{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "4161eb1f",
   "metadata": {},
   "source": [
    "## Resnet50 TensorFlow Pre-Trained Model Deployment With SageMaker Migration Toolkit\n",
    "\n",
    "- [SageMaker Migration Toolkit]()\n",
    "- [Resnet50 Model]()\n",
    "- Notebook Setting: conda_tf2_p38 Kernel\n",
    "- If you would like to deploy the model without using the API to abstract boto3 calls check the following [example](https://github.com/RamVegiraju/SageMaker-Deployment/tree/master/Misc/resnet-tf-sme)."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c9ea5c88",
   "metadata": {},
   "source": [
    "## API Setup\n",
    "\n",
    "1. Clone GitHub repo\n",
    "2. python setup.py bdist_wheel\n",
    "3. pip install dist/sagemaker_migration_toolkit-0.0.1-py3-none-any.whl\n",
    "4. sagemaker_migration-configure --module-name sagemaker_migration.configure\n",
    "\n",
    "Step 4 is where you provide your role for sagemaker execution, do this via the terminal in your notebook path. To get the role you are currently using you can also run ```sagemaker.get_execution_role()```."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "178343c9",
   "metadata": {},
   "source": [
    "Alternatively you can also install the package after cloning the repo with the following command in the notebook."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "8eb1f553",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Looking in indexes: https://pypi.org/simple, https://pip.repos.neuron.amazonaws.com\n",
      "Processing ./sagemaker-migration-toolkit/dist/sagemaker_migration_toolkit-0.0.1-py3-none-any.whl\n",
      "Requirement already satisfied: boto3>=1.24.0 in /home/ec2-user/anaconda3/envs/tensorflow2_p38/lib/python3.8/site-packages (from sagemaker-migration-toolkit==0.0.1) (1.26.5)\n",
      "Requirement already satisfied: sagemaker>=2.92.1 in /home/ec2-user/anaconda3/envs/tensorflow2_p38/lib/python3.8/site-packages (from sagemaker-migration-toolkit==0.0.1) (2.116.0)\n",
      "Requirement already satisfied: s3transfer<0.7.0,>=0.6.0 in /home/ec2-user/anaconda3/envs/tensorflow2_p38/lib/python3.8/site-packages (from boto3>=1.24.0->sagemaker-migration-toolkit==0.0.1) (0.6.0)\n",
      "Collecting botocore<1.30.0,>=1.29.5\n",
      "  Downloading botocore-1.29.15-py3-none-any.whl (9.9 MB)\n",
      "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m9.9/9.9 MB\u001b[0m \u001b[31m80.7 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m:00:01\u001b[0m00:01\u001b[0m\n",
      "\u001b[?25hRequirement already satisfied: jmespath<2.0.0,>=0.7.1 in /home/ec2-user/anaconda3/envs/tensorflow2_p38/lib/python3.8/site-packages (from boto3>=1.24.0->sagemaker-migration-toolkit==0.0.1) (0.10.0)\n",
      "Requirement already satisfied: google-pasta in /home/ec2-user/anaconda3/envs/tensorflow2_p38/lib/python3.8/site-packages (from sagemaker>=2.92.1->sagemaker-migration-toolkit==0.0.1) (0.2.0)\n",
      "Requirement already satisfied: importlib-metadata<5.0,>=1.4.0 in /home/ec2-user/anaconda3/envs/tensorflow2_p38/lib/python3.8/site-packages (from sagemaker>=2.92.1->sagemaker-migration-toolkit==0.0.1) (4.8.2)\n",
      "Requirement already satisfied: pandas in /home/ec2-user/anaconda3/envs/tensorflow2_p38/lib/python3.8/site-packages (from sagemaker>=2.92.1->sagemaker-migration-toolkit==0.0.1) (1.3.4)\n",
      "Requirement already satisfied: protobuf<4.0,>=3.1 in /home/ec2-user/anaconda3/envs/tensorflow2_p38/lib/python3.8/site-packages (from sagemaker>=2.92.1->sagemaker-migration-toolkit==0.0.1) (3.19.1)\n",
      "Requirement already satisfied: attrs<23,>=20.3.0 in /home/ec2-user/anaconda3/envs/tensorflow2_p38/lib/python3.8/site-packages (from sagemaker>=2.92.1->sagemaker-migration-toolkit==0.0.1) (21.2.0)\n",
      "Requirement already satisfied: pathos in /home/ec2-user/anaconda3/envs/tensorflow2_p38/lib/python3.8/site-packages (from sagemaker>=2.92.1->sagemaker-migration-toolkit==0.0.1) (0.2.8)\n",
      "Requirement already satisfied: schema in /home/ec2-user/anaconda3/envs/tensorflow2_p38/lib/python3.8/site-packages (from sagemaker>=2.92.1->sagemaker-migration-toolkit==0.0.1) (0.7.5)\n",
      "Requirement already satisfied: numpy<2.0,>=1.9.0 in /home/ec2-user/anaconda3/envs/tensorflow2_p38/lib/python3.8/site-packages (from sagemaker>=2.92.1->sagemaker-migration-toolkit==0.0.1) (1.20.3)\n",
      "Requirement already satisfied: protobuf3-to-dict<1.0,>=0.1.5 in /home/ec2-user/anaconda3/envs/tensorflow2_p38/lib/python3.8/site-packages (from sagemaker>=2.92.1->sagemaker-migration-toolkit==0.0.1) (0.1.5)\n",
      "Requirement already satisfied: smdebug-rulesconfig==1.0.1 in /home/ec2-user/anaconda3/envs/tensorflow2_p38/lib/python3.8/site-packages (from sagemaker>=2.92.1->sagemaker-migration-toolkit==0.0.1) (1.0.1)\n",
      "Requirement already satisfied: packaging>=20.0 in /home/ec2-user/anaconda3/envs/tensorflow2_p38/lib/python3.8/site-packages (from sagemaker>=2.92.1->sagemaker-migration-toolkit==0.0.1) (21.0)\n",
      "Requirement already satisfied: urllib3<1.27,>=1.25.4 in /home/ec2-user/anaconda3/envs/tensorflow2_p38/lib/python3.8/site-packages (from botocore<1.30.0,>=1.29.5->boto3>=1.24.0->sagemaker-migration-toolkit==0.0.1) (1.26.8)\n",
      "Requirement already satisfied: python-dateutil<3.0.0,>=2.1 in /home/ec2-user/anaconda3/envs/tensorflow2_p38/lib/python3.8/site-packages (from botocore<1.30.0,>=1.29.5->boto3>=1.24.0->sagemaker-migration-toolkit==0.0.1) (2.8.2)\n",
      "Requirement already satisfied: zipp>=0.5 in /home/ec2-user/anaconda3/envs/tensorflow2_p38/lib/python3.8/site-packages (from importlib-metadata<5.0,>=1.4.0->sagemaker>=2.92.1->sagemaker-migration-toolkit==0.0.1) (3.6.0)\n",
      "Requirement already satisfied: pyparsing>=2.0.2 in /home/ec2-user/anaconda3/envs/tensorflow2_p38/lib/python3.8/site-packages (from packaging>=20.0->sagemaker>=2.92.1->sagemaker-migration-toolkit==0.0.1) (3.0.6)\n",
      "Requirement already satisfied: six in /home/ec2-user/anaconda3/envs/tensorflow2_p38/lib/python3.8/site-packages (from protobuf3-to-dict<1.0,>=0.1.5->sagemaker>=2.92.1->sagemaker-migration-toolkit==0.0.1) (1.16.0)\n",
      "Requirement already satisfied: pytz>=2017.3 in /home/ec2-user/anaconda3/envs/tensorflow2_p38/lib/python3.8/site-packages (from pandas->sagemaker>=2.92.1->sagemaker-migration-toolkit==0.0.1) (2021.3)\n",
      "Requirement already satisfied: pox>=0.3.0 in /home/ec2-user/anaconda3/envs/tensorflow2_p38/lib/python3.8/site-packages (from pathos->sagemaker>=2.92.1->sagemaker-migration-toolkit==0.0.1) (0.3.0)\n",
      "Requirement already satisfied: ppft>=1.6.6.4 in /home/ec2-user/anaconda3/envs/tensorflow2_p38/lib/python3.8/site-packages (from pathos->sagemaker>=2.92.1->sagemaker-migration-toolkit==0.0.1) (1.6.6.4)\n",
      "Requirement already satisfied: multiprocess>=0.70.12 in /home/ec2-user/anaconda3/envs/tensorflow2_p38/lib/python3.8/site-packages (from pathos->sagemaker>=2.92.1->sagemaker-migration-toolkit==0.0.1) (0.70.12.2)\n",
      "Requirement already satisfied: dill>=0.3.4 in /home/ec2-user/anaconda3/envs/tensorflow2_p38/lib/python3.8/site-packages (from pathos->sagemaker>=2.92.1->sagemaker-migration-toolkit==0.0.1) (0.3.4)\n",
      "Requirement already satisfied: contextlib2>=0.5.5 in /home/ec2-user/anaconda3/envs/tensorflow2_p38/lib/python3.8/site-packages (from schema->sagemaker>=2.92.1->sagemaker-migration-toolkit==0.0.1) (21.6.0)\n",
      "Installing collected packages: botocore, sagemaker-migration-toolkit\n",
      "  Attempting uninstall: botocore\n",
      "    Found existing installation: botocore 1.24.19\n",
      "    Uninstalling botocore-1.24.19:\n",
      "      Successfully uninstalled botocore-1.24.19\n",
      "\u001b[31mERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.\n",
      "awscli 1.27.5 requires botocore==1.29.5, but you have botocore 1.29.15 which is incompatible.\n",
      "aiobotocore 2.0.1 requires botocore<1.22.9,>=1.22.8, but you have botocore 1.29.15 which is incompatible.\u001b[0m\u001b[31m\n",
      "\u001b[0mSuccessfully installed botocore-1.29.15 sagemaker-migration-toolkit-0.0.1\n",
      "\u001b[33mWARNING: You are using pip version 22.0.4; however, version 22.3.1 is available.\n",
      "You should consider upgrading via the '/home/ec2-user/anaconda3/envs/tensorflow2_p38/bin/python -m pip install --upgrade pip' command.\u001b[0m\u001b[33m\n",
      "\u001b[0m"
     ]
    }
   ],
   "source": [
    "!pip install sagemaker-migration-toolkit/dist/sagemaker_migration_toolkit-0.0.1-py3-none-any.whl"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3eba2b28",
   "metadata": {},
   "source": [
    "## Resnet50 Model Artifacts\n",
    "\n",
    "The API expects the model data to be serialized in the format compliant with the framework's serving stack. For [TensorFlow Serving](https://towardsdatascience.com/hosting-models-with-tf-serving-on-docker-aceff9fbf533) it should be the following structure.\n",
    "\n",
    "```\n",
    "model_artifact_version/\n",
    "    assets\n",
    "    variables/\n",
    "        variables.data\n",
    "        variables.index\n",
    "    keras_metadata.pb\n",
    "    saved_model.pb \n",
    "inference.py\n",
    "```\n",
    "\n",
    "For our example we don't need an inference.py, but this is mandatory for the package so we will make a dummy script. In the next few cells we will create the model artifacts for the pre-trained resnet50 model in the structure TensorFlow Serving expects."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "efd7c8d9",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'arn:aws:iam::474422712127:role/sagemaker-role-BYOC'"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import os\n",
    "import tensorflow as tf\n",
    "from tensorflow.keras.applications import resnet50\n",
    "from tensorflow.keras import backend\n",
    "import numpy as np\n",
    "from tensorflow.keras.preprocessing import image\n",
    "import sagemaker\n",
    "role = sagemaker.get_execution_role()\n",
    "role #pass this role in step 4 of API setup"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "b2a3c066",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/ec2-user/anaconda3/envs/tensorflow2_p38/lib/python3.8/site-packages/keras/backend.py:414: UserWarning: `tf.keras.backend.set_learning_phase` is deprecated and will be removed after 2020-10-11. To update it, simply pass a True/False value to the `training` argument of the `__call__` method of your layer or model.\n",
      "  warnings.warn('`tf.keras.backend.set_learning_phase` is deprecated and '\n",
      "2022-11-22 20:41:03.270035: E tensorflow/stream_executor/cuda/cuda_driver.cc:271] failed call to cuInit: CUDA_ERROR_NO_DEVICE: no CUDA-capable device is detected\n",
      "2022-11-22 20:41:03.270098: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:156] kernel driver does not appear to be running on this host (ip-172-16-96-163.ec2.internal): /proc/driver/nvidia/version does not exist\n",
      "2022-11-22 20:41:03.270842: I tensorflow/core/platform/cpu_feature_guard.cc:151] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations:  AVX2 AVX512F FMA\n",
      "To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Downloading data from https://storage.googleapis.com/tensorflow/keras-applications/resnet/resnet50_weights_tf_dim_ordering_tf_kernels.h5\n",
      "102973440/102967424 [==============================] - 0s 0us/step\n",
      "102981632/102967424 [==============================] - 0s 0us/step\n"
     ]
    }
   ],
   "source": [
    "tf.keras.backend.set_learning_phase(0)\n",
    "model = resnet50.ResNet50()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "af449863",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Directory  00002  Created \n",
      "WARNING:tensorflow:Compiled the loaded model, but the compiled metrics have yet to be built. `model.compile_metrics` will be empty until you train or evaluate the model.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-11-22 20:41:20.617096: W tensorflow/python/util/util.cc:368] Sets are not currently considered sequences, but this may change in the future, so consider avoiding using them.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: 00002/assets\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/ec2-user/anaconda3/envs/tensorflow2_p38/lib/python3.8/site-packages/keras/engine/functional.py:1410: CustomMaskWarning: Custom mask layers require a config and must override get_config. When loading, the custom mask layer must be passed to the custom_objects argument.\n",
      "  layer_config = serialize_layer_fn(layer)\n",
      "/home/ec2-user/anaconda3/envs/tensorflow2_p38/lib/python3.8/site-packages/keras/saving/saved_model/layer_serialization.py:112: CustomMaskWarning: Custom mask layers require a config and must override get_config. When loading, the custom mask layer must be passed to the custom_objects argument.\n",
      "  return generic_utils.serialize_keras_object(obj)\n"
     ]
    }
   ],
   "source": [
    "export_dir = \"00002\" #directory to store model artifacts\n",
    "model = tf.keras.applications.ResNet50()\n",
    "\n",
    "if not os.path.exists(export_dir):\n",
    "    os.makedirs(export_dir)\n",
    "    print(\"Directory \", export_dir, \" Created \")\n",
    "else:\n",
    "    print(\"Directory \", export_dir, \" already exists\")\n",
    "# Save to SavedModel\n",
    "model.save(export_dir, save_format=\"tf\", include_optimizer=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "90c4bb35",
   "metadata": {},
   "source": [
    "Here we will create a dummy inference.py script, you should replace this with your actual pre and post processing as necessary"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "88987147",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Writing inference.py\n"
     ]
    }
   ],
   "source": [
    "%%writefile inference.py\n",
    "import os\n",
    "import json"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "105fa447",
   "metadata": {},
   "source": [
    "## SageMaker Migration Toolkit Deployment"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "b97e2f17",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sagemaker_migration import frameworks as fwk\n",
    "import os"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "836f6d03",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create a TensorFlow Model Object Entity, you can create a real-time or serverless endpoint\n",
    "tf_model = fwk.TensorFlowModel(\n",
    "    version = '2.3.0',\n",
    "    model_data = '00002',\n",
    "    inference_option = 'real-time',\n",
    "    inference = 'inference.py',\n",
    "    instance_type = 'ml.m5.xlarge')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "0f36b6d2",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model data contains:  00002\n",
      "['assets', 'keras_metadata.pb', 'variables', 'saved_model.pb']\n",
      "00002\n",
      "inference.py\n",
      "model.tar.gz\n",
      "Uploading model.tar.gz to S3...\n",
      "Creating model in SageMaker...\n",
      "Created model: sm-model-tensorflow-2022-11-22-20-42-30\n",
      "Creating endpoint config in SageMaker...\n",
      "Created endpoint config: sm-endpoint-config-tensorflow-2022-11-22-20-42-30\n",
      "Creating endpoint in SageMaker...\n",
      "Endpoint creation in process: sm-endpoint-tensorflow-2022-11-22-20-42-30\n",
      "Monitoring endpoint creation...\n",
      "Creating\n",
      "Creating\n",
      "Creating\n",
      "Creating\n",
      "Creating\n",
      "Creating\n",
      "Creating\n",
      "InService\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{'EndpointName': 'sm-endpoint-tensorflow-2022-11-22-20-42-30',\n",
       " 'EndpointArn': 'arn:aws:sagemaker:us-east-1:474422712127:endpoint/sm-endpoint-tensorflow-2022-11-22-20-42-30',\n",
       " 'EndpointConfigName': 'sm-endpoint-config-tensorflow-2022-11-22-20-42-30',\n",
       " 'ProductionVariants': [{'VariantName': 'primaryvariant',\n",
       "   'DeployedImages': [{'SpecifiedImage': '763104351884.dkr.ecr.us-east-1.amazonaws.com/tensorflow-inference:2.3.0-cpu',\n",
       "     'ResolvedImage': '763104351884.dkr.ecr.us-east-1.amazonaws.com/tensorflow-inference@sha256:91ebb7428846c5f7b515d5d9b8389a14c73d0c5d02657f4a6413592124333278',\n",
       "     'ResolutionTime': datetime.datetime(2022, 11, 22, 20, 42, 31, 650000, tzinfo=tzlocal())}],\n",
       "   'CurrentWeight': 1.0,\n",
       "   'DesiredWeight': 1.0,\n",
       "   'CurrentInstanceCount': 1,\n",
       "   'DesiredInstanceCount': 1}],\n",
       " 'EndpointStatus': 'InService',\n",
       " 'CreationTime': datetime.datetime(2022, 11, 22, 20, 42, 31, 147000, tzinfo=tzlocal()),\n",
       " 'LastModifiedTime': datetime.datetime(2022, 11, 22, 20, 44, 9, 1000, tzinfo=tzlocal()),\n",
       " 'ResponseMetadata': {'RequestId': '107c4e22-7577-40ac-bfa3-3aabdb5042bc',\n",
       "  'HTTPStatusCode': 200,\n",
       "  'HTTPHeaders': {'x-amzn-requestid': '107c4e22-7577-40ac-bfa3-3aabdb5042bc',\n",
       "   'content-type': 'application/x-amz-json-1.1',\n",
       "   'content-length': '791',\n",
       "   'date': 'Tue, 22 Nov 2022 20:44:16 GMT'},\n",
       "  'RetryAttempts': 0}}"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tf_model.deploy_to_sagemaker()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9e2aa879",
   "metadata": {},
   "source": [
    "## Inference"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "8d92957a",
   "metadata": {},
   "outputs": [],
   "source": [
    "import boto3\n",
    "#runtime client for inference\n",
    "runtime = boto3.client(service_name=\"sagemaker-runtime\") #runtime boto3 client for inference\n",
    "#grab endpoint name\n",
    "endpoint_name = 'sm-endpoint-tensorflow-2022-11-22-20-42-30'"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "41fdd950",
   "metadata": {},
   "source": [
    "You can retrieve the data from the public Amazon datasets with the following curl command. I've also posted the same image in the repo incase you cannot retrieve.\n",
    "\n",
    "```\n",
    "file_name = 'kitten.jpg'\n",
    "!wget -q https://s3.amazonaws.com/model-server/inputs/kitten.jpg -O {file_name}\n",
    "```"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "ce02c89f",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'numpy.ndarray'>\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "from tensorflow.keras.preprocessing import image\n",
    "\n",
    "# preprocess image\n",
    "img = image.load_img(\"kitten.jpg\", target_size=(224, 224))\n",
    "x = image.img_to_array(img)\n",
    "x = np.expand_dims(x, axis=0)\n",
    "x = resnet50.preprocess_input(x)\n",
    "print(type(x))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "f2ffe237",
   "metadata": {},
   "outputs": [],
   "source": [
    "import json\n",
    "payload = json.dumps(x.tolist())\n",
    "#payload"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "4e92c9d7",
   "metadata": {},
   "outputs": [],
   "source": [
    "response = runtime.invoke_endpoint(EndpointName=endpoint_name,\n",
    "                                   ContentType='application/json',\n",
    "                                   Body=payload)\n",
    "result = json.loads(response['Body'].read().decode())\n",
    "#result"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "conda_tensorflow2_p38",
   "language": "python",
   "name": "conda_tensorflow2_p38"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}