sdk/python/endpoints/batch/imagenet-classifier-batch.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "source": [
    "# Use batch deployments for image file processing"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The following notebook demostrates how to use batch endpoints to deploy models that work with images. Particularly, we are going to deploy a TensorFlow model for the popular ImageNet classification problem."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This notebook requires:\n",
    "\n",
    "- `tensorflow`\n",
    "- `tensorflow_hub`\n",
    "- `pillow`\n",
    "- `azure-ai-ml`\n",
    "- `azureml-mlflow`\n",
    "- `pandas`\n",
    "- `scipy`"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 1. Connect to Azure Machine Learning Workspace\n",
    "\n",
    "The [workspace](https://docs.microsoft.com/en-us/azure/machine-learning/concept-workspace) is the top-level resource for Azure Machine Learning, providing a centralized place to work with all the artifacts you create when you use Azure Machine Learning. In this section we will connect to the workspace in which the job will be run.\n",
    "\n",
    "### 1.1. Import the required libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from azure.ai.ml import MLClient, Input\n",
    "from azure.ai.ml.entities import (\n",
    "    BatchEndpoint,\n",
    "    BatchDeployment,\n",
    "    Model,\n",
    "    AmlCompute,\n",
    "    Data,\n",
    "    BatchRetrySettings,\n",
    "    CodeConfiguration,\n",
    "    Environment,\n",
    ")\n",
    "from azure.ai.ml.constants import AssetTypes, BatchDeploymentOutputAction\n",
    "from azure.identity import DefaultAzureCredential"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "source": [
    "### 1.2. Configure workspace details and get a handle to the workspace\n",
    "\n",
    "To connect to a workspace, we need identifier parameters - a subscription, resource group and workspace name. We will use these details in the `MLClient` from `azure.ai.ml` to get a handle to the required Azure Machine Learning workspace. We use the default [default azure authentication](https://docs.microsoft.com/en-us/python/api/azure-identity/azure.identity.defaultazurecredential?view=azure-python) for this tutorial. Check the [configuration notebook](../../jobs/configuration.ipynb) for more details on how to configure credentials and connect to a workspace."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# enter details of your AML workspace\n",
    "subscription_id = \"<SUBSCRIPTION_ID>\"\n",
    "resource_group = \"<RESOURCE_GROUP>\"\n",
    "workspace = \"<AML_WORKSPACE_NAME>\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "jupyter": {
     "outputs_hidden": false,
     "source_hidden": false
    },
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "outputs": [],
   "source": [
    "ml_client = MLClient(\n",
    "    DefaultAzureCredential(), subscription_id, resource_group, workspace\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "source": [
    "## 2. Registering the model\n",
    "\n",
    "### 2.1 About the model\n",
    "\n",
    "Let's review how the model is built. The model was built using TensorFlow along with the RestNet architecture ([Identity Mappings in Deep Residual Networks](https://arxiv.org/abs/1603.05027)). This model has the following constraints that are important to keep in mind for deployment:\n",
    "\n",
    "* In work with images of size 244x244 (tensors of `(224, 224, 3)`).\n",
    "* It requires inputs to be scaled to the range `[0,1]`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "jupyter": {
     "outputs_hidden": false,
     "source_hidden": false
    },
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "outputs": [],
   "source": [
    "import tensorflow_hub as hub\n",
    "import tensorflow as tf\n",
    "\n",
    "model = tf.keras.Sequential(\n",
    "    [\n",
    "        hub.KerasLayer(\n",
    "            \"https://tfhub.dev/google/imagenet/resnet_v2_101/classification/5\"\n",
    "        ),\n",
    "    ]\n",
    ")\n",
    "model.build([None, None, None, 3])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "source": [
    "Testing if the model works:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "jupyter": {
     "outputs_hidden": false,
     "source_hidden": false
    },
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "outputs": [],
   "source": [
    "import PIL.Image as Image\n",
    "import numpy as np\n",
    "\n",
    "image_file = tf.keras.utils.get_file(\n",
    "    \"image.jpeg\",\n",
    "    \"https://azuremlexampledata.blob.core.windows.net/data/imagenet/goldfish.JPEG\",\n",
    ")\n",
    "img = Image.open(image_file).resize((244, 244))\n",
    "img = np.array(img) / 255.0\n",
    "batch_img = tf.expand_dims(img, axis=0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "source": [
    "Run the model:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "jupyter": {
     "outputs_hidden": false,
     "source_hidden": false
    },
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "outputs": [],
   "source": [
    "pred = model.predict(batch_img)\n",
    "pred_class = pred.argmax(axis=-1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "source": [
    "Getting the labels for ImageNet:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "jupyter": {
     "outputs_hidden": false,
     "source_hidden": false
    },
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "outputs": [],
   "source": [
    "labels_path = tf.keras.utils.get_file(\n",
    "    \"ImageNetLabels.txt\",\n",
    "    \"https://azuremlexampledata.blob.core.windows.net/data/imagenet/ImageNetLabels.txt\",\n",
    ")\n",
    "imagenet_labels = np.array(open(labels_path).read().splitlines())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "jupyter": {
     "outputs_hidden": false,
     "source_hidden": false
    },
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "outputs": [],
   "source": [
    "predicted_class_name = [\n",
    "    imagenet_labels[predicted_class] for predicted_class in pred_class\n",
    "]\n",
    "predicted_class_name"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "source": [
    "Let's save this model locally:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "jupyter": {
     "outputs_hidden": false,
     "source_hidden": false
    },
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "outputs": [],
   "source": [
    "model_local_path = \"imagenet-classifier/model\"\n",
    "model.save(model_local_path)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "source": [
    "### 2.2 Registering the model in the workspace\n",
    "\n",
    "We need to register the model in order to use it with Azure Machine Learning:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "jupyter": {
     "outputs_hidden": false,
     "source_hidden": false
    },
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "outputs": [],
   "source": [
    "model_name = \"imagenet-classifier\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "jupyter": {
     "outputs_hidden": false,
     "source_hidden": false
    },
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "outputs": [],
   "source": [
    "if not any(filter(lambda m: m.name == model_name, ml_client.models.list())):\n",
    "    print(f\"Model {model_name} is not registered. Creating...\")\n",
    "    model = ml_client.models.create_or_update(\n",
    "        Model(name=model_name, path=model_local_path, type=AssetTypes.CUSTOM_MODEL)\n",
    "    )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "source": [
    "Let's get a reference to the model:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "jupyter": {
     "outputs_hidden": false,
     "source_hidden": false
    },
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "outputs": [],
   "source": [
    "model = ml_client.models.get(name=model_name, label=\"latest\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "source": [
    "## 3 Create Batch Endpoint\n",
    "\n",
    "Batch endpoints are endpoints that are used batch inferencing on large volumes of data over a period of time. Batch endpoints receive pointers to data and run jobs asynchronously to process the data in parallel on compute clusters. Batch endpoints store outputs to a data store for further analysis.\n",
    "\n",
    "To create an online endpoint we will use `BatchEndpoint`. This class allows user to configure the following key aspects:\n",
    "- `name` - Name of the endpoint. Needs to be unique at the Azure region level\n",
    "- `auth_mode` - The authentication method for the endpoint. Currently only Azure Active Directory (Azure AD) token-based (`aad_token`) authentication is supported. \n",
    "- `defaults` - Default settings for the endpoint.\n",
    "   - `deployment_name` - Name of the deployment that will serve as the default deployment for the endpoint.\n",
    "- `description`- Description of the endpoint.\n",
    "\n",
    "### 3.1 Configure the endpoint\n",
    "\n",
    "First, let's create the endpoint that is going to host the batch deployments. To ensure that our endpoint name is unique, let's create a random suffix to append to it. \n",
    "\n",
    "> In general, you won't need to use this technique but you will use more meaningful names. Please skip the following cell if your case:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "jupyter": {
     "outputs_hidden": false,
     "source_hidden": false
    },
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "outputs": [],
   "source": [
    "import random\n",
    "import string\n",
    "\n",
    "# Creating a unique endpoint name by including a random suffix\n",
    "allowed_chars = string.ascii_lowercase + string.digits\n",
    "endpoint_suffix = \"\".join(random.choice(allowed_chars) for x in range(5))\n",
    "endpoint_name = \"imagenet-classifier-\" + endpoint_suffix"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's configure the endpoint:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "endpoint = BatchEndpoint(\n",
    "    name=endpoint_name,\n",
    "    description=\"An batch service to perform ImageNet image classification\",\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.2 Create the endpoint\n",
    "Using the `MLClient` created earlier, we will now create the Endpoint in the workspace. This command will start the endpoint creation and return a confirmation response while the endpoint creation continues."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "jupyter": {
     "outputs_hidden": false,
     "source_hidden": false
    },
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "outputs": [],
   "source": [
    "ml_client.batch_endpoints.begin_create_or_update(endpoint).result()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4. Create a batch deployment\n",
    "\n",
    "A deployment is a set of resources required for hosting the model that does the actual inferencing. We will create a deployment for our endpoint using the `BatchDeployment` class.\n",
    "\n",
    "### 4.1 Creating an scoring script to work with the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "jupyter": {
     "outputs_hidden": false,
     "source_hidden": false
    },
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "outputs": [],
   "source": [
    "%%writefile imagenet-classifier/code/batch_driver.py\n",
    "\n",
    "import os\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import tensorflow as tf\n",
    "from os.path import basename\n",
    "from PIL import Image\n",
    "from tensorflow.keras.models import load_model\n",
    "\n",
    "\n",
    "def init():\n",
    "    global model\n",
    "    global input_width\n",
    "    global input_height\n",
    "\n",
    "    # AZUREML_MODEL_DIR is an environment variable created during deployment\n",
    "    model_path = os.path.join(os.environ[\"AZUREML_MODEL_DIR\"], \"model\")\n",
    "\n",
    "    # load the model\n",
    "    model = load_model(model_path)\n",
    "    input_width = 244\n",
    "    input_height = 244\n",
    "\n",
    "def run(mini_batch):\n",
    "    results = []\n",
    "\n",
    "    for image in mini_batch:\n",
    "        data = Image.open(image).resize((input_width, input_height)) # Read and resize the image\n",
    "        data = np.array(data)/255.0 # Normalize\n",
    "        data_batch = tf.expand_dims(data, axis=0) # create a batch of size (1, 244, 244, 3)\n",
    "\n",
    "        # perform inference\n",
    "        pred = model.predict(data_batch)\n",
    "\n",
    "        # Compute probabilities, classes and labels\n",
    "        pred_prob = tf.math.reduce_max(tf.math.softmax(pred, axis=-1)).numpy()\n",
    "        pred_class = tf.math.argmax(pred, axis=-1).numpy()\n",
    "\n",
    "        results.append([basename(image), pred_class[0], pred_prob])\n",
    "\n",
    "    return pd.DataFrame(results)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "source": [
    "### 4.2 Creating the compute\n",
    "\n",
    "Batch deployments can run on any Azure ML compute that already exists in the workspace. That means that multiple batch deployments can share the same compute infrastructure. In this example, we are going to work on an AzureML compute cluster called `cpu-cluster`. Let's verify the compute exists on the workspace or create it otherwise."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "jupyter": {
     "outputs_hidden": false,
     "source_hidden": false
    },
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "outputs": [],
   "source": [
    "compute_name = \"cpu-cluster\"\n",
    "if not any(filter(lambda m: m.name == compute_name, ml_client.compute.list())):\n",
    "    print(f\"Compute {compute_name} is not created. Creating...\")\n",
    "    compute_cluster = AmlCompute(\n",
    "        name=compute_name, description=\"amlcompute\", min_instances=0, max_instances=5\n",
    "    )\n",
    "    ml_client.begin_create_or_update(compute_cluster).result()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "source": [
    "Compute may take time to be created. Let's wait for it:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "jupyter": {
     "outputs_hidden": false,
     "source_hidden": false
    },
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "outputs": [],
   "source": [
    "print(\"Waiting for compute\", end=\"\")\n",
    "while ml_client.compute.get(name=compute_name).provisioning_state == \"Creating\":\n",
    "    sleep(1)\n",
    "    print(\".\", end=\"\")\n",
    "\n",
    "print(\" [DONE]\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "source": [
    "### 4.3 Creating the environment\n",
    "\n",
    "Let's create the environment. In our case, our model runs on `TensorFlow`. Azure Machine Learning already has an environment with the required software installed, so we can reutilize this environment."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "jupyter": {
     "outputs_hidden": false,
     "source_hidden": false
    },
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "outputs": [],
   "source": [
    "environment = Environment(\n",
    "    conda_file=\"./imagenet-classifier/environment/conda.yml\",\n",
    "    image=\"mcr.microsoft.com/azureml/tensorflow-2.4-ubuntu18.04-py37-cpu-inference:latest\",\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "source": [
    " ### 4.4 Configuring the deployment\n",
    " \n",
    " We will create a deployment for our endpoint using the `BatchDeployment` class. This class allows user to configure the following key aspects.\n",
    "- `name` - Name of the deployment.\n",
    "- `endpoint_name` - Name of the endpoint to create the deployment under.\n",
    "- `model` - The model to use for the deployment. This value can be either a reference to an existing versioned model in the workspace or an inline model specification.\n",
    "- `environment` - The environment to use for the deployment. This value can be either a reference to an existing versioned environment in the workspace or an inline environment specification.\n",
    "- `code_path`- Path to the source code directory for scoring the model\n",
    "- `scoring_script` - Relative path to the scoring file in the source code directory\n",
    "- `compute` - Name of the compute target to execute the batch scoring jobs on\n",
    "- `instance_count`- The number of nodes to use for each batch scoring job.\t\t1\n",
    "- `max_concurrency_per_instance`- The maximum number of parallel scoring_script runs per instance.\n",
    "- `mini_batch_size`\t- The number of files the code_configuration.scoring_script can process in one `run()` call.\n",
    "- `retry_settings`- Retry settings for scoring each mini batch.\t\t\n",
    "   - `max_retries`- The maximum number of retries for a failed or timed-out mini batch (default is 3)\n",
    "   - `timeout`- The timeout in seconds for scoring a mini batch (default is 30)\n",
    "- `output_action`- Indicates how the output should be organized in the output file. Allowed values are `append_row` or `summary_only`. Default is `append_row`\n",
    "- `output_file_name`- Name of the batch scoring output file. Default is `predictions.csv`\n",
    "- `environment_variables`- Dictionary of environment variable name-value pairs to set for each batch scoring job.\n",
    "- `logging_level`- The log verbosity level.\tAllowed values are `warning`, `info`, `debug`. Default is `info`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "jupyter": {
     "outputs_hidden": false,
     "source_hidden": false
    },
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "outputs": [],
   "source": [
    "deployment = BatchDeployment(\n",
    "    name=\"imagenet-classifier-resnetv2\",\n",
    "    description=\"A ResNetV2 model architecture for performing ImageNet classification in batch\",\n",
    "    endpoint_name=endpoint.name,\n",
    "    model=model,\n",
    "    environment=environment,\n",
    "    code_configuration=CodeConfiguration(\n",
    "        code=\"./imagenet-classifier/code/\",\n",
    "        scoring_script=\"batch_driver.py\",\n",
    "    ),\n",
    "    compute=compute_name,\n",
    "    instance_count=2,\n",
    "    max_concurrency_per_instance=1,\n",
    "    mini_batch_size=10,\n",
    "    output_action=BatchDeploymentOutputAction.APPEND_ROW,\n",
    "    output_file_name=\"predictions.csv\",\n",
    "    retry_settings=BatchRetrySettings(max_retries=3, timeout=300),\n",
    "    logging_level=\"info\",\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4.5 Create the deployment\n",
    "Using the `MLClient` created earlier, we will now create the deployment in the workspace. This command will start the deployment creation and return a confirmation response while the deployment creation continues."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "jupyter": {
     "outputs_hidden": false,
     "source_hidden": false
    },
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "outputs": [],
   "source": [
    "ml_client.batch_deployments.begin_create_or_update(deployment).result()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4.6 Testing the deployment\n",
    "\n",
    "Once the deployment is created, it is ready to recieve jobs.\n",
    "\n",
    "#### 4.6.1 Creating a data asset\n",
    "\n",
    "Let's first register a data asset so we can run the job against it. This data asset is a folder containing 1000 images from the original ImageNet dataset. We are going to download it first and then create the data asset:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "jupyter": {
     "outputs_hidden": false,
     "source_hidden": false
    },
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "outputs": [],
   "source": [
    "!wget https://azuremlexampledata.blob.core.windows.net/data/imagenet/imagenet-1000.zip\n",
    "!unzip imagenet-1000.zip -d /tmp/imagenet-1000"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Registering a data asset:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "jupyter": {
     "outputs_hidden": false,
     "source_hidden": false
    },
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "outputs": [],
   "source": [
    "data_path = \"/tmp/imagenet-1000\"\n",
    "dataset_name = \"imagenet-sample-unlabeled\"\n",
    "\n",
    "imagenet_sample = Data(\n",
    "    path=data_path,\n",
    "    type=AssetTypes.URI_FOLDER,\n",
    "    description=\"A sample of 1000 images from the original ImageNet dataset\",\n",
    "    name=dataset_name,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "jupyter": {
     "outputs_hidden": false,
     "source_hidden": false
    },
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "outputs": [],
   "source": [
    "ml_client.data.create_or_update(imagenet_sample)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's wait for the data asset:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from time import sleep\n",
    "\n",
    "print(f\"Waiting for data asset {dataset_name}\", end=\"\")\n",
    "while not any(filter(lambda m: m.name == dataset_name, ml_client.data.list())):\n",
    "    sleep(10)\n",
    "    print(\".\", end=\"\")\n",
    "\n",
    "print(\" [DONE]\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "source": [
    "Let's get a reference of the new data asset:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "jupyter": {
     "outputs_hidden": false,
     "source_hidden": false
    },
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "outputs": [],
   "source": [
    "imagenet_sample = ml_client.data.get(name=dataset_name, label=\"latest\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 4.6.2 Creating an input for the deployment"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "jupyter": {
     "outputs_hidden": false,
     "source_hidden": false
    },
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "outputs": [],
   "source": [
    "input = Input(type=AssetTypes.URI_FOLDER, path=imagenet_sample.id)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "source": [
    "#### 4.6.3 Invoke the deployment\n",
    "\n",
    "Using the `MLClient` created earlier, we will get a handle to the endpoint. The endpoint can be invoked using the `invoke` command with the following parameters:\n",
    "- `name` - Name of the endpoint\n",
    "- `input_path` - Path where input data is present\n",
    "- `deployment_name` - Name of the specific deployment to test in an endpoint"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "jupyter": {
     "outputs_hidden": false,
     "source_hidden": false
    },
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "outputs": [],
   "source": [
    "job = ml_client.batch_endpoints.invoke(\n",
    "    endpoint_name=endpoint.name, deployment_name=deployment.name, input=input\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 4.6.4 Get the details of the invoked job\n",
    "\n",
    "Let us get details and logs of the invoked job:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "jupyter": {
     "outputs_hidden": false,
     "source_hidden": false
    },
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "outputs": [],
   "source": [
    "ml_client.jobs.get(job.name)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can wait for the job to finish using the following code:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(f\"Waiting for batch job deployment {job.name}\", end=\"\")\n",
    "while ml_client.jobs.get(name=job.name).status not in [\"Completed\", \"Failed\"]:\n",
    "    sleep(10)\n",
    "    print(\".\", end=\"\")\n",
    "\n",
    "print(\" [DONE]\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4.7 Exploring the results\n",
    "\n",
    "#### 4.7.1 Download the results\n",
    "\n",
    "The deployment creates a child job that executes the scoring. We can get the details of it using the following code:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "jupyter": {
     "outputs_hidden": false,
     "source_hidden": false
    },
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "outputs": [],
   "source": [
    "scoring_job = list(ml_client.jobs.list(parent_job_name=job.name))[0]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The outputs generated by the deployment job will be placed in an output named `score`:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "ml_client.jobs.download(name=scoring_job.name, download_path=\".\", output_name=\"score\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can read this data using pandas library:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "jupyter": {
     "outputs_hidden": false,
     "source_hidden": false
    },
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "\n",
    "score = pd.read_csv(\n",
    "    \"named-outputs/score/predictions.csv\",\n",
    "    header=None,\n",
    "    names=[\"file\", \"class\", \"probabilities\"],\n",
    "    sep=\" \",\n",
    ")\n",
    "score[\"label\"] = score[\"class\"].apply(lambda pred: imagenet_labels[pred])\n",
    "score"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "source": [
    "### 5. Setting the default deployment\n",
    "\n",
    "Once the deployment works correctly as we expect, we can update the default deployment to the new deployment so any invocation of the endpoint triggers the created deployment."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "jupyter": {
     "outputs_hidden": false,
     "source_hidden": false
    },
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "outputs": [],
   "source": [
    "endpoint = ml_client.batch_endpoints.get(endpoint_name)\n",
    "endpoint.defaults.deployment_name = deployment.name\n",
    "ml_client.batch_endpoints.begin_create_or_update(endpoint).result()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "source": [
    "We can see the endpoint URL as follows:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "jupyter": {
     "outputs_hidden": false,
     "source_hidden": false
    },
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "outputs": [],
   "source": [
    "endpoint = ml_client.batch_endpoints.get(endpoint_name)\n",
    "print(f\"The default deployment is {endpoint.defaults.deployment_name}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "source": [
    "## 6. (Optional) High throughput deployments\n",
    "\n",
    "We can achieve high throughput in deployments that score batches of images all at once instead of iterating one by one over the mini-batch. This kind of deployments can gain 5x of performance on CPU and 20x on a GPU (depending on hardware configuration and batching/parallelization)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 6.1 Creating an scoring script to work with the model in batch"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "jupyter": {
     "outputs_hidden": false,
     "source_hidden": false
    },
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "outputs": [],
   "source": [
    "%%writefile imagenet-classifier/code/batch_driver_ht.py\n",
    "\n",
    "import os\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import tensorflow as tf\n",
    "from tensorflow.keras.models import load_model\n",
    "\n",
    "\n",
    "def init():\n",
    "    global model\n",
    "    global input_width\n",
    "    global input_height\n",
    "\n",
    "    # AZUREML_MODEL_DIR is an environment variable created during deployment\n",
    "    model_path = os.path.join(os.environ[\"AZUREML_MODEL_DIR\"], \"model\")\n",
    "\n",
    "    # load the model\n",
    "    model = load_model(model_path)\n",
    "    input_width = 244\n",
    "    input_height = 244\n",
    "\n",
    "def decode_img(file_path):\n",
    "    file = tf.io.read_file(file_path)\n",
    "    img = tf.io.decode_jpeg(file, channels=3)\n",
    "    img = tf.image.resize(img, [input_width, input_height])\n",
    "    return img/255.\n",
    "\n",
    "def run(mini_batch):\n",
    "    images_ds = tf.data.Dataset.from_tensor_slices(mini_batch)\n",
    "    images_ds = images_ds.map(decode_img).batch(64)\n",
    "\n",
    "    # perform inference\n",
    "    pred = model.predict(images_ds)\n",
    "\n",
    "    # Compute probabilities, classes and labels\n",
    "    pred_prob = tf.math.reduce_max(tf.math.softmax(pred, axis=-1)).numpy()\n",
    "    pred_class = tf.math.argmax(pred, axis=-1).numpy()\n",
    "\n",
    "    return pd.DataFrame({ \n",
    "        \"file\": mini_batch, \n",
    "        \"class\": pred_class,\n",
    "        \"probability\": pred_prob, \n",
    "    })"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "source": [
    "### 6.2 Configuring a new deployment for the high performance inference"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "jupyter": {
     "outputs_hidden": false,
     "source_hidden": false
    },
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "outputs": [],
   "source": [
    "ht_deployment = BatchDeployment(\n",
    "    name=\"imagenet-classifier-resnetv2-ht\",\n",
    "    description=\"A ResNetV2 model architecture for performing ImageNet classification in batch (High throughput)\",\n",
    "    endpoint_name=endpoint.name,\n",
    "    model=model,\n",
    "    environment=environment,\n",
    "    code_configuration=CodeConfiguration(\n",
    "        code=\"./imagenet-classifier/code/\",\n",
    "        scoring_script=\"batch_driver_ht.py\",\n",
    "    ),\n",
    "    compute=compute_name,\n",
    "    instance_count=2,\n",
    "    max_concurrency_per_instance=1,\n",
    "    mini_batch_size=10,\n",
    "    output_action=BatchDeploymentOutputAction.APPEND_ROW,\n",
    "    output_file_name=\"predictions.csv\",\n",
    "    retry_settings=BatchRetrySettings(max_retries=3, timeout=300),\n",
    "    logging_level=\"info\",\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 6.3 Create the deployment"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "jupyter": {
     "outputs_hidden": false,
     "source_hidden": false
    },
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "outputs": [],
   "source": [
    "ml_client.batch_deployments.begin_create_or_update(ht_deployment).result()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 6.4 Invoke the new deployment"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "source": [
    "#### 6.4.1 Execute\n",
    "\n",
    "Let's execute this specific deployment now:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "jupyter": {
     "outputs_hidden": false,
     "source_hidden": false
    },
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "outputs": [],
   "source": [
    "job = ml_client.batch_endpoints.invoke(\n",
    "    endpoint_name=endpoint.name, deployment_name=ht_deployment.name, input=input\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 6.4.2 Get the details of the invoked job\n",
    "\n",
    "Let us get details and logs of the invoked job:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "ml_client.jobs.get(job.name)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can wait for the job to finish using the following code:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(f\"Waiting for batch deployment job {job.name}\", end=\"\")\n",
    "while ml_client.jobs.get(name=job.name).status not in [\"Completed\", \"Failed\"]:\n",
    "    sleep(10)\n",
    "    print(\".\", end=\"\")\n",
    "\n",
    "print(\" [DONE]\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 6.5 Exploring the results\n",
    "\n",
    "#### 6.5.1 Download the results\n",
    "\n",
    "The deployment creates a child job that executes the scoring. We can get the details of it using the following code:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "jupyter": {
     "outputs_hidden": false,
     "source_hidden": false
    },
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "outputs": [],
   "source": [
    "scoring_job = list(ml_client.jobs.list(parent_job_name=job.name))[0]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The outputs generated by the deployment job will be placed in an output named `score`:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "ml_client.jobs.download(name=scoring_job.name, download_path=\".\", output_name=\"score\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can read this data using pandas library:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "jupyter": {
     "outputs_hidden": false,
     "source_hidden": false
    },
    "nteract": {
     "transient": {
      "deleting": false
     }
    }
   },
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "\n",
    "score = pd.read_csv(\n",
    "    \"named-outputs/score/predictions.csv\",\n",
    "    header=None,\n",
    "    names=[\"file\", \"class\", \"probabilities\"],\n",
    "    sep=\" \",\n",
    ")\n",
    "score[\"label\"] = score[\"class\"].apply(lambda pred: imagenet_labels[pred])\n",
    "score"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 7. Clean up resources\n",
    "\n",
    "Clean-up the resources created. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "ml_client.batch_endpoints.begin_delete(endpoint_name)"
   ]
  }
 ],
 "metadata": {
  "kernel_info": {
   "name": "amlv2"
  },
  "kernelspec": {
   "display_name": "Python 3.10 - SDK V2",
   "language": "python",
   "name": "python310-sdkv2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.15"
  },
  "nteract": {
   "version": "nteract-front-end@1.0.0"
  },
  "vscode": {
   "interpreter": {
    "hash": "8d732042e5e620df2ddb4aad7f460808ed1754fa045785bdb47941e58456b253"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}