalizer.adapt(heading)\n",
    "costmapNormalizer.adapt(costmap)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "neither-basement",
   "metadata": {},
   "outputs": [],
   "source": [
    "planInput = Input((None, 2))\n",
    "headingInput = Input((2,))\n",
    "costmapInput = Input((20, 20, 1))\n",
    "\n",
    "planNormalized = planNormalizer(planInput)\n",
    "headingNormalized = headingNormalizer(headingInput)\n",
    "costmapNormalized = costmapNormalizer(costmapInput)\n",
    "\n",
    "planLatent = LSTM(32, input_shape=(None, 2), activation='tanh')(planNormalized)\n",
    "planLatent = Dense(16, activation='tanh')(planLatent)\n",
    "\n",
    "costmapLatent = Conv2D(16, (3, 3), activation=\"relu\",\n",
    "                       padding=\"same\", strides=(2, 2))(costmapNormalized)\n",
    "costmapLatent = Dropout(0.1)(costmapLatent)\n",
    "costmapLatent = Flatten()(costmapLatent)\n",
    "costmapLatent = Dense(8, activation=\"relu\")(costmapLatent)\n",
    "\n",
    "concatLayer = Concatenate()([planLatent, headingNormalized, costmapLatent])\n",
    "\n",
    "mergedLayer = Dense(16, activation='tanh')(concatLayer)\n",
    "mergedLayer = Dropout(0.1)(mergedLayer)\n",
    "mergedLayer = Dense(8, activation='tanh')(mergedLayer)\n",
    "output = Dense(2, activation='tanh')(mergedLayer)\n",
    "\n",
    "model = Model(inputs=[planInput, headingInput, costmapInput], outputs=output)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "handled-necklace",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "================================\n",
      "================================\n",
      "================================\n",
      "Model: \"model_1\"\n",
      "__________________________________________________________________________________________________\n",
      " Layer (type)                   Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      " input_6 (InputLayer)           [(None, 20, 20, 1)]  0           []                               \n",
      "                                                                                                  \n",
      " normalization_2 (Normalization  (None, 20, 20, 20)  41          ['input_6[0][0]']                \n",
      " )                                                                                                \n",
      "                                                                                                  \n",
      " input_4 (InputLayer)           [(None, None, 2)]    0           []                               \n",
      "                                                                                                  \n",
      " conv2d_1 (Conv2D)              (None, 10, 10, 16)   2896        ['normalization_2[1][0]']        \n",
      "                                                                                                  \n",
      " normalization (Normalization)  (None, None, 2)      5           ['input_4[0][0]']                \n",
      "                                                                                                  \n",
      " dropout_2 (Dropout)            (None, 10, 10, 16)   0           ['conv2d_1[0][0]']               \n",
      "                                                                                                  \n",
      " lstm_1 (LSTM)                  (None, 32)           4480        ['normalization[1][0]']          \n",
      "                                                                                                  \n",
      " input_5 (InputLayer)           [(None, 2)]          0           []                               \n",
      "                                                                                                  \n",
      " flatten_1 (Flatten)            (None, 1600)         0           ['dropout_2[0][0]']              \n",
      "                                                                                                  \n",
      " dense_5 (Dense)                (None, 16)           528         ['lstm_1[0][0]']                 \n",
      "                                                                                                  \n",
      " normalization_1 (Normalization  (None, 2)           5           ['input_5[0][0]']                \n",
      " )                                                                                                \n",
      "                                                                                                  \n",
      " dense_6 (Dense)                (None, 8)            12808       ['flatten_1[0][0]']              \n",
      "                                                                                                  \n",
      " concatenate_1 (Concatenate)    (None, 26)           0           ['dense_5[0][0]',                \n",
      "                                                                  'normalization_1[1][0]',        \n",
      "                                                                  'dense_6[0][0]']                \n",
      "                                                                                                  \n",
      " dense_7 (Dense)                (None, 16)           432         ['concatenate_1[0][0]']          \n",
      "                                                                                                  \n",
      " dropout_3 (Dropout)            (None, 16)           0           ['dense_7[0][0]']                \n",
      "                                                                                                  \n",
      " dense_8 (Dense)                (None, 8)            136         ['dropout_3[0][0]']              \n",
      "                                                                                                  \n",
      " dense_9 (Dense)                (None, 2)            18          ['dense_8[0][0]']                \n",
      "                                                                                                  \n",
      "==================================================================================================\n",
      "Total params: 21,349\n",
      "Trainable params: 21,298\n",
      "Non-trainable params: 51\n",
      "__________________________________________________________________________________________________\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.compile(loss='mse',\n",
    "              optimizer=tf.keras.optimizers.SGD(\n",
    "                  learning_rate=1e-2, decay=1e-6, momentum=0.9, nesterov=True),\n",
    "              metrics=['accuracy'])\n",
    "print(\"================================\")\n",
    "print(\"================================\")\n",
    "print(\"================================\")\n",
    "\n",
    "model.summary()\n",
    "tf.keras.utils.plot_model(model, to_file='model.png', show_shapes=True)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "pursuant-vinyl",
   "metadata": {},
   "outputs": [],
   "source": [
    "model_checkpoint_callback = tf.keras.callbacks.ModelCheckpoint(\n",
    "    filepath=f'{DATASET_ADDRESS}/checkpoint',\n",
    "    save_weights_only=False,\n",
    "    monitor='val_loss',\n",
    "    mode='min',\n",
    "    save_best_only=True)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "arctic-vault",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/150\n",
      "1821/1822 [============================>.] - ETA: 0s - loss: 0.1936 - accuracy: 0.5674"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:absl:Found untraced functions such as _jit_compiled_convolution_op, lstm_cell_1_layer_call_fn, lstm_cell_1_layer_call_and_return_conditional_losses while saving (showing 3 of 3). These functions will not be directly callable after loading.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: ../../bag/log_8/checkpoint/assets\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: ../../bag/log_8/checkpoint/assets\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1822/1822 [==============================] - 46s 25ms/step - loss: 0.1936 - accuracy: 0.5674 - val_loss: 0.2058 - val_accuracy: 0.5476\n",
      "Epoch 2/150\n",
      "1822/1822 [==============================] - 41s 22ms/step - loss: 0.1903 - accuracy: 0.5687 - val_loss: 0.2076 - val_accuracy: 0.5484\n",
      "Epoch 3/150\n",
      "1821/1822 [============================>.] - ETA: 0s - loss: 0.1880 - accuracy: 0.5711"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:absl:Found untraced functions such as _jit_compiled_convolution_op, lstm_cell_1_layer_call_fn, lstm_cell_1_layer_call_and_return_conditional_losses while saving (showing 3 of 3). These functions will not be directly callable after loading.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: ../../bag/log_8/checkpoint/assets\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: ../../bag/log_8/checkpoint/assets\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1822/1822 [==============================] - 44s 24ms/step - loss: 0.1880 - accuracy: 0.5712 - val_loss: 0.2032 - val_accuracy: 0.5607\n",
      "Epoch 4/150\n",
      "1822/1822 [==============================] - ETA: 0s - loss: 0.1835 - accuracy: 0.5758"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:absl:Found untraced functions such as _jit_compiled_convolution_op, lstm_cell_1_layer_call_fn, lstm_cell_1_layer_call_and_return_conditional_losses while saving (showing 3 of 3). These functions will not be directly callable after loading.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: ../../bag/log_8/checkpoint/assets\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: ../../bag/log_8/checkpoint/assets\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1822/1822 [==============================] - 44s 24ms/step - loss: 0.1835 - accuracy: 0.5758 - val_loss: 0.1976 - val_accuracy: 0.5763\n",
      "Epoch 5/150\n",
      "1821/1822 [============================>.] - ETA: 0s - loss: 0.1763 - accuracy: 0.6001"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:absl:Found untraced functions such as _jit_compiled_convolution_op, lstm_cell_1_layer_call_fn, lstm_cell_1_layer_call_and_return_conditional_losses while saving (showing 3 of 3). These functions will not be directly callable after loading.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: ../../bag/log_8/checkpoint/assets\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: ../../bag/log_8/checkpoint/assets\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1822/1822 [==============================] - 44s 24ms/step - loss: 0.1764 - accuracy: 0.6001 - val_loss: 0.1889 - val_accuracy: 0.6300\n",
      "Epoch 6/150\n",
      "1821/1822 [============================>.] - ETA: 0s - loss: 0.1659 - accuracy: 0.6457"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:absl:Found untraced functions such as _jit_compiled_convolution_op, lstm_cell_1_layer_call_fn, lstm_cell_1_layer_call_and_return_conditional_losses while saving (showing 3 of 3). These functions will not be directly callable after loading.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: ../../bag/log_8/checkpoint/assets\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: ../../bag/log_8/checkpoint/assets\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1822/1822 [==============================] - 44s 24ms/step - loss: 0.1659 - accuracy: 0.6457 - val_loss: 0.1733 - val_accuracy: 0.6803\n",
      "Epoch 7/150\n",
      "1820/1822 [============================>.] - ETA: 0s - loss: 0.1564 - accuracy: 0.6852"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:absl:Found untraced functions such as _jit_compiled_convolution_op, lstm_cell_1_layer_call_fn, lstm_cell_1_layer_call_and_return_conditional_losses while saving (showing 3 of 3). These functions will not be directly callable after loading.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: ../../bag/log_8/checkpoint/assets\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: ../../bag/log_8/checkpoint/assets\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1822/1822 [==============================] - 45s 24ms/step - loss: 0.1564 - accuracy: 0.6852 - val_loss: 0.1673 - val_accuracy: 0.7124\n",
      "Epoch 8/150\n",
      "1822/1822 [==============================] - ETA: 0s - loss: 0.1512 - accuracy: 0.7024"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:absl:Found untraced functions such as _jit_compiled_convolution_op, lstm_cell_1_layer_call_fn, lstm_cell_1_layer_call_and_return_conditional_losses while saving (showing 3 of 3). These functions will not be directly callable after loading.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: ../../bag/log_8/checkpoint/assets\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: ../../bag/log_8/checkpoint/assets\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1822/1822 [==============================] - 44s 24ms/step - loss: 0.1512 - accuracy: 0.7024 - val_loss: 0.1646 - val_accuracy: 0.7164\n",
      "Epoch 9/150\n",
      "1822/1822 [==============================] - ETA: 0s - loss: 0.1476 - accuracy: 0.7130"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:absl:Found untraced functions such as _jit_compiled_convolution_op, lstm_cell_1_layer_call_fn, lstm_cell_1_layer_call_and_return_conditional_losses while saving (showing 3 of 3). These functions will not be directly callable after loading.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: ../../bag/log_8/checkpoint/assets\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: ../../bag/log_8/checkpoint/assets\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1822/1822 [==============================] - 44s 24ms/step - loss: 0.1476 - accuracy: 0.7130 - val_loss: 0.1632 - val_accuracy: 0.7059\n",
      "Epoch 10/150\n",
      "1821/1822 [============================>.] - ETA: 0s - loss: 0.1444 - accuracy: 0.7212"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:absl:Found untraced functions such as _jit_compiled_convolution_op, lstm_cell_1_layer_call_fn, lstm_cell_1_layer_call_and_return_conditional_losses while saving (showing 3 of 3). These functions will not be directly callable after loading.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: ../../bag/log_8/checkpoint/assets\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: ../../bag/log_8/checkpoint/assets\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1822/1822 [==============================] - 44s 24ms/step - loss: 0.1444 - accuracy: 0.7212 - val_loss: 0.1623 - val_accuracy: 0.7105\n",
      "Epoch 11/150\n",
      "1821/1822 [============================>.] - ETA: 0s - loss: 0.1418 - accuracy: 0.7237"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:absl:Found untraced functions such as _jit_compiled_convolution_op, lstm_cell_1_layer_call_fn, lstm_cell_1_layer_call_and_return_conditional_losses while saving (showing 3 of 3). These functions will not be directly callable after loading.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: ../../bag/log_8/checkpoint/assets\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: ../../bag/log_8/checkpoint/assets\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1822/1822 [==============================] - 44s 24ms/step - loss: 0.1418 - accuracy: 0.7237 - val_loss: 0.1609 - val_accuracy: 0.7206\n",
      "Epoch 12/150\n",
      "1822/1822 [==============================] - 41s 22ms/step - loss: 0.1397 - accuracy: 0.7312 - val_loss: 0.1620 - val_accuracy: 0.7173\n",
      "Epoch 13/150\n",
      "1822/1822 [==============================] - 41s 23ms/step - loss: 0.1380 - accuracy: 0.7345 - val_loss: 0.1639 - val_accuracy: 0.7071\n",
      "Epoch 14/150\n",
      "1822/1822 [==============================] - ETA: 0s - loss: 0.1366 - accuracy: 0.7354"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:absl:Found untraced functions such as _jit_compiled_convolution_op, lstm_cell_1_layer_call_fn, lstm_cell_1_layer_call_and_return_conditional_losses while saving (showing 3 of 3). These functions will not be directly callable after loading.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: ../../bag/log_8/checkpoint/assets\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: ../../bag/log_8/checkpoint/assets\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1822/1822 [==============================] - 44s 24ms/step - loss: 0.1366 - accuracy: 0.7354 - val_loss: 0.1605 - val_accuracy: 0.7330\n",
      "Epoch 15/150\n",
      "1822/1822 [==============================] - 41s 23ms/step - loss: 0.1348 - accuracy: 0.7394 - val_loss: 0.1632 - val_accuracy: 0.7098\n",
      "Epoch 16/150\n",
      "1822/1822 [==============================] - 41s 23ms/step - loss: 0.1333 - accuracy: 0.7413 - val_loss: 0.1612 - val_accuracy: 0.7311\n",
      "Epoch 17/150\n",
      "1822/1822 [==============================] - ETA: 0s - loss: 0.1320 - accuracy: 0.7455"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:absl:Found untraced functions such as _jit_compiled_convolution_op, lstm_cell_1_layer_call_fn, lstm_cell_1_layer_call_and_return_conditional_losses while saving (showing 3 of 3). These functions will not be directly callable after loading.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: ../../bag/log_8/checkpoint/assets\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: ../../bag/log_8/checkpoint/assets\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1822/1822 [==============================] - 45s 24ms/step - loss: 0.1320 - accuracy: 0.7455 - val_loss: 0.1599 - val_accuracy: 0.7336\n",
      "Epoch 18/150\n",
      "1820/1822 [============================>.] - ETA: 0s - loss: 0.1309 - accuracy: 0.7473"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:absl:Found untraced functions such as _jit_compiled_convolution_op, lstm_cell_1_layer_call_fn, lstm_cell_1_layer_call_and_return_conditional_losses while saving (showing 3 of 3). These functions will not be directly callable after loading.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: ../../bag/log_8/checkpoint/assets\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: ../../bag/log_8/checkpoint/assets\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1822/1822 [==============================] - 45s 24ms/step - loss: 0.1308 - accuracy: 0.7473 - val_loss: 0.1594 - val_accuracy: 0.7523\n",
      "Epoch 19/150\n",
      "1822/1822 [==============================] - 41s 23ms/step - loss: 0.1296 - accuracy: 0.7497 - val_loss: 0.1621 - val_accuracy: 0.7247\n",
      "Epoch 20/150\n",
      "1822/1822 [==============================] - 41s 23ms/step - loss: 0.1282 - accuracy: 0.7516 - val_loss: 0.1634 - val_accuracy: 0.7267\n",
      "Epoch 21/150\n",
      "1822/1822 [==============================] - 41s 23ms/step - loss: 0.1274 - accuracy: 0.7535 - val_loss: 0.1631 - val_accuracy: 0.7267\n",
      "Epoch 22/150\n",
      "1822/1822 [==============================] - 41s 23ms/step - loss: 0.1264 - accuracy: 0.7541 - val_loss: 0.1630 - val_accuracy: 0.7385\n",
      "Epoch 23/150\n",
      "1822/1822 [==============================] - 41s 23ms/step - loss: 0.1254 - accuracy: 0.7576 - val_loss: 0.1612 - val_accuracy: 0.7361\n",
      "Epoch 24/150\n",
      "1820/1822 [============================>.] - ETA: 0s - loss: 0.1244 - accuracy: 0.7592"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:absl:Found untraced functions such as _jit_compiled_convolution_op, lstm_cell_1_layer_call_fn, lstm_cell_1_layer_call_and_return_conditional_losses while saving (showing 3 of 3). These functions will not be directly callable after loading.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: ../../bag/log_8/checkpoint/assets\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: ../../bag/log_8/checkpoint/assets\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1822/1822 [==============================] - 45s 24ms/step - loss: 0.1244 - accuracy: 0.7593 - val_loss: 0.1591 - val_accuracy: 0.7509\n",
      "Epoch 25/150\n",
      "1822/1822 [==============================] - 41s 22ms/step - loss: 0.1239 - accuracy: 0.7616 - val_loss: 0.1624 - val_accuracy: 0.7498\n",
      "Epoch 26/150\n",
      "1822/1822 [==============================] - 41s 23ms/step - loss: 0.1228 - accuracy: 0.7629 - val_loss: 0.1611 - val_accuracy: 0.7473\n",
      "Epoch 27/150\n",
      "1822/1822 [==============================] - 41s 23ms/step - loss: 0.1222 - accuracy: 0.7640 - val_loss: 0.1600 - val_accuracy: 0.7419\n",
      "Epoch 28/150\n",
      "1822/1822 [==============================] - 41s 22ms/step - loss: 0.1215 - accuracy: 0.7657 - val_loss: 0.1612 - val_accuracy: 0.7503\n",
      "Epoch 29/150\n",
      "1822/1822 [==============================] - 41s 23ms/step - loss: 0.1211 - accuracy: 0.7664 - val_loss: 0.1643 - val_accuracy: 0.7429\n",
      "Epoch 30/150\n",
      "1822/1822 [==============================] - 41s 23ms/step - loss: 0.1198 - accuracy: 0.7681 - val_loss: 0.1614 - val_accuracy: 0.7466\n",
      "Epoch 31/150\n",
      "1822/1822 [==============================] - 41s 23ms/step - loss: 0.1198 - accuracy: 0.7689 - val_loss: 0.1606 - val_accuracy: 0.7506\n",
      "Epoch 32/150\n",
      "1822/1822 [==============================] - 41s 23ms/step - loss: 0.1192 - accuracy: 0.7694 - val_loss: 0.1611 - val_accuracy: 0.7381\n",
      "Epoch 33/150\n",
      "1822/1822 [==============================] - 41s 22ms/step - loss: 0.1193 - accuracy: 0.7702 - val_loss: 0.1628 - val_accuracy: 0.7391\n",
      "Epoch 34/150\n",
      "1822/1822 [==============================] - 41s 22ms/step - loss: 0.1181 - accuracy: 0.7716 - val_loss: 0.1625 - val_accuracy: 0.7503\n",
      "Epoch 35/150\n",
      "1822/1822 [==============================] - 41s 22ms/step - loss: 0.1183 - accuracy: 0.7712 - val_loss: 0.1620 - val_accuracy: 0.7431\n",
      "Epoch 36/150\n",
      "1822/1822 [==============================] - 41s 22ms/step - loss: 0.1174 - accuracy: 0.7732 - val_loss: 0.1621 - val_accuracy: 0.7544\n",
      "Epoch 37/150\n",
      "1822/1822 [==============================] - 41s 22ms/step - loss: 0.1170 - accuracy: 0.7735 - val_loss: 0.1632 - val_accuracy: 0.7492\n",
      "Epoch 38/150\n",
      "1822/1822 [==============================] - 41s 22ms/step - loss: 0.1166 - accuracy: 0.7747 - val_loss: 0.1625 - val_accuracy: 0.7434\n",
      "Epoch 39/150\n",
      "1822/1822 [==============================] - 41s 22ms/step - loss: 0.1159 - accuracy: 0.7766 - val_loss: 0.1604 - val_accuracy: 0.7535\n",
      "Epoch 40/150\n",
      "1822/1822 [==============================] - 41s 23ms/step - loss: 0.1156 - accuracy: 0.7752 - val_loss: 0.1630 - val_accuracy: 0.7499\n",
      "Epoch 41/150\n",
      "1822/1822 [==============================] - 41s 23ms/step - loss: 0.1153 - accuracy: 0.7762 - val_loss: 0.1619 - val_accuracy: 0.7564\n",
      "Epoch 42/150\n",
      "1822/1822 [==============================] - 41s 22ms/step - loss: 0.1151 - accuracy: 0.7765 - val_loss: 0.1607 - val_accuracy: 0.7455\n",
      "Epoch 43/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1149 - accuracy: 0.7769 - val_loss: 0.1594 - val_accuracy: 0.7558\n",
      "Epoch 44/150\n",
      "1822/1822 [==============================] - 41s 23ms/step - loss: 0.1146 - accuracy: 0.7769 - val_loss: 0.1611 - val_accuracy: 0.7566\n",
      "Epoch 45/150\n",
      "1822/1822 [==============================] - 43s 24ms/step - loss: 0.1142 - accuracy: 0.7764 - val_loss: 0.1634 - val_accuracy: 0.7359\n",
      "Epoch 46/150\n",
      "1822/1822 [==============================] - 43s 24ms/step - loss: 0.1138 - accuracy: 0.7781 - val_loss: 0.1638 - val_accuracy: 0.7520\n",
      "Epoch 47/150\n",
      "1822/1822 [==============================] - 43s 24ms/step - loss: 0.1134 - accuracy: 0.7774 - val_loss: 0.1623 - val_accuracy: 0.7490\n",
      "Epoch 48/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1128 - accuracy: 0.7782 - val_loss: 0.1619 - val_accuracy: 0.7522\n",
      "Epoch 49/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1131 - accuracy: 0.7794 - val_loss: 0.1627 - val_accuracy: 0.7570\n",
      "Epoch 50/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1130 - accuracy: 0.7793 - val_loss: 0.1616 - val_accuracy: 0.7561\n",
      "Epoch 51/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1122 - accuracy: 0.7798 - val_loss: 0.1627 - val_accuracy: 0.7566\n",
      "Epoch 52/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1127 - accuracy: 0.7802 - val_loss: 0.1636 - val_accuracy: 0.7429\n",
      "Epoch 53/150\n",
      "1822/1822 [==============================] - 45s 25ms/step - loss: 0.1121 - accuracy: 0.7811 - val_loss: 0.1631 - val_accuracy: 0.7530\n",
      "Epoch 54/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1116 - accuracy: 0.7811 - val_loss: 0.1608 - val_accuracy: 0.7571\n",
      "Epoch 55/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1108 - accuracy: 0.7827 - val_loss: 0.1616 - val_accuracy: 0.7559\n",
      "Epoch 56/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1112 - accuracy: 0.7821 - val_loss: 0.1629 - val_accuracy: 0.7562\n",
      "Epoch 57/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1111 - accuracy: 0.7830 - val_loss: 0.1644 - val_accuracy: 0.7481\n",
      "Epoch 58/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1105 - accuracy: 0.7830 - val_loss: 0.1654 - val_accuracy: 0.7541\n",
      "Epoch 59/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1103 - accuracy: 0.7834 - val_loss: 0.1613 - val_accuracy: 0.7623\n",
      "Epoch 60/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1105 - accuracy: 0.7820 - val_loss: 0.1614 - val_accuracy: 0.7559\n",
      "Epoch 61/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1099 - accuracy: 0.7837 - val_loss: 0.1648 - val_accuracy: 0.7588\n",
      "Epoch 62/150\n",
      "1822/1822 [==============================] - 41s 23ms/step - loss: 0.1107 - accuracy: 0.7834 - val_loss: 0.1640 - val_accuracy: 0.7537\n",
      "Epoch 63/150\n",
      "1822/1822 [==============================] - 41s 23ms/step - loss: 0.1105 - accuracy: 0.7822 - val_loss: 0.1612 - val_accuracy: 0.7568\n",
      "Epoch 64/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1099 - accuracy: 0.7827 - val_loss: 0.1630 - val_accuracy: 0.7611\n",
      "Epoch 65/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1093 - accuracy: 0.7844 - val_loss: 0.1618 - val_accuracy: 0.7557\n",
      "Epoch 66/150\n",
      "1822/1822 [==============================] - 41s 23ms/step - loss: 0.1091 - accuracy: 0.7841 - val_loss: 0.1611 - val_accuracy: 0.7497\n",
      "Epoch 67/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1091 - accuracy: 0.7842 - val_loss: 0.1617 - val_accuracy: 0.7627\n",
      "Epoch 68/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1086 - accuracy: 0.7854 - val_loss: 0.1607 - val_accuracy: 0.7635\n",
      "Epoch 69/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1080 - accuracy: 0.7861 - val_loss: 0.1636 - val_accuracy: 0.7564\n",
      "Epoch 70/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1083 - accuracy: 0.7870 - val_loss: 0.1628 - val_accuracy: 0.7449\n",
      "Epoch 71/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1083 - accuracy: 0.7864 - val_loss: 0.1636 - val_accuracy: 0.7609\n",
      "Epoch 72/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1111 - accuracy: 0.7788 - val_loss: 0.1624 - val_accuracy: 0.7482\n",
      "Epoch 73/150\n",
      "1822/1822 [==============================] - 41s 23ms/step - loss: 0.1088 - accuracy: 0.7837 - val_loss: 0.1618 - val_accuracy: 0.7649\n",
      "Epoch 74/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1084 - accuracy: 0.7840 - val_loss: 0.1638 - val_accuracy: 0.7553\n",
      "Epoch 75/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1079 - accuracy: 0.7858 - val_loss: 0.1641 - val_accuracy: 0.7525\n",
      "Epoch 76/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1076 - accuracy: 0.7866 - val_loss: 0.1626 - val_accuracy: 0.7551\n",
      "Epoch 77/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1074 - accuracy: 0.7861 - val_loss: 0.1655 - val_accuracy: 0.7544\n",
      "Epoch 78/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1071 - accuracy: 0.7872 - val_loss: 0.1607 - val_accuracy: 0.7645\n",
      "Epoch 79/150\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1071 - accuracy: 0.7882 - val_loss: 0.1626 - val_accuracy: 0.7695\n",
      "Epoch 80/150\n",
      "1822/1822 [==============================] - 41s 23ms/step - loss: 0.1072 - accuracy: 0.7870 - val_loss: 0.1636 - val_accuracy: 0.7548\n",
      "Epoch 81/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1072 - accuracy: 0.7881 - val_loss: 0.1641 - val_accuracy: 0.7505\n",
      "Epoch 82/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1067 - accuracy: 0.7877 - val_loss: 0.1623 - val_accuracy: 0.7634\n",
      "Epoch 83/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1070 - accuracy: 0.7878 - val_loss: 0.1629 - val_accuracy: 0.7623\n",
      "Epoch 84/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1066 - accuracy: 0.7880 - val_loss: 0.1620 - val_accuracy: 0.7621\n",
      "Epoch 85/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1066 - accuracy: 0.7880 - val_loss: 0.1633 - val_accuracy: 0.7604\n",
      "Epoch 86/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1061 - accuracy: 0.7885 - val_loss: 0.1637 - val_accuracy: 0.7650\n",
      "Epoch 87/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1064 - accuracy: 0.7889 - val_loss: 0.1621 - val_accuracy: 0.7685\n",
      "Epoch 88/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1061 - accuracy: 0.7895 - val_loss: 0.1608 - val_accuracy: 0.7649\n",
      "Epoch 89/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1059 - accuracy: 0.7896 - val_loss: 0.1603 - val_accuracy: 0.7691\n",
      "Epoch 90/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1058 - accuracy: 0.7902 - val_loss: 0.1625 - val_accuracy: 0.7630\n",
      "Epoch 91/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1052 - accuracy: 0.7893 - val_loss: 0.1617 - val_accuracy: 0.7650\n",
      "Epoch 92/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1053 - accuracy: 0.7913 - val_loss: 0.1615 - val_accuracy: 0.7577\n",
      "Epoch 93/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1054 - accuracy: 0.7906 - val_loss: 0.1628 - val_accuracy: 0.7540\n",
      "Epoch 94/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1052 - accuracy: 0.7891 - val_loss: 0.1623 - val_accuracy: 0.7630\n",
      "Epoch 95/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1053 - accuracy: 0.7907 - val_loss: 0.1611 - val_accuracy: 0.7597\n",
      "Epoch 96/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1051 - accuracy: 0.7916 - val_loss: 0.1609 - val_accuracy: 0.7682\n",
      "Epoch 97/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1055 - accuracy: 0.7915 - val_loss: 0.1626 - val_accuracy: 0.7619\n",
      "Epoch 98/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1051 - accuracy: 0.7898 - val_loss: 0.1596 - val_accuracy: 0.7712\n",
      "Epoch 99/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1045 - accuracy: 0.7908 - val_loss: 0.1612 - val_accuracy: 0.7552\n",
      "Epoch 100/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1046 - accuracy: 0.7915 - val_loss: 0.1604 - val_accuracy: 0.7697\n",
      "Epoch 101/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1046 - accuracy: 0.7911 - val_loss: 0.1616 - val_accuracy: 0.7670\n",
      "Epoch 102/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1044 - accuracy: 0.7919 - val_loss: 0.1618 - val_accuracy: 0.7698\n",
      "Epoch 103/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1043 - accuracy: 0.7910 - val_loss: 0.1615 - val_accuracy: 0.7673\n",
      "Epoch 104/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1073 - accuracy: 0.7828 - val_loss: 0.1601 - val_accuracy: 0.7634\n",
      "Epoch 105/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1052 - accuracy: 0.7884 - val_loss: 0.1618 - val_accuracy: 0.7630\n",
      "Epoch 106/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1047 - accuracy: 0.7895 - val_loss: 0.1615 - val_accuracy: 0.7667\n",
      "Epoch 107/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1042 - accuracy: 0.7920 - val_loss: 0.1620 - val_accuracy: 0.7616\n",
      "Epoch 108/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1038 - accuracy: 0.7929 - val_loss: 0.1602 - val_accuracy: 0.7721\n",
      "Epoch 109/150\n",
      "1822/1822 [==============================] - ETA: 0s - loss: 0.1034 - accuracy: 0.7916"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:absl:Found untraced functions such as _jit_compiled_convolution_op, lstm_cell_1_layer_call_fn, lstm_cell_1_layer_call_and_return_conditional_losses while saving (showing 3 of 3). These functions will not be directly callable after loading.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: ../../bag/log_8/checkpoint/assets\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: ../../bag/log_8/checkpoint/assets\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1822/1822 [==============================] - 48s 26ms/step - loss: 0.1034 - accuracy: 0.7916 - val_loss: 0.1568 - val_accuracy: 0.7723\n",
      "Epoch 110/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1032 - accuracy: 0.7924 - val_loss: 0.1627 - val_accuracy: 0.7568\n",
      "Epoch 111/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1050 - accuracy: 0.7877 - val_loss: 0.1626 - val_accuracy: 0.7540\n",
      "Epoch 112/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1048 - accuracy: 0.7885 - val_loss: 0.1622 - val_accuracy: 0.7646\n",
      "Epoch 113/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1047 - accuracy: 0.7915 - val_loss: 0.1653 - val_accuracy: 0.7556\n",
      "Epoch 114/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1045 - accuracy: 0.7925 - val_loss: 0.1631 - val_accuracy: 0.7608\n",
      "Epoch 115/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1041 - accuracy: 0.7917 - val_loss: 0.1620 - val_accuracy: 0.7685\n",
      "Epoch 116/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1036 - accuracy: 0.7927 - val_loss: 0.1619 - val_accuracy: 0.7664\n",
      "Epoch 117/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1080 - accuracy: 0.7774 - val_loss: 0.1629 - val_accuracy: 0.7572\n",
      "Epoch 118/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1053 - accuracy: 0.7853 - val_loss: 0.1610 - val_accuracy: 0.7696\n",
      "Epoch 119/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1045 - accuracy: 0.7893 - val_loss: 0.1638 - val_accuracy: 0.7557\n",
      "Epoch 120/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1042 - accuracy: 0.7900 - val_loss: 0.1616 - val_accuracy: 0.7681\n",
      "Epoch 121/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1038 - accuracy: 0.7921 - val_loss: 0.1590 - val_accuracy: 0.7704\n",
      "Epoch 122/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1037 - accuracy: 0.7912 - val_loss: 0.1638 - val_accuracy: 0.7560\n",
      "Epoch 123/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1033 - accuracy: 0.7927 - val_loss: 0.1637 - val_accuracy: 0.7691\n",
      "Epoch 124/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1029 - accuracy: 0.7931 - val_loss: 0.1618 - val_accuracy: 0.7671\n",
      "Epoch 125/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1032 - accuracy: 0.7917 - val_loss: 0.1618 - val_accuracy: 0.7582\n",
      "Epoch 126/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1032 - accuracy: 0.7930 - val_loss: 0.1646 - val_accuracy: 0.7712\n",
      "Epoch 127/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1024 - accuracy: 0.7958 - val_loss: 0.1620 - val_accuracy: 0.7700\n",
      "Epoch 128/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1022 - accuracy: 0.7949 - val_loss: 0.1628 - val_accuracy: 0.7693\n",
      "Epoch 129/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1029 - accuracy: 0.7940 - val_loss: 0.1629 - val_accuracy: 0.7671\n",
      "Epoch 130/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1021 - accuracy: 0.7965 - val_loss: 0.1621 - val_accuracy: 0.7766\n",
      "Epoch 131/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1017 - accuracy: 0.7962 - val_loss: 0.1618 - val_accuracy: 0.7670\n",
      "Epoch 132/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1020 - accuracy: 0.7963 - val_loss: 0.1605 - val_accuracy: 0.7704\n",
      "Epoch 133/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1020 - accuracy: 0.7956 - val_loss: 0.1605 - val_accuracy: 0.7727\n",
      "Epoch 134/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1021 - accuracy: 0.7963 - val_loss: 0.1599 - val_accuracy: 0.7733\n",
      "Epoch 135/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1019 - accuracy: 0.7966 - val_loss: 0.1627 - val_accuracy: 0.7664\n",
      "Epoch 136/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1013 - accuracy: 0.7972 - val_loss: 0.1632 - val_accuracy: 0.7650\n",
      "Epoch 137/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1014 - accuracy: 0.7966 - val_loss: 0.1611 - val_accuracy: 0.7630\n",
      "Epoch 138/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1012 - accuracy: 0.7970 - val_loss: 0.1612 - val_accuracy: 0.7700\n",
      "Epoch 139/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1013 - accuracy: 0.7972 - val_loss: 0.1621 - val_accuracy: 0.7677\n",
      "Epoch 140/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1016 - accuracy: 0.7966 - val_loss: 0.1614 - val_accuracy: 0.7662\n",
      "Epoch 141/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1011 - accuracy: 0.7977 - val_loss: 0.1634 - val_accuracy: 0.7648\n",
      "Epoch 142/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1014 - accuracy: 0.7961 - val_loss: 0.1608 - val_accuracy: 0.7682\n",
      "Epoch 143/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1010 - accuracy: 0.7967 - val_loss: 0.1617 - val_accuracy: 0.7718\n",
      "Epoch 144/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1006 - accuracy: 0.7975 - val_loss: 0.1618 - val_accuracy: 0.7644\n",
      "Epoch 145/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1003 - accuracy: 0.7979 - val_loss: 0.1628 - val_accuracy: 0.7737\n",
      "Epoch 146/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1002 - accuracy: 0.7980 - val_loss: 0.1610 - val_accuracy: 0.7653\n",
      "Epoch 147/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1003 - accuracy: 0.7988 - val_loss: 0.1625 - val_accuracy: 0.7732\n",
      "Epoch 148/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1026 - accuracy: 0.7934 - val_loss: 0.1626 - val_accuracy: 0.7568\n",
      "Epoch 149/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1017 - accuracy: 0.7945 - val_loss: 0.1611 - val_accuracy: 0.7669\n",
      "Epoch 150/150\n",
      "1822/1822 [==============================] - 42s 23ms/step - loss: 0.1013 - accuracy: 0.7968 - val_loss: 0.1611 - val_accuracy: 0.7659\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x7fe485b00a00>"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.fit(\n",
    "    x=[plan[:-validationLength],\n",
    "      heading[:-validationLength],\n",
    "      costmap[:-validationLength]],\n",
    "    y=cmdVel[:-validationLength],\n",
    "    epochs=150,\n",
    "    batch_size=128,\n",
    "    validation_split=0.2,\n",
    "    callbacks=[model_checkpoint_callback]\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "worse-foster",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "120/120 [==============================] - 1s 8ms/step - loss: 0.1683 - accuracy: 0.7806\n",
      "================================\n",
      "evaluation accurecy: 78%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:absl:Found untraced functions such as _jit_compiled_convolution_op, lstm_cell_1_layer_call_fn, lstm_cell_1_layer_call_and_return_conditional_losses while saving (showing 3 of 3). These functions will not be directly callable after loading.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: ../../bag/log_8/deepMPCModel78/assets\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: ../../bag/log_8/deepMPCModel78/assets\n"
     ]
    }
   ],
   "source": [
    "q = model.evaluate(x=[plan[-validationLength:], heading[-validationLength:], costmap[-validationLength:]],\n",
    "                   y=cmdVel[-validationLength:],\n",
    "                   batch_size=128)\n",
    "q = int(np.array(q).round(2)[1]*100)\n",
    "print(\"================================\")\n",
    "print(f\"evaluation accurecy: {q}%\")\n",
    "\n",
    "address = f\"{DATASET_ADDRESS}/deepMPCModel{q}\"\n",
    "model.save(filepath=address,\n",
    "           overwrite=True, include_optimizer=True)\n",
    "model = load_model(address)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "optional-tactics",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 1s 1s/step\n",
      "sample predicted output: [[ 0.20998025 -0.02068261]]\n"
     ]
    }
   ],
   "source": [
    "i = random.randint(0, plan.shape[0]-1)\n",
    "vel = model.predict([plan[None, i], heading[None, i], costmap[None, i]])\n",
    "print(f\"sample predicted output: {vel}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ready-testament",
   "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": "3.8.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}