conv3d_lstm_model.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "source": [
    "import numpy as np\r\n",
    "import os\r\n",
    "import imutils\r\n",
    "import dlib\r\n",
    "import cv2\r\n",
    "\r\n",
    "import imageio\r\n",
    "from imutils import face_utils\r\n",
    "\r\n",
    "\r\n",
    "import time\r\n",
    "import os"
   ],
   "outputs": [],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "source": [
    "people = ['F01','F02','F04','F05','F06','F07','F08','F09','F10','F11','M01','M02','M04','M07','M08']\r\n",
    "folder_enum = ['01','02','03','04','05','06','07','08', '09', '10']\r\n",
    "instances = ['01','02','03','04','05','06','07','08', '09', '10']\r\n",
    "data_types = ['words']\r\n",
    "words = ['Begin', 'Choose', 'Connection', 'Navigation', 'Next', 'Previous', 'Start', 'Stop', 'Hello', 'Web']\r\n",
    "\r\n",
    "import pickle\r\n",
    "\r\n",
    "from keras.utils import np_utils, generic_utils\r\n",
    "\r\n",
    "# WITH ONE HOT ENCODING \r\n",
    "\r\n",
    "with open('pickle_train_test_ohe/X_train.pkl', 'rb') as f:\r\n",
    "   X_train = pickle.load(f)\r\n",
    "\r\n",
    "with open('pickle_train_test_ohe/y_train.pkl', 'rb') as f:\r\n",
    "   y_train = pickle.load(f)\r\n",
    "\r\n",
    "with open('pickle_train_test_ohe/X_test.pkl', 'rb') as f:\r\n",
    "   X_test = pickle.load(f)\r\n",
    "\r\n",
    "with open('pickle_train_test_ohe/y_test.pkl', 'rb') as f:\r\n",
    "   y_test = pickle.load(f)\r\n",
    "\r\n",
    "with open('pickle_train_test_ohe/X_val.pkl', 'rb') as f:\r\n",
    "   X_val = pickle.load(f)\r\n",
    "\r\n",
    "with open('pickle_train_test_ohe/y_val.pkl', 'rb') as f:\r\n",
    "   y_val = pickle.load(f)\r\n",
    "\r\n",
    "# WITHOUT ONE HOT ENCODING \r\n",
    "\"\"\"\r\n",
    "with open('pickle_train_test/X_train.pkl', 'rb') as f:\r\n",
    "   X_train = pickle.load(f)\r\n",
    "\r\n",
    "with open('pickle_train_test/y_train.pkl', 'rb') as f:\r\n",
    "   y_train = pickle.load(f)\r\n",
    "\r\n",
    "with open('pickle_train_test/X_test.pkl', 'rb') as f:\r\n",
    "   X_test = pickle.load(f)\r\n",
    "\r\n",
    "with open('pickle_train_test/y_test.pkl', 'rb') as f:\r\n",
    "   y_test = pickle.load(f)\r\n",
    "\r\n",
    "with open('pickle_train_test/X_val.pkl', 'rb') as f:\r\n",
    "   X_val = pickle.load(f)\r\n",
    "\r\n",
    "with open('pickle_train_test/y_val.pkl', 'rb') as f:\r\n",
    "   y_val = pickle.load(f)\r\n",
    "\"\"\"\r\n",
    "\r\n",
    "print(X_train.shape)\r\n",
    "print(y_train.shape)\r\n",
    "print(X_test.shape)\r\n",
    "print(y_test.shape)\r\n",
    "print(X_val.shape)\r\n",
    "print(y_val.shape)"
   ],
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": [
      "(1200, 22, 100, 100, 1)\n",
      "(1200, 10)\n",
      "(150, 22, 100, 100, 1)\n",
      "(150, 10)\n",
      "(150, 22, 100, 100, 1)\n",
      "(150, 10)\n"
     ]
    }
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "source": [
    "X_train[0].shape"
   ],
   "outputs": [
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": [
       "(22, 100, 100, 1)"
      ]
     },
     "metadata": {},
     "execution_count": 3
    }
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "source": [
    "y_train.shape"
   ],
   "outputs": [
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": [
       "(1200, 10)"
      ]
     },
     "metadata": {},
     "execution_count": 4
    }
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "source": [
    "\"\"\"from sklearn.preprocessing import LabelEncoder\r\n",
    "le = LabelEncoder()\r\n",
    "le.fit(y_train)\r\n",
    "y_train_enc = le.transform(y_train)\r\n",
    "#y_test_enc = le.transform(y_test)\"\"\""
   ],
   "outputs": [
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": [
       "'from sklearn.preprocessing import LabelEncoder\\nle = LabelEncoder()\\nle.fit(y_train)\\ny_train_enc = le.transform(y_train)\\n#y_test_enc = le.transform(y_test)'"
      ]
     },
     "metadata": {},
     "execution_count": 5
    }
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "source": [
    "from tensorflow.keras.layers import Conv3D, MaxPooling3D\r\n",
    "from tensorflow.keras.layers import Dense, Dropout, Flatten\r\n",
    "from tensorflow.keras.models import Sequential\r\n",
    "from tensorflow.keras.layers import Activation, ZeroPadding3D, TimeDistributed, LSTM, GRU, Reshape\r\n",
    "from tensorflow.keras.utils import plot_model\r\n",
    "from tensorflow.keras.layers import BatchNormalization\r\n",
    "import matplotlib.pyplot as plt\r\n",
    "\r\n",
    "from tensorflow import keras\r\n"
   ],
   "outputs": [],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "source": [
    "y_train[0]"
   ],
   "outputs": [
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": [
       "array([0., 0., 0., 0., 1., 0., 0., 0., 0., 0.])"
      ]
     },
     "metadata": {},
     "execution_count": 7
    }
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "source": [
    "\"\"\"from keras.utils import to_categorical\r\n",
    "\r\n",
    "y_train_1 = to_categorical(y_train,num_classes = 10, dtype =\"float32\")\r\n",
    "\"\"\""
   ],
   "outputs": [
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": [
       "'from keras.utils import to_categorical\\n\\ny_train_1 = to_categorical(y_train,num_classes = 10, dtype =\"float32\")\\n'"
      ]
     },
     "metadata": {},
     "execution_count": 8
    }
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "source": [
    "#y_train_1.dtype"
   ],
   "outputs": [],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "source": [
    "#CNN + LSTM\r\n",
    "\r\n",
    "model = Sequential()\r\n",
    "\r\n",
    "# 1st layer group\r\n",
    "model.add(Conv3D(32, (3, 3, 3), strides = 1, input_shape=(22, 100, 100, 1), activation='relu', padding='valid'))\r\n",
    "\r\n",
    "model.add(MaxPooling3D(pool_size=(2, 2, 2), strides=2))\r\n",
    "\r\n",
    "model.add(Conv3D(64, (3, 3, 3), activation='relu', strides=1))\r\n",
    "model.add(MaxPooling3D(pool_size=(2, 2, 2), strides=2))\r\n",
    "\r\n",
    "model.add(Conv3D(128, (3, 3, 3), activation='relu', strides=1))\r\n",
    "model.add(MaxPooling3D(pool_size=(2, 2, 2), strides=2))\r\n",
    "\r\n",
    "model.add(Reshape((128, 1*10*10)))  \r\n",
    "\r\n",
    "# LSTMS - Recurrent Network Layer\r\n",
    "model.add(LSTM(32, return_sequences=True))\r\n",
    "model.add(Dropout(.5))\r\n",
    "\r\n",
    "model.add((Flatten()))\r\n",
    "\r\n",
    "# # FC layers group\r\n",
    "model.add(Dense(2048, activation='relu'))\r\n",
    "model.add(Dropout(.5))\r\n",
    "model.add(Dense(1024, activation='relu'))\r\n",
    "model.add(Dropout(.5))\r\n",
    "\r\n",
    "model.add(Dense(10, activation='softmax'))\r\n",
    "\r\n",
    "model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])\r\n",
    "\r\n",
    "model.summary()\r\n",
    "\r\n"
   ],
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv3d (Conv3D)              (None, 20, 98, 98, 32)    896       \n",
      "_________________________________________________________________\n",
      "max_pooling3d (MaxPooling3D) (None, 10, 49, 49, 32)    0         \n",
      "_________________________________________________________________\n",
      "conv3d_1 (Conv3D)            (None, 8, 47, 47, 64)     55360     \n",
      "_________________________________________________________________\n",
      "max_pooling3d_1 (MaxPooling3 (None, 4, 23, 23, 64)     0         \n",
      "_________________________________________________________________\n",
      "conv3d_2 (Conv3D)            (None, 2, 21, 21, 128)    221312    \n",
      "_________________________________________________________________\n",
      "max_pooling3d_2 (MaxPooling3 (None, 1, 10, 10, 128)    0         \n",
      "_________________________________________________________________\n",
      "reshape (Reshape)            (None, 128, 100)          0         \n",
      "_________________________________________________________________\n",
      "lstm (LSTM)                  (None, 128, 32)           17024     \n",
      "_________________________________________________________________\n",
      "dropout (Dropout)            (None, 128, 32)           0         \n",
      "_________________________________________________________________\n",
      "flatten (Flatten)            (None, 4096)              0         \n",
      "_________________________________________________________________\n",
      "dense (Dense)                (None, 2048)              8390656   \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 2048)              0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 1024)              2098176   \n",
      "_________________________________________________________________\n",
      "dropout_2 (Dropout)          (None, 1024)              0         \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 10)                10250     \n",
      "=================================================================\n",
      "Total params: 10,793,674\n",
      "Trainable params: 10,793,674\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "source": [
    "'''import visualkeras\r\n",
    "from PIL import ImageFont\r\n",
    "\r\n",
    "font = ImageFont.truetype(\"arial.ttf\", 28, encoding=\"unic\")'''"
   ],
   "outputs": [
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": [
       "'import visualkeras\\nfrom PIL import ImageFont\\n\\nfont = ImageFont.truetype(\"arial.ttf\", 28, encoding=\"unic\")'"
      ]
     },
     "metadata": {},
     "execution_count": 11
    }
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "source": [
    "#visualkeras.layered_view(model, legend=True, font=font) "
   ],
   "outputs": [],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "source": [
    "#plot_model(model, to_file='Vis_speech_model_plot.png', show_shapes=True, show_layer_names=True)\r\n"
   ],
   "outputs": [],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "source": [
    "from tensorflow.keras.callbacks import EarlyStopping\r\n",
    "\r\n",
    "early_stop = EarlyStopping(monitor=\"val_loss\", patience=4)\r\n",
    "\r\n",
    "t1 = time.time()\r\n",
    "history = model.fit(X_train,y_train, validation_data=(X_val, y_val), epochs=35)\r\n",
    "t2 = time.time()\r\n",
    "print()\r\n",
    "print(f\"Training time : {t2 - t1} secs.\")\r\n"
   ],
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": [
      "Epoch 1/35\n",
      "38/38 [==============================] - 975s 26s/step - loss: 2.3174 - accuracy: 0.0620 - val_loss: 2.2953 - val_accuracy: 0.1000\n",
      "Epoch 2/35\n"
     ]
    }
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "source": [
    "\"\"\"Training and validation accuracy\"\"\"\r\n",
    "\r\n",
    "# Plot training & validation accuracy values\r\n",
    "plt.plot(history.history['accuracy'])\r\n",
    "plt.plot(history.history['val_accuracy'])\r\n",
    "plt.title('Model accuracy')\r\n",
    "plt.ylabel('Accuracy')\r\n",
    "plt.xlabel('Epoch')\r\n",
    "plt.legend(['Train', 'Validation'], loc='upper left')\r\n",
    "plt.show()\r\n",
    "\r\n",
    "# Plot training & validation loss values\r\n",
    "plt.plot(history.history['loss'])\r\n",
    "plt.plot(history.history['val_loss'])\r\n",
    "plt.title('Model loss')\r\n",
    "plt.ylabel('Loss')\r\n",
    "plt.xlabel('Epoch')\r\n",
    "#plt.xlim(1, 15)\r\n",
    "# plt.ylim(0, 3)\r\n",
    "plt.legend(['Train', 'Validation'], loc='upper left')\r\n",
    "plt.show()\r\n",
    "\r\n"
   ],
   "outputs": [
    {
     "output_type": "error",
     "ename": "Error",
     "evalue": "Kernel is dead",
     "traceback": [
      "Error: Kernel is dead",
      "at g._sendKernelShellControl (c:\\Users\\mprad\\.vscode\\extensions\\ms-toolsai.jupyter-2021.8.1054968649\\out\\client\\extension.js:52:852938)",
      "at g.sendShellMessage (c:\\Users\\mprad\\.vscode\\extensions\\ms-toolsai.jupyter-2021.8.1054968649\\out\\client\\extension.js:52:852707)",
      "at g.requestExecute (c:\\Users\\mprad\\.vscode\\extensions\\ms-toolsai.jupyter-2021.8.1054968649\\out\\client\\extension.js:52:855249)",
      "at d.requestExecute (c:\\Users\\mprad\\.vscode\\extensions\\ms-toolsai.jupyter-2021.8.1054968649\\out\\client\\extension.js:37:313885)",
      "at w.requestExecute (c:\\Users\\mprad\\.vscode\\extensions\\ms-toolsai.jupyter-2021.8.1054968649\\out\\client\\extension.js:24:134487)",
      "at w.executeCodeCell (c:\\Users\\mprad\\.vscode\\extensions\\ms-toolsai.jupyter-2021.8.1054968649\\out\\client\\extension.js:90:319834)",
      "at w.execute (c:\\Users\\mprad\\.vscode\\extensions\\ms-toolsai.jupyter-2021.8.1054968649\\out\\client\\extension.js:90:319389)",
      "at w.start (c:\\Users\\mprad\\.vscode\\extensions\\ms-toolsai.jupyter-2021.8.1054968649\\out\\client\\extension.js:90:315205)",
      "at runMicrotasks (<anonymous>)",
      "at processTicksAndRejections (internal/process/task_queues.js:93:5)",
      "at async t.CellExecutionQueue.executeQueuedCells (c:\\Users\\mprad\\.vscode\\extensions\\ms-toolsai.jupyter-2021.8.1054968649\\out\\client\\extension.js:90:329732)",
      "at async t.CellExecutionQueue.start (c:\\Users\\mprad\\.vscode\\extensions\\ms-toolsai.jupyter-2021.8.1054968649\\out\\client\\extension.js:90:329272)"
     ]
    }
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "source": [
    "model.save('complete_saved_model_35e')\r\n"
   ],
   "outputs": [
    {
     "output_type": "stream",
     "name": "stderr",
     "text": [
      "WARNING:absl:Found untraced functions such as lstm_cell_3_layer_call_fn, lstm_cell_3_layer_call_and_return_conditional_losses, lstm_cell_3_layer_call_fn, lstm_cell_3_layer_call_and_return_conditional_losses, lstm_cell_3_layer_call_and_return_conditional_losses while saving (showing 5 of 5). These functions will not be directly callable after loading.\n",
      "WARNING:absl:Found untraced functions such as lstm_cell_3_layer_call_fn, lstm_cell_3_layer_call_and_return_conditional_losses, lstm_cell_3_layer_call_fn, lstm_cell_3_layer_call_and_return_conditional_losses, lstm_cell_3_layer_call_and_return_conditional_losses while saving (showing 5 of 5). These functions will not be directly callable after loading.\n"
     ]
    },
    {
     "output_type": "stream",
     "name": "stdout",
     "text": [
      "INFO:tensorflow:Assets written to: complete_saved_model_30e\\assets\n"
     ]
    },
    {
     "output_type": "stream",
     "name": "stderr",
     "text": [
      "INFO:tensorflow:Assets written to: complete_saved_model_30e\\assets\n"
     ]
    }
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "source": [
    "\"\"\"#CNN + LSTM\r\n",
    "\r\n",
    "model = Sequential()\r\n",
    "\r\n",
    "#model.add(Flatten(input_shape=(X_train.shape)))\r\n",
    "# 1st layer group\r\n",
    "model.add(Conv3D(32, (3, 3, 3), strides = 1, input_shape=(22, 100, 100, 1), activation='relu', padding='valid'))\r\n",
    "#model.add(Conv3D(32, (3, 3, 3), strides = 1,input_shape= X_train.shape, activation='relu', padding='valid'))\r\n",
    "\r\n",
    "model.add(MaxPooling3D(pool_size=(2, 2, 2), strides=2))\r\n",
    "\r\n",
    "model.add(Conv3D(64, (3, 3, 3), activation='relu', strides=1))\r\n",
    "model.add(MaxPooling3D(pool_size=(2, 2, 2), strides=2))\r\n",
    "\r\n",
    "model.add(Conv3D(128, (3, 3, 3), activation='relu', strides=1))\r\n",
    "model.add(MaxPooling3D(pool_size=(2, 2, 2), strides=2))\r\n",
    "#(None, 1, 10, 10, 128)\r\n",
    "#shape = model.get_output_shape_at(0)\r\n",
    "#shape=model.layers[0].compute_output_shape((22, 100, 100, 1))\r\n",
    "#model.add(Reshape((shape[-1],shape[1]*shape[2]*shape[3])))\r\n",
    "\r\n",
    "model.add(Reshape((128, 1*10*10)))  \r\n",
    "\r\n",
    "# LSTMS - Recurrent Network Layer\r\n",
    "model.add(LSTM(32, return_sequences=True))\r\n",
    "model.add(Dropout(.5))\r\n",
    "\r\n",
    "model.add((Flatten()))\r\n",
    "\r\n",
    "# # FC layers group\r\n",
    "model.add(Dense(2048, activation='relu'))\r\n",
    "model.add(Dropout(.5))\r\n",
    "model.add(Dense(1024, activation='relu'))\r\n",
    "model.add(Dropout(.5))\r\n",
    "\r\n",
    "#model.add((Flatten()))\r\n",
    "#model.add(Dense(100, activation='relu'))\r\n",
    "\r\n",
    "model.add(Dense(10, activation='softmax'))\r\n",
    "\r\n",
    "model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])\r\n",
    "\r\n",
    "model.summary()\"\"\""
   ],
   "outputs": [
    {
     "output_type": "error",
     "ename": "SyntaxError",
     "evalue": "invalid syntax (<ipython-input-17-f2b438463ee6>, line 1)",
     "traceback": [
      "\u001b[1;36m  File \u001b[1;32m\"<ipython-input-17-f2b438463ee6>\"\u001b[1;36m, line \u001b[1;32m1\u001b[0m\n\u001b[1;33m    $\u001b[0m\n\u001b[1;37m    ^\u001b[0m\n\u001b[1;31mSyntaxError\u001b[0m\u001b[1;31m:\u001b[0m invalid syntax\n"
     ]
    }
   ],
   "metadata": {}
  },
  {
   "cell_type": "markdown",
   "source": [
    "# Testing Saved Model on Test Set"
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "source": [
    "from tensorflow import keras\r\n",
    "model_2 = keras.models.load_model('F:/MIRACLE-V1/ALR/complete_saved_model_30e/')"
   ],
   "outputs": [],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "source": [
    "train_loss, train_acc = model_2.evaluate(X_train,y_train)\r\n",
    "val_loss, val_acc   = model_2.evaluate(X_val,y_val)\r\n",
    "test_loss, test_acc   = model_2.evaluate(X_test,y_test)\r\n",
    "\r\n",
    "print(\"final train accuracy = {:.2f} , validation accuracy = {:.2f} , Test accuracy = {:.2f} \".format(train_acc*100, val_acc*100,test_acc*100))"
   ],
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": [
      "38/38 [==============================] - 155s 4s/step - loss: 0.0234 - accuracy: 0.9950\n",
      "5/5 [==============================] - 19s 4s/step - loss: 0.5918 - accuracy: 0.8467\n",
      "5/5 [==============================] - 24s 5s/step - loss: 0.4110 - accuracy: 0.9133\n",
      "final train accuracy = 99.50 , validation accuracy = 84.67 , Test accuracy = 91.33 \n"
     ]
    }
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "source": [
    "words = ['Begin', 'Choose', 'Connection', 'Navigation', 'Next', 'Previous', 'Start', 'Stop', 'Hello', 'Web']\r\n"
   ],
   "outputs": [],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "source": [
    "ypred = model_2.predict(X_test)\r\n",
    "\r\n",
    "ypred[0]\r\n",
    "\r\n",
    "predicted_words = [words[i] for i in np.argmax(ypred, axis=1)]\r\n",
    "actual_words = [words[i] for i in np.argmax(y_test, axis=1)]\r\n",
    "\r\n",
    "correct = 0\r\n",
    "for p, a in zip(predicted_words, actual_words):\r\n",
    "    if p == a:\r\n",
    "        correct += 1\r\n",
    "#     print(f\"Predicted : {p} \\t Actual : {a}\")\r\n",
    "\r\n",
    "accuracy = correct/len(actual_words)\r\n",
    "print(f\"Accuracy = {accuracy} on completely unseen data\")\r\n"
   ],
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": [
      "Accuracy = 0.9133333333333333 on completely unseen data\n"
     ]
    }
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "source": [
    "model_2.evaluate(X_test, y_test)\r\n",
    "\r\n",
    "model_2.metrics_names\r\n",
    "\r\n",
    "my_data = X_test[0]\r\n",
    "my_data.shape\r\n",
    "\r\n",
    "# my_data\r\n",
    "\r\n",
    "X_train[0].shape\r\n",
    "\r\n",
    "my_data = my_data.reshape(1,22,100,100,1)\r\n",
    "\r\n",
    "ans = model_2.predict(my_data)\r\n",
    "np.argmax(ans,)\r\n",
    "\r\n",
    "y_test[0]\r\n"
   ],
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": [
      "5/5 [==============================] - 17s 3s/step - loss: 0.4110 - accuracy: 0.9133\n"
     ]
    },
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": [
       "array([0., 0., 0., 0., 1., 0., 0., 0., 0., 0.])"
      ]
     },
     "metadata": {},
     "execution_count": 18
    }
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "source": [
    "from sklearn import preprocessing\r\n",
    "le = preprocessing.LabelEncoder()"
   ],
   "outputs": [],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "source": [
    "actual_words_le=le.fit_transform(actual_words)\r\n",
    "predicted_words_le=le.transform(predicted_words)"
   ],
   "outputs": [],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "source": [
    "import pandas as pd"
   ],
   "outputs": [],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "source": [
    "from sklearn.metrics import classification_report\r\n",
    "\r\n",
    "target_names = words\r\n",
    "\r\n",
    "clsf_report = pd.DataFrame(classification_report(y_true = actual_words_le, y_pred = predicted_words_le, \r\n",
    "                                                target_names=target_names,output_dict=True)).transpose()"
   ],
   "outputs": [],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "source": [
    "clsf_report.style.background_gradient(cmap='viridis',)"
   ],
   "outputs": [
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": [
       "<pandas.io.formats.style.Styler at 0x25fe888f2e0>"
      ],
      "text/html": [
       "<style  type=\"text/css\" >\n",
       "#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow0_col0,#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow8_col0{\n",
       "            background-color:  #365d8d;\n",
       "            color:  #f1f1f1;\n",
       "        }#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow0_col1,#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow3_col1,#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow8_col1{\n",
       "            background-color:  #35b779;\n",
       "            color:  #000000;\n",
       "        }#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow0_col2,#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow8_col2{\n",
       "            background-color:  #32658e;\n",
       "            color:  #f1f1f1;\n",
       "        }#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow0_col3,#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow1_col3,#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow2_col3,#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow3_col3,#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow4_col3,#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow5_col3,#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow6_col3,#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow7_col3,#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow8_col3,#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow9_col3{\n",
       "            background-color:  #482374;\n",
       "            color:  #f1f1f1;\n",
       "        }#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow1_col0,#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow5_col0{\n",
       "            background-color:  #22a785;\n",
       "            color:  #000000;\n",
       "        }#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow1_col1,#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow5_col1,#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow7_col0{\n",
       "            background-color:  #31688e;\n",
       "            color:  #f1f1f1;\n",
       "        }#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow1_col2,#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow5_col2{\n",
       "            background-color:  #38598c;\n",
       "            color:  #f1f1f1;\n",
       "        }#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow2_col0{\n",
       "            background-color:  #2eb37c;\n",
       "            color:  #000000;\n",
       "        }#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow2_col1,#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow4_col0,#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow4_col1,#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow4_col2,#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow7_col1,#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow9_col0,#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow11_col3,#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow12_col3{\n",
       "            background-color:  #fde725;\n",
       "            color:  #000000;\n",
       "        }#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow2_col2{\n",
       "            background-color:  #6ccd5a;\n",
       "            color:  #000000;\n",
       "        }#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow3_col0,#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow6_col1,#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow6_col2,#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow9_col1,#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow10_col3{\n",
       "            background-color:  #440154;\n",
       "            color:  #f1f1f1;\n",
       "        }#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow3_col2{\n",
       "            background-color:  #472c7a;\n",
       "            color:  #f1f1f1;\n",
       "        }#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow6_col0,#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow7_col2{\n",
       "            background-color:  #1fa088;\n",
       "            color:  #000000;\n",
       "        }#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow9_col2{\n",
       "            background-color:  #3e4989;\n",
       "            color:  #f1f1f1;\n",
       "        }#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow10_col0{\n",
       "            background-color:  #20928c;\n",
       "            color:  #000000;\n",
       "        }#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow10_col1,#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow11_col1,#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow12_col1{\n",
       "            background-color:  #1fa188;\n",
       "            color:  #000000;\n",
       "        }#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow10_col2{\n",
       "            background-color:  #2a768e;\n",
       "            color:  #f1f1f1;\n",
       "        }#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow11_col0,#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow12_col0{\n",
       "            background-color:  #1f988b;\n",
       "            color:  #000000;\n",
       "        }#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow11_col2,#T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow12_col2{\n",
       "            background-color:  #2b758e;\n",
       "            color:  #f1f1f1;\n",
       "        }</style><table id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292c\" ><thead>    <tr>        <th class=\"blank level0\" ></th>        <th class=\"col_heading level0 col0\" >precision</th>        <th class=\"col_heading level0 col1\" >recall</th>        <th class=\"col_heading level0 col2\" >f1-score</th>        <th class=\"col_heading level0 col3\" >support</th>    </tr></thead><tbody>\n",
       "                <tr>\n",
       "                        <th id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292clevel0_row0\" class=\"row_heading level0 row0\" >Begin</th>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow0_col0\" class=\"data row0 col0\" >0.875000</td>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow0_col1\" class=\"data row0 col1\" >0.933333</td>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow0_col2\" class=\"data row0 col2\" >0.903226</td>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow0_col3\" class=\"data row0 col3\" >15.000000</td>\n",
       "            </tr>\n",
       "            <tr>\n",
       "                        <th id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292clevel0_row1\" class=\"row_heading level0 row1\" >Choose</th>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow1_col0\" class=\"data row1 col0\" >0.928571</td>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow1_col1\" class=\"data row1 col1\" >0.866667</td>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow1_col2\" class=\"data row1 col2\" >0.896552</td>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow1_col3\" class=\"data row1 col3\" >15.000000</td>\n",
       "            </tr>\n",
       "            <tr>\n",
       "                        <th id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292clevel0_row2\" class=\"row_heading level0 row2\" >Connection</th>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow2_col0\" class=\"data row2 col0\" >0.937500</td>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow2_col1\" class=\"data row2 col1\" >1.000000</td>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow2_col2\" class=\"data row2 col2\" >0.967742</td>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow2_col3\" class=\"data row2 col3\" >15.000000</td>\n",
       "            </tr>\n",
       "            <tr>\n",
       "                        <th id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292clevel0_row3\" class=\"row_heading level0 row3\" >Navigation</th>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow3_col0\" class=\"data row3 col0\" >0.823529</td>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow3_col1\" class=\"data row3 col1\" >0.933333</td>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow3_col2\" class=\"data row3 col2\" >0.875000</td>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow3_col3\" class=\"data row3 col3\" >15.000000</td>\n",
       "            </tr>\n",
       "            <tr>\n",
       "                        <th id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292clevel0_row4\" class=\"row_heading level0 row4\" >Next</th>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow4_col0\" class=\"data row4 col0\" >1.000000</td>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow4_col1\" class=\"data row4 col1\" >1.000000</td>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow4_col2\" class=\"data row4 col2\" >1.000000</td>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow4_col3\" class=\"data row4 col3\" >15.000000</td>\n",
       "            </tr>\n",
       "            <tr>\n",
       "                        <th id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292clevel0_row5\" class=\"row_heading level0 row5\" >Previous</th>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow5_col0\" class=\"data row5 col0\" >0.928571</td>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow5_col1\" class=\"data row5 col1\" >0.866667</td>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow5_col2\" class=\"data row5 col2\" >0.896552</td>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow5_col3\" class=\"data row5 col3\" >15.000000</td>\n",
       "            </tr>\n",
       "            <tr>\n",
       "                        <th id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292clevel0_row6\" class=\"row_heading level0 row6\" >Start</th>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow6_col0\" class=\"data row6 col0\" >0.923077</td>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow6_col1\" class=\"data row6 col1\" >0.800000</td>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow6_col2\" class=\"data row6 col2\" >0.857143</td>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow6_col3\" class=\"data row6 col3\" >15.000000</td>\n",
       "            </tr>\n",
       "            <tr>\n",
       "                        <th id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292clevel0_row7\" class=\"row_heading level0 row7\" >Stop</th>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow7_col0\" class=\"data row7 col0\" >0.882353</td>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow7_col1\" class=\"data row7 col1\" >1.000000</td>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow7_col2\" class=\"data row7 col2\" >0.937500</td>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow7_col3\" class=\"data row7 col3\" >15.000000</td>\n",
       "            </tr>\n",
       "            <tr>\n",
       "                        <th id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292clevel0_row8\" class=\"row_heading level0 row8\" >Hello</th>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow8_col0\" class=\"data row8 col0\" >0.875000</td>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow8_col1\" class=\"data row8 col1\" >0.933333</td>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow8_col2\" class=\"data row8 col2\" >0.903226</td>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow8_col3\" class=\"data row8 col3\" >15.000000</td>\n",
       "            </tr>\n",
       "            <tr>\n",
       "                        <th id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292clevel0_row9\" class=\"row_heading level0 row9\" >Web</th>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow9_col0\" class=\"data row9 col0\" >1.000000</td>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow9_col1\" class=\"data row9 col1\" >0.800000</td>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow9_col2\" class=\"data row9 col2\" >0.888889</td>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow9_col3\" class=\"data row9 col3\" >15.000000</td>\n",
       "            </tr>\n",
       "            <tr>\n",
       "                        <th id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292clevel0_row10\" class=\"row_heading level0 row10\" >accuracy</th>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow10_col0\" class=\"data row10 col0\" >0.913333</td>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow10_col1\" class=\"data row10 col1\" >0.913333</td>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow10_col2\" class=\"data row10 col2\" >0.913333</td>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow10_col3\" class=\"data row10 col3\" >0.913333</td>\n",
       "            </tr>\n",
       "            <tr>\n",
       "                        <th id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292clevel0_row11\" class=\"row_heading level0 row11\" >macro avg</th>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow11_col0\" class=\"data row11 col0\" >0.917360</td>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow11_col1\" class=\"data row11 col1\" >0.913333</td>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow11_col2\" class=\"data row11 col2\" >0.912583</td>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow11_col3\" class=\"data row11 col3\" >150.000000</td>\n",
       "            </tr>\n",
       "            <tr>\n",
       "                        <th id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292clevel0_row12\" class=\"row_heading level0 row12\" >weighted avg</th>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow12_col0\" class=\"data row12 col0\" >0.917360</td>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow12_col1\" class=\"data row12 col1\" >0.913333</td>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow12_col2\" class=\"data row12 col2\" >0.912583</td>\n",
       "                        <td id=\"T_d17e4ef6_cf58_11ec_b79d_8c47be1b292crow12_col3\" class=\"data row12 col3\" >150.000000</td>\n",
       "            </tr>\n",
       "    </tbody></table>"
      ]
     },
     "metadata": {},
     "execution_count": 23
    }
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "source": [
    "import seaborn as sns\r\n",
    "import matplotlib.pyplot as plt  "
   ],
   "outputs": [],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "source": [
    "from sklearn.metrics import confusion_matrix\r\n",
    "\r\n",
    "conf_matrix=confusion_matrix(actual_words_le, predicted_words_le)\r\n",
    "\r\n",
    "# Creating a dataframe for a array-formatted Confusion matrix,so it will be easy for plotting.\r\n",
    "cm_df = pd.DataFrame(conf_matrix,\r\n",
    "                     index = words, \r\n",
    "                     columns = words)\r\n",
    "#Plotting the confusion matrix\r\n",
    "plt.figure(figsize=(10,10))\r\n",
    "sns.heatmap(cm_df, annot=True)\r\n",
    "plt.title('Confusion Matrix for Uttered Word ( Epochs= 30 )')\r\n",
    "plt.ylabel('Actal Values')\r\n",
    "plt.xlabel('Predicted Values')\r\n",
    "plt.show()"
   ],
   "outputs": [
    {
     "output_type": "display_data",
     "data": {
      "text/plain": [
       "<Figure size 720x720 with 2 Axes>"
      ],
      "image/png": "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"
     },
     "metadata": {
      "needs_background": "light"
     }
    }
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "source": [
    "# Commented out IPython magic to ensure Python compatibility.\r\n",
    "#Model Saving\r\n",
    "# %pwd\r\n",
    "\r\n",
    "# model.save_weights('saved_model/')\r\n",
    "\r\n",
    "#model.save('complete_saved_model_15e')"
   ],
   "outputs": [
    {
     "output_type": "stream",
     "name": "stderr",
     "text": [
      "WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses, lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 5 of 5). These functions will not be directly callable after loading.\n",
      "WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses, lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 5 of 5). These functions will not be directly callable after loading.\n"
     ]
    },
    {
     "output_type": "stream",
     "name": "stdout",
     "text": [
      "INFO:tensorflow:Assets written to: complete_saved_model_10e\\assets\n"
     ]
    },
    {
     "output_type": "stream",
     "name": "stderr",
     "text": [
      "INFO:tensorflow:Assets written to: complete_saved_model_10e\\assets\n"
     ]
    }
   ],
   "metadata": {}
  }
 ],
 "metadata": {
  "orig_nbformat": 4,
  "language_info": {
   "name": "python",
   "version": "3.8.5",
   "mimetype": "text/x-python",
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "pygments_lexer": "ipython3",
   "nbconvert_exporter": "python",
   "file_extension": ".py"
  },
  "kernelspec": {
   "name": "python3",
   "display_name": "Python 3.8.5 64-bit (conda)"
  },
  "interpreter": {
   "hash": "b64de0ba76e769a7c3088015f258876bdb4240a199c36c507f14d2fb10bd5d74"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}