bidirection-lstm.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    " # \n",
    " https://analyticsindiamag.com/complete-guide-to-bidirectional-lstm-with-python-codes/\n",
    "\n",
    "ANN - artificial neural network\n",
    "CNN - \n",
    "\n",
    "\n",
    "RNN recurrent neural network\n",
    "* uses feedback loops\n",
    "* creates \"memory\"\n",
    "* independent variable becomes dependent in next layer\n",
    "* used for speech recognition, language modeling, image classification\n",
    "* standard RNN contains a single layer\n",
    "\n",
    "LSTM has more complex repeating structure\n",
    "* four layers\n",
    "    - Neural net\n",
    "    - Poitwise Operation\n",
    "    - Vector transformer\n",
    "    - Concatenate\n",
    "    - Copy\n",
    "\n",
    "* BI-LSTM\n",
    "    - sequence info in both dirs (future & past)\n",
    "    - input flows two directions.\n",
    "    \n",
    "\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Defaulting to user installation because normal site-packages is not writeable\n",
      "Requirement already satisfied: tensorflow in /home/brianh/.local/lib/python3.10/site-packages (2.9.1)\n",
      "Requirement already satisfied: opt-einsum>=2.3.2 in /home/brianh/.local/lib/python3.10/site-packages (from tensorflow) (3.3.0)\n",
      "Requirement already satisfied: tensorflow-estimator<2.10.0,>=2.9.0rc0 in /home/brianh/.local/lib/python3.10/site-packages (from tensorflow) (2.9.0)\n",
      "Requirement already satisfied: tensorflow-io-gcs-filesystem>=0.23.1 in /home/brianh/.local/lib/python3.10/site-packages (from tensorflow) (0.26.0)\n",
      "Requirement already satisfied: protobuf<3.20,>=3.9.2 in /home/brianh/.local/lib/python3.10/site-packages (from tensorflow) (3.19.4)\n",
      "Requirement already satisfied: keras-preprocessing>=1.1.1 in /home/brianh/.local/lib/python3.10/site-packages (from tensorflow) (1.1.2)\n",
      "Requirement already satisfied: google-pasta>=0.1.1 in /home/brianh/.local/lib/python3.10/site-packages (from tensorflow) (0.2.0)\n",
      "Requirement already satisfied: gast<=0.4.0,>=0.2.1 in /home/brianh/.local/lib/python3.10/site-packages (from tensorflow) (0.4.0)\n",
      "Requirement already satisfied: termcolor>=1.1.0 in /home/brianh/.local/lib/python3.10/site-packages (from tensorflow) (1.1.0)\n",
      "Requirement already satisfied: setuptools in /usr/lib/python3/dist-packages (from tensorflow) (59.6.0)\n",
      "Requirement already satisfied: wrapt>=1.11.0 in /home/brianh/.local/lib/python3.10/site-packages (from tensorflow) (1.14.1)\n",
      "Requirement already satisfied: absl-py>=1.0.0 in /home/brianh/.local/lib/python3.10/site-packages (from tensorflow) (1.1.0)\n",
      "Requirement already satisfied: packaging in /usr/lib/python3/dist-packages (from tensorflow) (21.3)\n",
      "Requirement already satisfied: grpcio<2.0,>=1.24.3 in /home/brianh/.local/lib/python3.10/site-packages (from tensorflow) (1.46.3)\n",
      "Requirement already satisfied: libclang>=13.0.0 in /home/brianh/.local/lib/python3.10/site-packages (from tensorflow) (14.0.1)\n",
      "Requirement already satisfied: six>=1.12.0 in /usr/lib/python3/dist-packages (from tensorflow) (1.16.0)\n",
      "Requirement already satisfied: keras<2.10.0,>=2.9.0rc0 in /home/brianh/.local/lib/python3.10/site-packages (from tensorflow) (2.9.0)\n",
      "Requirement already satisfied: tensorboard<2.10,>=2.9 in /home/brianh/.local/lib/python3.10/site-packages (from tensorflow) (2.9.1)\n",
      "Requirement already satisfied: typing-extensions>=3.6.6 in /home/brianh/.local/lib/python3.10/site-packages (from tensorflow) (4.2.0)\n",
      "Requirement already satisfied: astunparse>=1.6.0 in /home/brianh/.local/lib/python3.10/site-packages (from tensorflow) (1.6.3)\n",
      "Requirement already satisfied: flatbuffers<2,>=1.12 in /home/brianh/.local/lib/python3.10/site-packages (from tensorflow) (1.12)\n",
      "Requirement already satisfied: h5py>=2.9.0 in /home/brianh/.local/lib/python3.10/site-packages (from tensorflow) (3.7.0)\n",
      "Requirement already satisfied: numpy>=1.20 in /home/brianh/.local/lib/python3.10/site-packages (from tensorflow) (1.22.4)\n",
      "Requirement already satisfied: wheel<1.0,>=0.23.0 in /usr/lib/python3/dist-packages (from astunparse>=1.6.0->tensorflow) (0.37.1)\n",
      "Requirement already satisfied: google-auth<3,>=1.6.3 in /home/brianh/.local/lib/python3.10/site-packages (from tensorboard<2.10,>=2.9->tensorflow) (2.7.0)\n",
      "Requirement already satisfied: google-auth-oauthlib<0.5,>=0.4.1 in /home/brianh/.local/lib/python3.10/site-packages (from tensorboard<2.10,>=2.9->tensorflow) (0.4.6)\n",
      "Requirement already satisfied: markdown>=2.6.8 in /usr/lib/python3/dist-packages (from tensorboard<2.10,>=2.9->tensorflow) (3.3.6)\n",
      "Requirement already satisfied: tensorboard-data-server<0.7.0,>=0.6.0 in /home/brianh/.local/lib/python3.10/site-packages (from tensorboard<2.10,>=2.9->tensorflow) (0.6.1)\n",
      "Requirement already satisfied: tensorboard-plugin-wit>=1.6.0 in /home/brianh/.local/lib/python3.10/site-packages (from tensorboard<2.10,>=2.9->tensorflow) (1.8.1)\n",
      "Requirement already satisfied: requests<3,>=2.21.0 in /usr/lib/python3/dist-packages (from tensorboard<2.10,>=2.9->tensorflow) (2.25.1)\n",
      "Requirement already satisfied: werkzeug>=1.0.1 in /home/brianh/.local/lib/python3.10/site-packages (from tensorboard<2.10,>=2.9->tensorflow) (2.1.2)\n",
      "Requirement already satisfied: rsa<5,>=3.1.4 in /home/brianh/.local/lib/python3.10/site-packages (from google-auth<3,>=1.6.3->tensorboard<2.10,>=2.9->tensorflow) (4.8)\n",
      "Requirement already satisfied: pyasn1-modules>=0.2.1 in /usr/lib/python3/dist-packages (from google-auth<3,>=1.6.3->tensorboard<2.10,>=2.9->tensorflow) (0.2.1)\n",
      "Requirement already satisfied: cachetools<6.0,>=2.0.0 in /home/brianh/.local/lib/python3.10/site-packages (from google-auth<3,>=1.6.3->tensorboard<2.10,>=2.9->tensorflow) (5.2.0)\n",
      "Requirement already satisfied: requests-oauthlib>=0.7.0 in /home/brianh/.local/lib/python3.10/site-packages (from google-auth-oauthlib<0.5,>=0.4.1->tensorboard<2.10,>=2.9->tensorflow) (1.3.1)\n",
      "Requirement already satisfied: oauthlib>=3.0.0 in /usr/lib/python3/dist-packages (from requests-oauthlib>=0.7.0->google-auth-oauthlib<0.5,>=0.4.1->tensorboard<2.10,>=2.9->tensorflow) (3.2.0)\n",
      "Requirement already satisfied: pyasn1>=0.1.3 in /usr/lib/python3/dist-packages (from rsa<5,>=3.1.4->google-auth<3,>=1.6.3->tensorboard<2.10,>=2.9->tensorflow) (0.4.8)\n",
      "Defaulting to user installation because normal site-packages is not writeable\n",
      "Requirement already satisfied: keras in /home/brianh/.local/lib/python3.10/site-packages (2.9.0)\n",
      "Defaulting to user installation because normal site-packages is not writeable\n",
      "Requirement already satisfied: keras.preprocessing in /home/brianh/.local/lib/python3.10/site-packages (1.1.2)\n",
      "Requirement already satisfied: numpy>=1.9.1 in /home/brianh/.local/lib/python3.10/site-packages (from keras.preprocessing) (1.22.4)\n",
      "Requirement already satisfied: six>=1.9.0 in /usr/lib/python3/dist-packages (from keras.preprocessing) (1.16.0)\n",
      "Defaulting to user installation because normal site-packages is not writeable\n",
      "Requirement already satisfied: keras.models in /home/brianh/.local/lib/python3.10/site-packages (0.0.7)\n",
      "Requirement already satisfied: Pillow in /home/brianh/.local/lib/python3.10/site-packages (from keras.models) (9.1.1)\n",
      "Requirement already satisfied: keras in /home/brianh/.local/lib/python3.10/site-packages (from keras.models) (2.9.0)\n",
      "Requirement already satisfied: opencv-python in /home/brianh/.local/lib/python3.10/site-packages (from keras.models) (4.5.5.64)\n",
      "Requirement already satisfied: spacy in /home/brianh/.local/lib/python3.10/site-packages (from keras.models) (3.3.1)\n",
      "Requirement already satisfied: pathlib in /home/brianh/.local/lib/python3.10/site-packages (from keras.models) (1.0.1)\n",
      "Requirement already satisfied: numpy in /home/brianh/.local/lib/python3.10/site-packages (from keras.models) (1.22.4)\n",
      "Requirement already satisfied: pydantic!=1.8,!=1.8.1,<1.9.0,>=1.7.4 in /home/brianh/.local/lib/python3.10/site-packages (from spacy->keras.models) (1.8.2)\n",
      "Requirement already satisfied: blis<0.8.0,>=0.4.0 in /home/brianh/.local/lib/python3.10/site-packages (from spacy->keras.models) (0.7.7)\n",
      "Requirement already satisfied: langcodes<4.0.0,>=3.2.0 in /home/brianh/.local/lib/python3.10/site-packages (from spacy->keras.models) (3.3.0)\n",
      "Requirement already satisfied: murmurhash<1.1.0,>=0.28.0 in /home/brianh/.local/lib/python3.10/site-packages (from spacy->keras.models) (1.0.7)\n",
      "Requirement already satisfied: spacy-loggers<2.0.0,>=1.0.0 in /home/brianh/.local/lib/python3.10/site-packages (from spacy->keras.models) (1.0.2)\n",
      "Requirement already satisfied: wasabi<1.1.0,>=0.9.1 in /home/brianh/.local/lib/python3.10/site-packages (from spacy->keras.models) (0.9.1)\n",
      "Requirement already satisfied: thinc<8.1.0,>=8.0.14 in /home/brianh/.local/lib/python3.10/site-packages (from spacy->keras.models) (8.0.17)\n",
      "Requirement already satisfied: requests<3.0.0,>=2.13.0 in /usr/lib/python3/dist-packages (from spacy->keras.models) (2.25.1)\n",
      "Requirement already satisfied: cymem<2.1.0,>=2.0.2 in /home/brianh/.local/lib/python3.10/site-packages (from spacy->keras.models) (2.0.6)\n",
      "Requirement already satisfied: spacy-legacy<3.1.0,>=3.0.9 in /home/brianh/.local/lib/python3.10/site-packages (from spacy->keras.models) (3.0.9)\n",
      "Requirement already satisfied: jinja2 in /usr/lib/python3/dist-packages (from spacy->keras.models) (3.0.3)\n",
      "Requirement already satisfied: packaging>=20.0 in /usr/lib/python3/dist-packages (from spacy->keras.models) (21.3)\n",
      "Requirement already satisfied: srsly<3.0.0,>=2.4.3 in /home/brianh/.local/lib/python3.10/site-packages (from spacy->keras.models) (2.4.3)\n",
      "Requirement already satisfied: setuptools in /usr/lib/python3/dist-packages (from spacy->keras.models) (59.6.0)\n",
      "Requirement already satisfied: pathy>=0.3.5 in /home/brianh/.local/lib/python3.10/site-packages (from spacy->keras.models) (0.6.1)\n",
      "Requirement already satisfied: catalogue<2.1.0,>=2.0.6 in /home/brianh/.local/lib/python3.10/site-packages (from spacy->keras.models) (2.0.7)\n",
      "Requirement already satisfied: tqdm<5.0.0,>=4.38.0 in /home/brianh/.local/lib/python3.10/site-packages (from spacy->keras.models) (4.64.0)\n",
      "Requirement already satisfied: preshed<3.1.0,>=3.0.2 in /home/brianh/.local/lib/python3.10/site-packages (from spacy->keras.models) (3.0.6)\n",
      "Requirement already satisfied: typer<0.5.0,>=0.3.0 in /home/brianh/.local/lib/python3.10/site-packages (from spacy->keras.models) (0.4.1)\n",
      "Requirement already satisfied: smart-open<6.0.0,>=5.0.0 in /home/brianh/.local/lib/python3.10/site-packages (from pathy>=0.3.5->spacy->keras.models) (5.2.1)\n",
      "Requirement already satisfied: typing-extensions>=3.7.4.3 in /home/brianh/.local/lib/python3.10/site-packages (from pydantic!=1.8,!=1.8.1,<1.9.0,>=1.7.4->spacy->keras.models) (4.2.0)\n",
      "Requirement already satisfied: click<9.0.0,>=7.1.1 in /usr/lib/python3/dist-packages (from typer<0.5.0,>=0.3.0->spacy->keras.models) (8.0.3)\n",
      "Defaulting to user installation because normal site-packages is not writeable\n",
      "\u001b[31mERROR: Could not find a version that satisfies the requirement keras.layers (from versions: none)\u001b[0m\u001b[31m\n",
      "\u001b[0m\u001b[31mERROR: No matching distribution found for keras.layers\u001b[0m\u001b[31m\n",
      "\u001b[0mDefaulting to user installation because normal site-packages is not writeable\n",
      "Requirement already satisfied: keras.datasets in /home/brianh/.local/lib/python3.10/site-packages (0.1.0)\n",
      "Defaulting to user installation because normal site-packages is not writeable\n",
      "Collecting keras.utils\n",
      "  Downloading keras-utils-1.0.13.tar.gz (2.4 kB)\n",
      "  Preparing metadata (setup.py) ... \u001b[?25ldone\n",
      "\u001b[?25hRequirement already satisfied: Keras>=2.1.5 in /home/brianh/.local/lib/python3.10/site-packages (from keras.utils) (2.9.0)\n",
      "Building wheels for collected packages: keras.utils\n",
      "  Building wheel for keras.utils (setup.py) ... \u001b[?25ldone\n",
      "\u001b[?25h  Created wheel for keras.utils: filename=keras_utils-1.0.13-py3-none-any.whl size=2656 sha256=60156838251a2a70637cb2bc8590b5a9c478107580393f722d4726abc0c6c17b\n",
      "  Stored in directory: /home/brianh/.cache/pip/wheels/5c/c0/b3/0c332de4fd71f3733ea6d61697464b7ae4b2b5ff0300e6ca7a\n",
      "Successfully built keras.utils\n",
      "Installing collected packages: keras.utils\n",
      "Successfully installed keras.utils-1.0.13\n"
     ]
    }
   ],
   "source": [
    "#!/bin/sh\n",
    "\n",
    "!pip3 install tensorflow\n",
    "\n",
    "# https://matplotlib.org/stable/users/installing/index.html\n",
    "!pip3 install matplotlib\n",
    "\n",
    "!pip3 install keras\n",
    "!pip3 install keras.preprocessing\n",
    "!pip3 install keras.models\n",
    "!pip3 install keras.layers \n",
    "!pip3 install keras.datasets\n",
    "!pip3 install keras.utils\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    " # https://analyticsindiamag.com/complete-guide-to-bidirectional-lstm-with-python-codes/\n",
    " \n",
    " import numpy as np\n",
    " from keras.preprocessing import sequence\n",
    " from keras.models import Sequential\n",
    " from keras.layers import Dense, Dropout, Embedding, LSTM, Bidirectional\n",
    " from keras.datasets import imdb \n",
    " from keras.utils.data_utils import pad_sequences\n",
    "\n",
    "\n",
    " "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    " n_unique_words = 10000 # cut texts after this number of words\n",
    " maxlen = 200\n",
    " batch_size = 128 \n",
    "\n",
    " "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "(x_train, y_train),(x_test, y_test) = imdb.load_data(num_words=n_unique_words)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "convert data into matrices\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    " x_train = keras.utils.data_utils.pad_sequences(x_train, maxlen=maxlen)\n",
    " x_test = keras.utils.data_utils.pad_sequences(x_test, maxlen=maxlen)\n",
    " y_train = np.array(y_train)\n",
    " y_test = np.array(y_test) "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* conda install -c conda-forge cudatoolkit=11.2 cudnn=8.1.0\n",
    "* export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$CONDA_PREFIX/lib/\n",
    "* python3 -m pip install tensorflow\n",
    "\n",
    "python3 -c \"import tensorflow as tf; print(tf.config.list_physical_devices('GPU'))\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "wget https://developer.download.nvidia.com/compute/cuda/repos/wsl-ubuntu/x86_64/cuda-wsl-ubuntu.pin\n",
    "sudo mv cuda-wsl-ubuntu.pin /etc/apt/preferences.d/cuda-repository-pin-600\n",
    "sudo apt-key adv --fetch-keys https://developer.download.nvidia.com/compute/cuda/repos/wsl-ubuntu/x86_64/3bf863cc.pub\n",
    "sudo add-apt-repository \"deb https://developer.download.nvidia.com/compute/cuda/repos/wsl-ubuntu/x86_64/ /\"\n",
    "sudo apt-get update\n",
    "sudo apt-get -y install cuda"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "create a bi-lstm\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    " model = Sequential()\n",
    " model.add(Embedding(n_unique_words, 128, input_length=maxlen))\n",
    " model.add(Bidirectional(LSTM(64)))\n",
    " model.add(Dropout(0.5))\n",
    " model.add(Dense(1, activation='sigmoid'))\n",
    " model.compile(loss='binary_crossentropy', optimizer='adam', metrics=['accuracy']) "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "fit the model with data that we loaded"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/12\n",
      "196/196 [==============================] - 129s 633ms/step - loss: 0.4789 - accuracy: 0.7570 - val_loss: 0.3159 - val_accuracy: 0.8666\n",
      "Epoch 2/12\n",
      "196/196 [==============================] - 124s 636ms/step - loss: 0.2473 - accuracy: 0.9047 - val_loss: 0.3220 - val_accuracy: 0.8710\n",
      "Epoch 3/12\n",
      "196/196 [==============================] - 162s 828ms/step - loss: 0.1858 - accuracy: 0.9334 - val_loss: 0.3554 - val_accuracy: 0.8663\n",
      "Epoch 4/12\n",
      "196/196 [==============================] - 152s 778ms/step - loss: 0.1367 - accuracy: 0.9521 - val_loss: 0.3912 - val_accuracy: 0.8635\n",
      "Epoch 5/12\n",
      "196/196 [==============================] - 154s 788ms/step - loss: 0.1063 - accuracy: 0.9639 - val_loss: 0.3932 - val_accuracy: 0.8574\n",
      "Epoch 6/12\n",
      "196/196 [==============================] - 174s 891ms/step - loss: 0.0939 - accuracy: 0.9677 - val_loss: 0.4324 - val_accuracy: 0.8559\n",
      "Epoch 7/12\n",
      "196/196 [==============================] - 180s 919ms/step - loss: 0.0916 - accuracy: 0.9692 - val_loss: 0.4440 - val_accuracy: 0.8541\n",
      "Epoch 8/12\n",
      "196/196 [==============================] - 165s 842ms/step - loss: 0.0633 - accuracy: 0.9792 - val_loss: 0.5421 - val_accuracy: 0.8434\n",
      "Epoch 9/12\n",
      "196/196 [==============================] - 153s 781ms/step - loss: 0.0488 - accuracy: 0.9850 - val_loss: 0.5677 - val_accuracy: 0.8473\n",
      "Epoch 10/12\n",
      "196/196 [==============================] - 144s 738ms/step - loss: 0.0434 - accuracy: 0.9864 - val_loss: 0.6773 - val_accuracy: 0.8539\n",
      "Epoch 11/12\n",
      "196/196 [==============================] - 129s 656ms/step - loss: 0.0516 - accuracy: 0.9817 - val_loss: 0.5790 - val_accuracy: 0.8535\n",
      "Epoch 12/12\n",
      "196/196 [==============================] - 124s 631ms/step - loss: 0.0362 - accuracy: 0.9896 - val_loss: 0.6458 - val_accuracy: 0.8312\n",
      "[0.4789074957370758, 0.2472827285528183, 0.18583106994628906, 0.13666650652885437, 0.10632037371397018, 0.09387798607349396, 0.09163457900285721, 0.06334238499403, 0.04882904887199402, 0.043383948504924774, 0.051626671105623245, 0.03616974502801895]\n",
      "[0.7569599747657776, 0.9047200083732605, 0.9333599805831909, 0.9521200060844421, 0.9638800024986267, 0.9676799774169922, 0.9692000150680542, 0.979200005531311, 0.985040009021759, 0.9864400029182434, 0.981719970703125, 0.9896000027656555]\n"
     ]
    }
   ],
   "source": [
    " history=model.fit(x_train, y_train,\n",
    "           batch_size=batch_size,\n",
    "           epochs=12,\n",
    "           validation_data=[x_test, y_test])\n",
    " print(history.history['loss'])\n",
    " print(history.history['accuracy']) "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "measure bi-lstm performance\n",
    "\n",
    "this is good: accuracy line is near 1, loss line is near zero "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    " from matplotlib import pyplot\n",
    " pyplot.plot(history.history['loss'])\n",
    " pyplot.plot(history.history['accuracy'])\n",
    " pyplot.title('model loss vs accuracy')\n",
    " pyplot.xlabel('epoch')\n",
    " pyplot.legend(['loss', 'accuracy'], loc='upper right')\n",
    " pyplot.show() "
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3.10.4 64-bit",
   "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": "3.10.4"
  },
  "orig_nbformat": 4,
  "vscode": {
   "interpreter": {
    "hash": "916dbcbb3f70747c44a77c7bcd40155683ae19c65e1c03b4aa3499c5328201f1"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}