ex_09_cifar_ray_air_example.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Ray AI Runtime: An end-to-end for image classification example\n",
    "\n",
    "[Ray AI Runtime (AIR)](https://docs.ray.io/en/latest/ray-air/getting-started.html) is a scalable and unified toolkit for machine learning (ML) applications. AIR enables simple scaling of individual workloads and end-to-end workflows using popular ML frameworks, all in just Python.\n",
    "\n",
    "AIR builds on Ray’s best-in-class libraries for [Preprocessing](https://docs.ray.io/en/latest/data/dataset.html#datasets), [Training](https://docs.ray.io/en/latest/train/train.html#train-docs), [Tuning](https://docs.ray.io/en/latest/tune/index.html#tune-main), [Scoring](https://docs.ray.io/en/latest/ray-air/predictors.html#air-predictors), [Serving](https://docs.ray.io/en/latest/serve/index.html#rayserve), and [Reinforcement Learning](https://docs.ray.io/en/latest/rllib/index.html#rllib-index) to bring together an ecosystem of integrations. \n",
    "\n",
    "Out of the box with Ray AIR, you can easily build the common machine line pipeline with AIR's components.\n",
    "\n",
    "<img src=\"images/air_ml_workflow.png\" width=\"60%\" height=\"30%\">\n",
    "\n",
    "This notebook employs each of these components to illustrate how in six distinct steps, using each of the Ray AIR [components and its APIs](https://docs.ray.io/en/latest/ray-air/package-ref.html#ray-air-api), to build a common ML pipeline for an image classification.\n",
    "\n",
    "\n",
    "### The CIFAR-10 dataset for classification\n",
    "\n",
    "The CIFAR-10 dataset consists of 60,000 32x32 colour images in 10 classes, with 6000 images per class. There are 50000 training images and 10000 test images.\n",
    "\n",
    "The dataset is divided into five training batches and one test batch, each with 10000 images. The test batch contains exactly 1000 randomly-selected images from each class. The training batches contain the remaining images in random order, but some training batches may contain more images from one class than another. Between them, the training batches contain exactly 5000 images from each class.\n",
    "\n",
    "Here are the classes in the dataset, as well as 10 random images from each:\n",
    "\n",
    "<img src=\"images/cifar-10.png\" width=\"60%\" height=\"30%\">\n",
    "\n",
    "[Cifar-10 source](https://www.cs.toronto.edu/~kriz/cifar.html)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Learning objectives\n",
    "\n",
    "  * Use AIR's PyTorch Trainers, Tuners, Checkpoint, Batch & Online inference predictors\n",
    "  * Understand how Ray data integrates with Ray AIR for data ingestion and prediction\n",
    "  * Use out-of-box Preprocessors\n",
    "  * Load model from the best model checkpoint and use for batch inference\n",
    "  * Deploy best checkpoint model and use for online inference\n",
    "  * Using all the Ray discrete components to write an end-to-end ML application in a single Python script\n",
    "  * Use PyTorch for data distributed training (DDP)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "import ray\n",
    "\n",
    "from ray import serve\n",
    "from ray.serve import PredictorDeployment\n",
    " \n",
    "import cifar_utils\n",
    "\n",
    "from ray.train.torch import TorchTrainer\n",
    "from ray.tune import Tuner\n",
    "from ray.tune import TuneConfig\n",
    "from ray.air.config import RunConfig\n",
    "from ray.air.config import ScalingConfig\n",
    "from ray.air.config import CheckpointConfig\n",
    "\n",
    "from ray.train.torch import TorchPredictor\n",
    "from ray.train.batch_predictor import BatchPredictor\n",
    "\n",
    "import torch\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import requests"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2023-01-03 16:27:59,384\tINFO worker.py:1529 -- Started a local Ray instance. View the dashboard at \u001b[1m\u001b[32mhttp://127.0.0.1:8265 \u001b[39m\u001b[22m\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "    <div style=\"margin-left: 50px;display: flex;flex-direction: row;align-items: center\">\n",
       "        <h3 style=\"color: var(--jp-ui-font-color0)\">Ray</h3>\n",
       "        <svg version=\"1.1\" id=\"ray\" width=\"3em\" viewBox=\"0 0 144.5 144.6\" style=\"margin-left: 3em;margin-right: 3em\">\n",
       "            <g id=\"layer-1\">\n",
       "                <path fill=\"#00a2e9\" class=\"st0\" d=\"M97.3,77.2c-3.8-1.1-6.2,0.9-8.3,5.1c-3.5,6.8-9.9,9.9-17.4,9.6S58,88.1,54.8,81.2c-1.4-3-3-4-6.3-4.1\n",
       "                    c-5.6-0.1-9.9,0.1-13.1,6.4c-3.8,7.6-13.6,10.2-21.8,7.6C5.2,88.4-0.4,80.5,0,71.7c0.1-8.4,5.7-15.8,13.8-18.2\n",
       "                    c8.4-2.6,17.5,0.7,22.3,8c1.3,1.9,1.3,5.2,3.6,5.6c3.9,0.6,8,0.2,12,0.2c1.8,0,1.9-1.6,2.4-2.8c3.5-7.8,9.7-11.8,18-11.9\n",
       "                    c8.2-0.1,14.4,3.9,17.8,11.4c1.3,2.8,2.9,3.6,5.7,3.3c1-0.1,2,0.1,3,0c2.8-0.5,6.4,1.7,8.1-2.7s-2.3-5.5-4.1-7.5\n",
       "                    c-5.1-5.7-10.9-10.8-16.1-16.3C84,38,81.9,37.1,78,38.3C66.7,42,56.2,35.7,53,24.1C50.3,14,57.3,2.8,67.7,0.5\n",
       "                    C78.4-2,89,4.7,91.5,15.3c0.1,0.3,0.1,0.5,0.2,0.8c0.7,3.4,0.7,6.9-0.8,9.8c-1.7,3.2-0.8,5,1.5,7.2c6.7,6.5,13.3,13,19.8,19.7\n",
       "                    c1.8,1.8,3,2.1,5.5,1.2c9.1-3.4,17.9-0.6,23.4,7c4.8,6.9,4.6,16.1-0.4,22.9c-5.4,7.2-14.2,9.9-23.1,6.5c-2.3-0.9-3.5-0.6-5.1,1.1\n",
       "                    c-6.7,6.9-13.6,13.7-20.5,20.4c-1.8,1.8-2.5,3.2-1.4,5.9c3.5,8.7,0.3,18.6-7.7,23.6c-7.9,5-18.2,3.8-24.8-2.9\n",
       "                    c-6.4-6.4-7.4-16.2-2.5-24.3c4.9-7.8,14.5-11,23.1-7.8c3,1.1,4.7,0.5,6.9-1.7C91.7,98.4,98,92.3,104.2,86c1.6-1.6,4.1-2.7,2.6-6.2\n",
       "                    c-1.4-3.3-3.8-2.5-6.2-2.6C99.8,77.2,98.9,77.2,97.3,77.2z M72.1,29.7c5.5,0.1,9.9-4.3,10-9.8c0-0.1,0-0.2,0-0.3\n",
       "                    C81.8,14,77,9.8,71.5,10.2c-5,0.3-9,4.2-9.3,9.2c-0.2,5.5,4,10.1,9.5,10.3C71.8,29.7,72,29.7,72.1,29.7z M72.3,62.3\n",
       "                    c-5.4-0.1-9.9,4.2-10.1,9.7c0,0.2,0,0.3,0,0.5c0.2,5.4,4.5,9.7,9.9,10c5.1,0.1,9.9-4.7,10.1-9.8c0.2-5.5-4-10-9.5-10.3\n",
       "                    C72.6,62.3,72.4,62.3,72.3,62.3z M115,72.5c0.1,5.4,4.5,9.7,9.8,9.9c5.6-0.2,10-4.8,10-10.4c-0.2-5.4-4.6-9.7-10-9.7\n",
       "                    c-5.3-0.1-9.8,4.2-9.9,9.5C115,72.1,115,72.3,115,72.5z M19.5,62.3c-5.4,0.1-9.8,4.4-10,9.8c-0.1,5.1,5.2,10.4,10.2,10.3\n",
       "                    c5.6-0.2,10-4.9,9.8-10.5c-0.1-5.4-4.5-9.7-9.9-9.6C19.6,62.3,19.5,62.3,19.5,62.3z M71.8,134.6c5.9,0.2,10.3-3.9,10.4-9.6\n",
       "                    c0.5-5.5-3.6-10.4-9.1-10.8c-5.5-0.5-10.4,3.6-10.8,9.1c0,0.5,0,0.9,0,1.4c-0.2,5.3,4,9.8,9.3,10\n",
       "                    C71.6,134.6,71.7,134.6,71.8,134.6z\"/>\n",
       "            </g>\n",
       "        </svg>\n",
       "        <table>\n",
       "            <tr>\n",
       "                <td style=\"text-align: left\"><b>Python version:</b></td>\n",
       "                <td style=\"text-align: left\"><b>3.8.13</b></td>\n",
       "            </tr>\n",
       "            <tr>\n",
       "                <td style=\"text-align: left\"><b>Ray version:</b></td>\n",
       "                <td style=\"text-align: left\"><b> 2.2.0</b></td>\n",
       "            </tr>\n",
       "            <tr>\n",
       "    <td style=\"text-align: left\"><b>Dashboard:</b></td>\n",
       "    <td style=\"text-align: left\"><b><a href=\"http://127.0.0.1:8265\" target=\"_blank\">http://127.0.0.1:8265</a></b></td>\n",
       "</tr>\n",
       "\n",
       "        </table>\n",
       "    </div>\n",
       "</div>\n"
      ],
      "text/plain": [
       "RayContext(dashboard_url='127.0.0.1:8265', python_version='3.8.13', ray_version='2.2.0', ray_commit='b6af0887ee5f2e460202133791ad941a41f15beb', address_info={'node_ip_address': '127.0.0.1', 'raylet_ip_address': '127.0.0.1', 'redis_address': None, 'object_store_address': '/tmp/ray/session_2023-01-03_16-27-57_332349_64338/sockets/plasma_store', 'raylet_socket_name': '/tmp/ray/session_2023-01-03_16-27-57_332349_64338/sockets/raylet', 'webui_url': '127.0.0.1:8265', 'session_dir': '/tmp/ray/session_2023-01-03_16-27-57_332349_64338', 'metrics_export_port': 62140, 'gcs_address': '127.0.0.1:63927', 'address': '127.0.0.1:63927', 'dashboard_agent_listen_port': 52365, 'node_id': '33257489bfc08dd7933c534e2127d12d45babc34ecd3e1b814f21fbb'})"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "if ray.is_initialized():\n",
    "    ray.shutdown()\n",
    "ray.init()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'CPU': 10.0,\n",
       " 'object_store_memory': 2147483648.0,\n",
       " 'node:127.0.0.1': 1.0,\n",
       " 'memory': 46051208397.0}"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ray.cluster_resources()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Step 1. Fetch Data\n",
    "\n",
    "Use Ray data to fetch data from a data source. Here we use a simple built-in `SimpleTorchDatasouce`. For preprocessing needs, such as scaling or normalizing, \n",
    "Ray AIR provides a host of preprocessors you can use; these [preprocessors](https://docs.ray.io/en/latest/ray-air/preprocessors.html) can be supplied downstream to `Trainers` and `BatchPredictors` to preprocess your \n",
    "input before training, tuning, or scoring, relieving you the onus of doing it yourself. \n",
    "\n",
    "<img src=\"images/data_prep.png\" width=\"360%\" height=\"30%\">"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Files already downloaded and verified\n",
      "Files already downloaded and verified\n"
     ]
    }
   ],
   "source": [
    "train_torchvision_dataset = cifar_utils.train_dataset_factory()\n",
    "test_torchvision_dataset = cifar_utils.test_dataset_factory()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(Dataset CIFAR10\n",
       "     Number of datapoints: 50000\n",
       "     Root location: /Users/jules/data\n",
       "     Split: Train\n",
       "     StandardTransform\n",
       " Transform: Compose(\n",
       "                ToTensor()\n",
       "                Normalize(mean=(0.5, 0.5, 0.5), std=(0.5, 0.5, 0.5))\n",
       "            ),\n",
       " torchvision.datasets.cifar.CIFAR10)"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_torchvision_dataset, type(train_torchvision_dataset) "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Note**: The above are PyTorch torchvision data types. We want to convert\n",
    "them into Ray Dataset types."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Step 2: Create a preprocessor to transform data\n",
    "* Convert Torchvisiion data into Ray data\n",
    "* Transform data into numpy so we can convert to Pandas DataFrame because internally Ray data uses Pandas by default as it internal representation\n",
    "* Use dataset `map_batches`() to convert "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(Dataset(num_blocks=200, num_rows=50000, schema=<class 'tuple'>),\n",
       " ray.data.dataset.Dataset)"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_dataset: ray.data.Dataset = ray.data.from_torch(train_torchvision_dataset)\n",
    "test_dataset: ray.data.Dataset = ray.data.from_torch(test_torchvision_dataset)\n",
    "train_dataset, type(train_dataset)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Map_Batches: 100%|██████████████████████████████████████████████████████████████████████████████████████| 200/200 [00:01<00:00, 108.76it/s]\n",
      "Map_Batches: 100%|██████████████████████████████████████████████████████████████████████████████████████| 200/200 [00:00<00:00, 535.46it/s]\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "(Dataset(num_blocks=200, num_rows=50000, schema={image: ArrowTensorType(shape=(3, 32, 32), dtype=float), label: int64}),\n",
       " ray.data.dataset.Dataset)"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_dataset = train_dataset.map_batches(cifar_utils.convert_batch_to_numpy)\n",
    "test_dataset = test_dataset.map_batches(cifar_utils.convert_batch_to_numpy)\n",
    "train_dataset, type(train_dataset)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Step 3: Create a `TorchTrainer` for PyTorch DDP training\n",
    "This trainer is then fed into tuner for doing some hyper-parameter optimization. Ray AIR supports popular [Trainers](https://docs.ray.io/en/latest/ray-air/trainer.html)\n",
    "for the common ML libraries: PyTorch, TensorFlow, XGBoost, scikit-learn. These can be used alongside [Tuner](https://docs.ray.io/en/latest/ray-air/tuner.html)\n",
    "\n",
    "<img src=\"images/trainer.png\" width=\"60%\" height=\"30%\">"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainer = TorchTrainer(train_loop_per_worker=cifar_utils.train_loop_per_worker,\n",
    "                        datasets={\"train\": train_dataset},\n",
    "                        scaling_config=ScalingConfig(num_workers=6, use_gpu=False), # change to True if on GPU\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Step 4: Create the Tuner \n",
    "\n",
    "Tuning a model over a hyperparameters space is a common workload and paramount to get better accuracy. Here we tune the relevant tune configuration such as `lr`, `batch_size`, `epochs`.\n",
    "However, you can use SOTA [search algorithms](https://docs.ray.io/en/latest/tune/api_docs/suggestion.html) and [schedulers](https://docs.ray.io/en/latest/tune/api_docs/schedulers.html) alongside [Tuner](https://docs.ray.io/en/latest/tune/api_docs/execution.html#ray.tune.Tuner) to control your trials, depending on your model requirements and desired results.\n",
    "\n",
    "\n",
    "<img src=\"images/tuner.png\" width=\"60%\" height=\"30%\">"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "tuner = Tuner(trainer,\n",
    "        param_space={\n",
    "            \"train_loop_config\": {\n",
    "                \"lr\": ray.tune.grid_search([0.01]),\n",
    "                \"epochs\": 15, \"batch_size\": 32,\n",
    "                \"verbose\": 1}\n",
    "        },\n",
    "        # specific tune metrics to collect and checkpoint\n",
    "        # during trials\n",
    "        tune_config=TuneConfig(metric=\"train_loss\", mode=\"min\"),\n",
    "        run_config=RunConfig(checkpoint_config=CheckpointConfig(num_to_keep=1, \n",
    "                        checkpoint_score_attribute=\"train_loss\", \n",
    "                        checkpoint_score_order=\"min\")\n",
    "        )\n",
    "    )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Once we have set up our Trainer and Tuner, all we need to create distributed training and trials using the familiar API .`.fit`() to launch our distributed training and tuning."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div class=\"tuneStatus\">\n",
       "  <div style=\"display: flex;flex-direction: row\">\n",
       "    <div style=\"display: flex;flex-direction: column;\">\n",
       "      <h3>Tune Status</h3>\n",
       "      <table>\n",
       "<tbody>\n",
       "<tr><td>Current time:</td><td>2023-01-03 20:11:44</td></tr>\n",
       "<tr><td>Running for: </td><td>03:41:36.22        </td></tr>\n",
       "<tr><td>Memory:      </td><td>21.6/64.0 GiB      </td></tr>\n",
       "</tbody>\n",
       "</table>\n",
       "    </div>\n",
       "    <div class=\"vDivider\"></div>\n",
       "    <div class=\"systemInfo\">\n",
       "      <h3>System Info</h3>\n",
       "      Using FIFO scheduling algorithm.<br>Resources requested: 7.0/10 CPUs, 0/0 GPUs, 0.0/42.89 GiB heap, 0.0/2.0 GiB objects\n",
       "    </div>\n",
       "    \n",
       "  </div>\n",
       "  <div class=\"hDivider\"></div>\n",
       "  <div class=\"trialStatus\">\n",
       "    <h3>Trial Status</h3>\n",
       "    <table>\n",
       "<thead>\n",
       "<tr><th>Trial name              </th><th>status  </th><th>loc            </th><th style=\"text-align: right;\">  train_loop_config/lr</th><th style=\"text-align: right;\">  iter</th><th style=\"text-align: right;\">  total time (s)</th><th style=\"text-align: right;\">  train_loss</th><th style=\"text-align: right;\">  _timestamp</th><th style=\"text-align: right;\">  _time_this_iter_s</th></tr>\n",
       "</thead>\n",
       "<tbody>\n",
       "<tr><td>TorchTrainer_f133b_00000</td><td>RUNNING </td><td>127.0.0.1:82654</td><td style=\"text-align: right;\">                  0.01</td><td style=\"text-align: right;\">    14</td><td style=\"text-align: right;\">         12917.4</td><td style=\"text-align: right;\">     0.09737</td><td style=\"text-align: right;\">  1672805127</td><td style=\"text-align: right;\">            451.827</td></tr>\n",
       "</tbody>\n",
       "</table>\n",
       "  </div>\n",
       "</div>\n",
       "<style>\n",
       ".tuneStatus {\n",
       "  color: var(--jp-ui-font-color1);\n",
       "}\n",
       ".tuneStatus .systemInfo {\n",
       "  display: flex;\n",
       "  flex-direction: column;\n",
       "}\n",
       ".tuneStatus td {\n",
       "  white-space: nowrap;\n",
       "}\n",
       ".tuneStatus .trialStatus {\n",
       "  display: flex;\n",
       "  flex-direction: column;\n",
       "}\n",
       ".tuneStatus h3 {\n",
       "  font-weight: bold;\n",
       "}\n",
       ".tuneStatus .hDivider {\n",
       "  border-bottom-width: var(--jp-border-width);\n",
       "  border-bottom-color: var(--jp-border-color0);\n",
       "  border-bottom-style: solid;\n",
       "}\n",
       ".tuneStatus .vDivider {\n",
       "  border-left-width: var(--jp-border-width);\n",
       "  border-left-color: var(--jp-border-color0);\n",
       "  border-left-style: solid;\n",
       "  margin: 0.5em 1em 0.5em 1em;\n",
       "}\n",
       "</style>\n"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82663)\u001b[0m 2023-01-03 16:30:12,809\tINFO config.py:86 -- Setting up process group for: env:// [rank=0, world_size=6]\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82665)\u001b[0m [W ProcessGroupGloo.cpp:724] Warning: Unable to resolve hostname to a (local) address. Using the loopback address as fallback. Manually set the network interface to bind to with GLOO_SOCKET_IFNAME. (function operator())\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82667)\u001b[0m [W ProcessGroupGloo.cpp:724] Warning: Unable to resolve hostname to a (local) address. Using the loopback address as fallback. Manually set the network interface to bind to with GLOO_SOCKET_IFNAME. (function operator())\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82666)\u001b[0m [W ProcessGroupGloo.cpp:724] Warning: Unable to resolve hostname to a (local) address. Using the loopback address as fallback. Manually set the network interface to bind to with GLOO_SOCKET_IFNAME. (function operator())\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82664)\u001b[0m [W ProcessGroupGloo.cpp:724] Warning: Unable to resolve hostname to a (local) address. Using the loopback address as fallback. Manually set the network interface to bind to with GLOO_SOCKET_IFNAME. (function operator())\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82663)\u001b[0m [W ProcessGroupGloo.cpp:724] Warning: Unable to resolve hostname to a (local) address. Using the loopback address as fallback. Manually set the network interface to bind to with GLOO_SOCKET_IFNAME. (function operator())\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82668)\u001b[0m [W ProcessGroupGloo.cpp:724] Warning: Unable to resolve hostname to a (local) address. Using the loopback address as fallback. Manually set the network interface to bind to with GLOO_SOCKET_IFNAME. (function operator())\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82663)\u001b[0m 2023-01-03 16:30:15,551\tINFO train_loop_utils.py:270 -- Moving model to device: cpu\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82663)\u001b[0m 2023-01-03 16:30:15,552\tINFO train_loop_utils.py:330 -- Wrapping provided model in DistributedDataParallel.\n",
      "  0%|          | 0/15 [00:00<?, ?it/s]m \n",
      "  0%|          | 0/15 [00:00<?, ?it/s]m \n",
      "  0%|          | 0/15 [00:00<?, ?it/s]m \n",
      "  0%|          | 0/15 [00:00<?, ?it/s]m \n",
      "  0%|          | 0/15 [00:00<?, ?it/s]m \n",
      "  0%|          | 0/15 [00:00<?, ?it/s]m \n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82666)\u001b[0m Training epoch:1/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82667)\u001b[0m Training epoch:1/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82665)\u001b[0m Training epoch:1/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82668)\u001b[0m Training epoch:1/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82663)\u001b[0m Training epoch:1/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82664)\u001b[0m Training epoch:1/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82666)\u001b[0m training loss: 1.448 | epoch: 1/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82667)\u001b[0m training loss: 1.451 | epoch: 1/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82665)\u001b[0m training loss: 1.441 | epoch: 1/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82668)\u001b[0m training loss: 1.440 | epoch: 1/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82663)\u001b[0m training loss: 1.442 | epoch: 1/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82664)\u001b[0m training loss: 1.437 | epoch: 1/15 | batch: 261\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div class=\"trialProgress\">\n",
       "  <h3>Trial Progress</h3>\n",
       "  <table>\n",
       "<thead>\n",
       "<tr><th>Trial name              </th><th style=\"text-align: right;\">  _time_this_iter_s</th><th style=\"text-align: right;\">  _timestamp</th><th style=\"text-align: right;\">  _training_iteration</th><th>date               </th><th>done  </th><th>episodes_total  </th><th>experiment_id                   </th><th>hostname             </th><th style=\"text-align: right;\">  iterations_since_restore</th><th>node_ip  </th><th style=\"text-align: right;\">  pid</th><th>should_checkpoint  </th><th style=\"text-align: right;\">  time_since_restore</th><th style=\"text-align: right;\">  time_this_iter_s</th><th style=\"text-align: right;\">  time_total_s</th><th style=\"text-align: right;\">  timestamp</th><th style=\"text-align: right;\">  timesteps_since_restore</th><th>timesteps_total  </th><th style=\"text-align: right;\">  train_loss</th><th style=\"text-align: right;\">  training_iteration</th><th>trial_id   </th><th style=\"text-align: right;\">  warmup_time</th></tr>\n",
       "</thead>\n",
       "<tbody>\n",
       "<tr><td>TorchTrainer_f133b_00000</td><td style=\"text-align: right;\">            451.827</td><td style=\"text-align: right;\">  1672805127</td><td style=\"text-align: right;\">                   14</td><td>2023-01-03_20-05-27</td><td>False </td><td>                </td><td>a7d2c21a29224497bbfe5e5a130030fc</td><td>Juless-MacBook-Pro-16</td><td style=\"text-align: right;\">                        14</td><td>127.0.0.1</td><td style=\"text-align: right;\">82654</td><td>True               </td><td style=\"text-align: right;\">             12917.4</td><td style=\"text-align: right;\">           451.827</td><td style=\"text-align: right;\">       12917.4</td><td style=\"text-align: right;\"> 1672805127</td><td style=\"text-align: right;\">                        0</td><td>                 </td><td style=\"text-align: right;\">     0.09737</td><td style=\"text-align: right;\">                  14</td><td>f133b_00000</td><td style=\"text-align: right;\">     0.313893</td></tr>\n",
       "</tbody>\n",
       "</table>\n",
       "</div>\n",
       "<style>\n",
       ".trialProgress {\n",
       "  display: flex;\n",
       "  flex-direction: column;\n",
       "  color: var(--jp-ui-font-color1);\n",
       "}\n",
       ".trialProgress h3 {\n",
       "  font-weight: bold;\n",
       "}\n",
       ".trialProgress td {\n",
       "  white-space: nowrap;\n",
       "}\n",
       "</style>\n"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82666)\u001b[0m Training epoch:2/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82667)\u001b[0m Training epoch:2/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82665)\u001b[0m Training epoch:2/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82668)\u001b[0m Training epoch:2/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82663)\u001b[0m Training epoch:2/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82664)\u001b[0m Training epoch:2/15 | batch_size:32 | lr:0.01\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "  7%|▋         | 1/15 [07:09<1:40:19, 429.93s/it]\n",
      "  7%|▋         | 1/15 [07:09<1:40:19, 429.93s/it]\n",
      "  7%|▋         | 1/15 [07:09<1:40:18, 429.93s/it]\n",
      "  7%|▋         | 1/15 [07:09<1:40:19, 429.93s/it]\n",
      "  7%|▋         | 1/15 [07:09<1:40:19, 429.93s/it]\n",
      "  7%|▋         | 1/15 [07:09<1:40:19, 429.93s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82666)\u001b[0m training loss: 0.948 | epoch: 2/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82667)\u001b[0m training loss: 0.951 | epoch: 2/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82665)\u001b[0m training loss: 0.927 | epoch: 2/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82668)\u001b[0m training loss: 0.933 | epoch: 2/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82663)\u001b[0m training loss: 0.946 | epoch: 2/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82664)\u001b[0m training loss: 0.935 | epoch: 2/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82666)\u001b[0m Training epoch:3/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82667)\u001b[0m Training epoch:3/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82665)\u001b[0m Training epoch:3/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82668)\u001b[0m Training epoch:3/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82663)\u001b[0m Training epoch:3/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82664)\u001b[0m Training epoch:3/15 | batch_size:32 | lr:0.01\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 13%|█▎        | 2/15 [14:09<1:31:48, 423.71s/it]\n",
      " 13%|█▎        | 2/15 [14:09<1:31:48, 423.71s/it]\n",
      " 13%|█▎        | 2/15 [14:09<1:31:48, 423.71s/it]\n",
      " 13%|█▎        | 2/15 [14:09<1:31:48, 423.71s/it]\n",
      " 13%|█▎        | 2/15 [14:09<1:31:48, 423.71s/it]\n",
      " 13%|█▎        | 2/15 [14:09<1:31:48, 423.71s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82665)\u001b[0m training loss: 0.697 | epoch: 3/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82663)\u001b[0m training loss: 0.721 | epoch: 3/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82666)\u001b[0m training loss: 0.714 | epoch: 3/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82667)\u001b[0m training loss: 0.736 | epoch: 3/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82668)\u001b[0m training loss: 0.717 | epoch: 3/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82664)\u001b[0m training loss: 0.723 | epoch: 3/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82666)\u001b[0m Training epoch:4/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82667)\u001b[0m Training epoch:4/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82665)\u001b[0m Training epoch:4/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82668)\u001b[0m Training epoch:4/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82663)\u001b[0m Training epoch:4/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82664)\u001b[0m Training epoch:4/15 | batch_size:32 | lr:0.01\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 20%|██        | 3/15 [25:57<1:50:44, 553.73s/it]\n",
      " 20%|██        | 3/15 [25:57<1:50:44, 553.73s/it]\n",
      " 20%|██        | 3/15 [25:57<1:50:44, 553.73s/it]\n",
      " 20%|██        | 3/15 [25:57<1:50:44, 553.73s/it]\n",
      " 20%|██        | 3/15 [25:57<1:50:44, 553.73s/it]\n",
      " 20%|██        | 3/15 [25:57<1:50:44, 553.73s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82666)\u001b[0m training loss: 0.578 | epoch: 4/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82667)\u001b[0m training loss: 0.589 | epoch: 4/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82665)\u001b[0m training loss: 0.562 | epoch: 4/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82668)\u001b[0m training loss: 0.589 | epoch: 4/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82663)\u001b[0m training loss: 0.581 | epoch: 4/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82664)\u001b[0m training loss: 0.593 | epoch: 4/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82666)\u001b[0m Training epoch:5/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82667)\u001b[0m Training epoch:5/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82665)\u001b[0m Training epoch:5/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82668)\u001b[0m Training epoch:5/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82663)\u001b[0m Training epoch:5/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82664)\u001b[0m Training epoch:5/15 | batch_size:32 | lr:0.01\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 27%|██▋       | 4/15 [33:31<1:34:15, 514.16s/it]\n",
      " 27%|██▋       | 4/15 [33:31<1:34:15, 514.16s/it]\n",
      " 27%|██▋       | 4/15 [33:31<1:34:15, 514.16s/it]\n",
      " 27%|██▋       | 4/15 [33:31<1:34:15, 514.16s/it]\n",
      " 27%|██▋       | 4/15 [33:31<1:34:15, 514.16s/it]\n",
      " 27%|██▋       | 4/15 [33:31<1:34:15, 514.16s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82666)\u001b[0m training loss: 0.475 | epoch: 5/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82667)\u001b[0m training loss: 0.478 | epoch: 5/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82665)\u001b[0m training loss: 0.464 | epoch: 5/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82668)\u001b[0m training loss: 0.483 | epoch: 5/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82663)\u001b[0m training loss: 0.476 | epoch: 5/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82664)\u001b[0m training loss: 0.487 | epoch: 5/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82666)\u001b[0m Training epoch:6/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82667)\u001b[0m Training epoch:6/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82665)\u001b[0m Training epoch:6/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82668)\u001b[0m Training epoch:6/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82663)\u001b[0m Training epoch:6/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82664)\u001b[0m Training epoch:6/15 | batch_size:32 | lr:0.01\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 33%|███▎      | 5/15 [41:07<1:22:11, 493.13s/it]\n",
      " 33%|███▎      | 5/15 [41:07<1:22:11, 493.13s/it]\n",
      " 33%|███▎      | 5/15 [41:07<1:22:11, 493.12s/it]\n",
      " 33%|███▎      | 5/15 [41:07<1:22:11, 493.12s/it]\n",
      " 33%|███▎      | 5/15 [41:07<1:22:11, 493.12s/it]\n",
      " 33%|███▎      | 5/15 [41:07<1:22:11, 493.12s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82666)\u001b[0m training loss: 0.393 | epoch: 6/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82667)\u001b[0m training loss: 0.400 | epoch: 6/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82665)\u001b[0m training loss: 0.390 | epoch: 6/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82668)\u001b[0m training loss: 0.400 | epoch: 6/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82663)\u001b[0m training loss: 0.403 | epoch: 6/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82664)\u001b[0m training loss: 0.407 | epoch: 6/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82666)\u001b[0m Training epoch:7/15 | batch_size:32 | lr:0.01\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 40%|████      | 6/15 [48:40<1:11:56, 479.58s/it]\n",
      " 40%|████      | 6/15 [48:40<1:11:56, 479.58s/it]\n",
      " 40%|████      | 6/15 [48:40<1:11:56, 479.58s/it]\n",
      " 40%|████      | 6/15 [48:40<1:11:56, 479.58s/it]\n",
      " 40%|████      | 6/15 [48:40<1:11:56, 479.58s/it]\n",
      " 40%|████      | 6/15 [48:40<1:11:56, 479.58s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82667)\u001b[0m Training epoch:7/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82665)\u001b[0m Training epoch:7/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82668)\u001b[0m Training epoch:7/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82663)\u001b[0m Training epoch:7/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82664)\u001b[0m Training epoch:7/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82666)\u001b[0m training loss: 0.321 | epoch: 7/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82667)\u001b[0m training loss: 0.331 | epoch: 7/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82665)\u001b[0m training loss: 0.319 | epoch: 7/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82668)\u001b[0m training loss: 0.326 | epoch: 7/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82663)\u001b[0m training loss: 0.333 | epoch: 7/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82664)\u001b[0m training loss: 0.331 | epoch: 7/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82666)\u001b[0m Training epoch:8/15 | batch_size:32 | lr:0.01\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 47%|████▋     | 7/15 [56:06<1:02:29, 468.74s/it]\n",
      " 47%|████▋     | 7/15 [56:06<1:02:29, 468.74s/it]\n",
      " 47%|████▋     | 7/15 [56:06<1:02:29, 468.74s/it]\n",
      " 47%|████▋     | 7/15 [56:06<1:02:29, 468.74s/it]\n",
      " 47%|████▋     | 7/15 [56:06<1:02:29, 468.74s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82665)\u001b[0m Training epoch:8/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82668)\u001b[0m Training epoch:8/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82663)\u001b[0m Training epoch:8/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82664)\u001b[0m Training epoch:8/15 | batch_size:32 | lr:0.01\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 47%|████▋     | 7/15 [56:06<1:02:30, 468.77s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82667)\u001b[0m Training epoch:8/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82666)\u001b[0m training loss: 0.260 | epoch: 8/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82667)\u001b[0m training loss: 0.262 | epoch: 8/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82665)\u001b[0m training loss: 0.254 | epoch: 8/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82668)\u001b[0m training loss: 0.266 | epoch: 8/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82663)\u001b[0m training loss: 0.286 | epoch: 8/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82664)\u001b[0m training loss: 0.266 | epoch: 8/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82666)\u001b[0m Training epoch:9/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82667)\u001b[0m Training epoch:9/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82665)\u001b[0m Training epoch:9/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82668)\u001b[0m Training epoch:9/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82663)\u001b[0m Training epoch:9/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82664)\u001b[0m Training epoch:9/15 | batch_size:32 | lr:0.01\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 53%|█████▎    | 8/15 [1:18:00<1:26:03, 737.68s/it]\n",
      " 53%|█████▎    | 8/15 [1:18:00<1:26:03, 737.69s/it]\n",
      " 53%|█████▎    | 8/15 [1:18:00<1:26:03, 737.69s/it]\n",
      " 53%|█████▎    | 8/15 [1:18:00<1:26:03, 737.69s/it]\n",
      " 53%|█████▎    | 8/15 [1:18:00<1:26:03, 737.69s/it]\n",
      " 53%|█████▎    | 8/15 [1:18:00<1:26:03, 737.69s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82666)\u001b[0m training loss: 0.220 | epoch: 9/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82667)\u001b[0m training loss: 0.228 | epoch: 9/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82665)\u001b[0m training loss: 0.221 | epoch: 9/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82668)\u001b[0m training loss: 0.227 | epoch: 9/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82663)\u001b[0m training loss: 0.242 | epoch: 9/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82664)\u001b[0m training loss: 0.236 | epoch: 9/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82666)\u001b[0m Training epoch:10/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82667)\u001b[0m Training epoch:10/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82665)\u001b[0m Training epoch:10/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82668)\u001b[0m Training epoch:10/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82663)\u001b[0m Training epoch:10/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82664)\u001b[0m Training epoch:10/15 | batch_size:32 | lr:0.01\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 60%|██████    | 9/15 [1:25:29<1:04:45, 647.60s/it]\n",
      " 60%|██████    | 9/15 [1:25:29<1:04:45, 647.60s/it]\n",
      " 60%|██████    | 9/15 [1:25:29<1:04:45, 647.60s/it]\n",
      " 60%|██████    | 9/15 [1:25:29<1:04:45, 647.60s/it]\n",
      " 60%|██████    | 9/15 [1:25:29<1:04:45, 647.60s/it]\n",
      " 60%|██████    | 9/15 [1:25:29<1:04:45, 647.60s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82666)\u001b[0m training loss: 0.173 | epoch: 10/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82667)\u001b[0m training loss: 0.190 | epoch: 10/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82665)\u001b[0m training loss: 0.175 | epoch: 10/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82668)\u001b[0m training loss: 0.187 | epoch: 10/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82663)\u001b[0m training loss: 0.194 | epoch: 10/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82664)\u001b[0m training loss: 0.183 | epoch: 10/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82666)\u001b[0m Training epoch:11/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82667)\u001b[0m Training epoch:11/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82665)\u001b[0m Training epoch:11/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82668)\u001b[0m Training epoch:11/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82663)\u001b[0m Training epoch:11/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82664)\u001b[0m Training epoch:11/15 | batch_size:32 | lr:0.01\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 67%|██████▋   | 10/15 [2:00:46<1:31:44, 1100.97s/it]\n",
      " 67%|██████▋   | 10/15 [2:00:46<1:31:44, 1100.97s/it]\n",
      " 67%|██████▋   | 10/15 [2:00:46<1:31:44, 1100.97s/it]\n",
      " 67%|██████▋   | 10/15 [2:00:46<1:31:44, 1100.98s/it]\n",
      " 67%|██████▋   | 10/15 [2:00:46<1:31:44, 1100.97s/it]\n",
      " 67%|██████▋   | 10/15 [2:00:46<1:31:44, 1100.97s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82666)\u001b[0m training loss: 0.160 | epoch: 11/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82667)\u001b[0m training loss: 0.158 | epoch: 11/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82665)\u001b[0m training loss: 0.155 | epoch: 11/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82668)\u001b[0m training loss: 0.166 | epoch: 11/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82663)\u001b[0m training loss: 0.163 | epoch: 11/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82664)\u001b[0m training loss: 0.164 | epoch: 11/15 | batch: 261\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 73%|███████▎  | 11/15 [3:12:44<2:19:02, 2085.63s/it]\n",
      " 73%|███████▎  | 11/15 [3:12:44<2:19:02, 2085.66s/it]\n",
      " 73%|███████▎  | 11/15 [3:12:44<2:19:02, 2085.66s/it]\n",
      " 73%|███████▎  | 11/15 [3:12:44<2:19:02, 2085.67s/it]\n",
      " 73%|███████▎  | 11/15 [3:12:44<2:19:02, 2085.66s/it]\n",
      " 73%|███████▎  | 11/15 [3:12:44<2:19:02, 2085.66s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82666)\u001b[0m Training epoch:12/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82667)\u001b[0m Training epoch:12/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82665)\u001b[0m Training epoch:12/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82668)\u001b[0m Training epoch:12/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82663)\u001b[0m Training epoch:12/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82664)\u001b[0m Training epoch:12/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82666)\u001b[0m training loss: 0.146 | epoch: 12/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82667)\u001b[0m training loss: 0.130 | epoch: 12/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82665)\u001b[0m training loss: 0.133 | epoch: 12/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82668)\u001b[0m training loss: 0.139 | epoch: 12/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82663)\u001b[0m training loss: 0.140 | epoch: 12/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82664)\u001b[0m training loss: 0.144 | epoch: 12/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82666)\u001b[0m Training epoch:13/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82667)\u001b[0m Training epoch:13/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82665)\u001b[0m Training epoch:13/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82668)\u001b[0m Training epoch:13/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82663)\u001b[0m Training epoch:13/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82664)\u001b[0m Training epoch:13/15 | batch_size:32 | lr:0.01\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 80%|████████  | 12/15 [3:20:15<1:19:25, 1588.36s/it]\n",
      " 80%|████████  | 12/15 [3:20:15<1:19:25, 1588.35s/it]\n",
      " 80%|████████  | 12/15 [3:20:15<1:19:25, 1588.35s/it]\n",
      " 80%|████████  | 12/15 [3:20:15<1:19:25, 1588.35s/it]\n",
      " 80%|████████  | 12/15 [3:20:15<1:19:25, 1588.35s/it]\n",
      " 80%|████████  | 12/15 [3:20:15<1:19:25, 1588.35s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82666)\u001b[0m training loss: 0.106 | epoch: 13/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82667)\u001b[0m training loss: 0.119 | epoch: 13/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82665)\u001b[0m training loss: 0.105 | epoch: 13/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82668)\u001b[0m training loss: 0.121 | epoch: 13/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82663)\u001b[0m training loss: 0.111 | epoch: 13/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82664)\u001b[0m training loss: 0.111 | epoch: 13/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82666)\u001b[0m Training epoch:14/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82667)\u001b[0m Training epoch:14/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82665)\u001b[0m Training epoch:14/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82668)\u001b[0m Training epoch:14/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82663)\u001b[0m Training epoch:14/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82664)\u001b[0m Training epoch:14/15 | batch_size:32 | lr:0.01\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 87%|████████▋ | 13/15 [3:27:40<41:24, 1242.02s/it]  \n",
      " 87%|████████▋ | 13/15 [3:27:40<41:24, 1242.02s/it]  \n",
      " 87%|████████▋ | 13/15 [3:27:40<41:24, 1242.02s/it]  \n",
      " 87%|████████▋ | 13/15 [3:27:40<41:24, 1242.02s/it]  \n",
      " 87%|████████▋ | 13/15 [3:27:40<41:24, 1242.02s/it]  \n",
      " 87%|████████▋ | 13/15 [3:27:40<41:24, 1242.02s/it]  \n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82666)\u001b[0m training loss: 0.092 | epoch: 14/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82667)\u001b[0m training loss: 0.102 | epoch: 14/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82665)\u001b[0m training loss: 0.093 | epoch: 14/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82668)\u001b[0m training loss: 0.090 | epoch: 14/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82663)\u001b[0m training loss: 0.097 | epoch: 14/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82664)\u001b[0m training loss: 0.092 | epoch: 14/15 | batch: 261\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82666)\u001b[0m Training epoch:15/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82667)\u001b[0m Training epoch:15/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82665)\u001b[0m Training epoch:15/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82668)\u001b[0m Training epoch:15/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82663)\u001b[0m Training epoch:15/15 | batch_size:32 | lr:0.01\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82664)\u001b[0m Training epoch:15/15 | batch_size:32 | lr:0.01\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 93%|█████████▎| 14/15 [3:35:12<16:43, 1003.35s/it]\n",
      " 93%|█████████▎| 14/15 [3:35:12<16:43, 1003.34s/it]\n",
      " 93%|█████████▎| 14/15 [3:35:12<16:43, 1003.34s/it]\n",
      " 93%|█████████▎| 14/15 [3:35:12<16:43, 1003.34s/it]\n",
      " 93%|█████████▎| 14/15 [3:35:12<16:43, 1003.34s/it]\n",
      " 93%|█████████▎| 14/15 [3:35:12<16:43, 1003.34s/it]\n"
     ]
    }
   ],
   "source": [
    "# Run the tuner, which will call trainer for each trial with\n",
    "# the parameters configurations as part of its HPO and trains it in parallel\n",
    "# with a shard of its own data. This is distributed training and tunning in parallel\n",
    "\n",
    "results = tuner.fit()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "TorchCheckpoint(local_path=/Users/jules/ray_results/TorchTrainer_2023-01-03_16-29-34/TorchTrainer_f133b_00000_0_lr=0.0100_2023-01-03_16-30-08/checkpoint_000014)"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82667)\u001b[0m /opt/homebrew/Caskroom/miniforge/base/envs/ray-core-tutorial/lib/python3.8/multiprocessing/resource_tracker.py:216: UserWarning: resource_tracker: There appear to be 1 leaked semaphore objects to clean up at shutdown\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82667)\u001b[0m   warnings.warn('resource_tracker: There appear to be %d '\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82664)\u001b[0m /opt/homebrew/Caskroom/miniforge/base/envs/ray-core-tutorial/lib/python3.8/multiprocessing/resource_tracker.py:216: UserWarning: resource_tracker: There appear to be 1 leaked semaphore objects to clean up at shutdown\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82664)\u001b[0m   warnings.warn('resource_tracker: There appear to be %d '\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82663)\u001b[0m /opt/homebrew/Caskroom/miniforge/base/envs/ray-core-tutorial/lib/python3.8/multiprocessing/resource_tracker.py:216: UserWarning: resource_tracker: There appear to be 1 leaked semaphore objects to clean up at shutdown\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82663)\u001b[0m   warnings.warn('resource_tracker: There appear to be %d '\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82668)\u001b[0m /opt/homebrew/Caskroom/miniforge/base/envs/ray-core-tutorial/lib/python3.8/multiprocessing/resource_tracker.py:216: UserWarning: resource_tracker: There appear to be 1 leaked semaphore objects to clean up at shutdown\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82668)\u001b[0m   warnings.warn('resource_tracker: There appear to be %d '\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82665)\u001b[0m /opt/homebrew/Caskroom/miniforge/base/envs/ray-core-tutorial/lib/python3.8/multiprocessing/resource_tracker.py:216: UserWarning: resource_tracker: There appear to be 1 leaked semaphore objects to clean up at shutdown\n",
      "\u001b[2m\u001b[36m(RayTrainWorker pid=82665)\u001b[0m   warnings.warn('resource_tracker: There appear to be %d '\n"
     ]
    }
   ],
   "source": [
    "# Get the best checkpoint result\n",
    "best_checkpoint = results.get_best_result(metric=\"train_loss\", mode=\"min\").checkpoint\n",
    "best_checkpoint"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "best_checkpoint.to_directory(\"./best_checkpoint\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Step 5: Use `TorchPredictor` using the `Checkpoint` object\n",
    "The `TorchPredictor` allows batch inference at scale:\n",
    "* Test our model with [`TorchPredictor`](https://docs.ray.io/en/latest/_modules/ray/train/torch/torch_predictor.html?highlight=TorchPredictor) using the [`Checkpoint`](https://docs.ray.io/en/latest/ray-air/key-concepts.html#checkpoints) object.\n",
    "* Fetch the best_checkpoint from the  checkpoint. \n",
    "* Use batch predictor to test the entire batch in one go\n",
    "\n",
    "\n",
    "<img src=\"images/batch_predictor.png\" width=\"60%\" height=\"30%\">"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Map_Batches: 100%|██████████████████████████████████████████████████████████████████████████████████████| 200/200 [00:01<00:00, 137.85it/s]\n"
     ]
    }
   ],
   "source": [
    "# drop the label column\n",
    "predict_dataset = test_dataset.drop_columns(cols=[\"label\"])\n",
    "\n",
    "# Create our BatchPredictor from the best checkpoint obtained above after all the trials are finished\n",
    "batch_predictor = BatchPredictor.from_checkpoint(\n",
    "    checkpoint=best_checkpoint,\n",
    "    predictor_cls=TorchPredictor,\n",
    "    model=cifar_utils.ResNet18(),\n",
    "    use_gpu=False # set it to True if GPU available\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2023-01-03 20:14:00,527\tWARNING compute.py:520 -- `batch_size` is set to 4096, which reduces parallelism from 200 to 3. If the performance is worse than expected, this may indicate that the batch size is too large or the input block size is too small. To reduce batch size, consider decreasing `batch_size` or use the default in `map_batches`. To increase input block size, consider decreasing `parallelism` in read.\n",
      "Map_Batches:   0%|                                                                                                   | 0/3 [00:00<?, ?it/s]2023-01-03 20:14:00,830\tWARNING worker.py:1851 -- Warning: The actor BlockWorker is very large (42 MiB). Check that its definition is not implicitly capturing a large array or other object in scope. Tip: use ray.put() to put large objects in the Ray object store.\n",
      "Map Progress (10 actors 3 pending):   0%|                                                                            | 0/3 [00:12<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[2m\u001b[1m\u001b[33m(scheduler +5h7m15s)\u001b[0m Warning: The following resource request cannot be scheduled right now: {'CPU': 1.0}. This is likely due to all cluster resources being claimed by actors. Consider creating fewer actors or adding more nodes to this Ray cluster.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Map Progress (10 actors 3 pending):  33%|██████████████████████▋                                             | 1/3 [00:30<01:00, 30.10s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[2m\u001b[1m\u001b[33m(scheduler +5h7m50s)\u001b[0m Warning: The following resource request cannot be scheduled right now: {'CPU': 1.0}. This is likely due to all cluster resources being claimed by actors. Consider creating fewer actors or adding more nodes to this Ray cluster.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Map Progress (10 actors 3 pending): 100%|████████████████████████████████████████████████████████████████████| 3/3 [01:29<00:00, 29.97s/it]\n"
     ]
    }
   ],
   "source": [
    "# Do prediction at scale over the entire batch\n",
    "output: ray.data.Dataset = batch_predictor.predict(\n",
    "    data=test_dataset, dtype=torch.float, \n",
    "    feature_columns=[\"image\"], \n",
    "    keep_columns=[\"label\"],\n",
    "    #num_gpus_per_worker=1,       # uncomment if CUDA available\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Map_Batches: 100%|███████████████████████████████████████████████████████████████████████████████████████████| 3/3 [00:01<00:00,  1.86it/s]\n",
      "Map_Batches: 100%|██████████████████████████████████████████████████████████████████████████████████████████| 3/3 [00:00<00:00, 267.28it/s]\n",
      "Shuffle Map: 100%|██████████████████████████████████████████████████████████████████████████████████████████| 3/3 [00:00<00:00, 376.79it/s]\n",
      "Shuffle Reduce: 100%|███████████████████████████████████████████████████████████████████████████████████████| 1/1 [00:00<00:00, 192.71it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Prediction accuracy from the test data of 10,000 images: 83.66%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "# Get the predictions from the output and convert them into classes\n",
    "predictions = output.map_batches(cifar_utils.convert_logits_to_classes, batch_format=\"pandas\")\n",
    "\n",
    "# Get all predictions for test_dataset \n",
    "scores = predictions.map_batches(cifar_utils.calculate_prediction_scores)\n",
    "\n",
    "# compute total prediction accuracy. That is all predictions equal to ground truth\n",
    "# That is, predictated accurately.\n",
    "total_acc = scores.sum(on=\"correct\") / scores.count()\n",
    "\n",
    "print(f\"Prediction accuracy from the test data of 10,000 images: {total_acc * 100:.2f}%\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Step 6: Online prediction\n",
    " * Deploy our model to the network for online prediction\n",
    " * Use `TorchPredictor` and [`PredictorDeployment`](https://docs.ray.io/en/latest/ray-air/package-ref.html?highlight=PredictorDeployment#ray.serve.air_integrations.PredictorDeployment) APIs\n",
    " * Start a Ray serve in detached mode, with two replicas\n",
    " \n",
    " This will automatically deploy the model from the lastest checkpoint. You can adjust num of replicas in the options.\n",
    " \n",
    " <img src=\"images/online_predictor.png\" width=\"60%\" height=\"25%\">\n",
    " "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\u001b[2m\u001b[36m(ServeController pid=9403)\u001b[0m INFO 2023-01-03 20:16:01,722 controller 9403 http_state.py:129 - Starting HTTP proxy with name 'SERVE_CONTROLLER_ACTOR:SERVE_PROXY_ACTOR-33257489bfc08dd7933c534e2127d12d45babc34ecd3e1b814f21fbb' on node '33257489bfc08dd7933c534e2127d12d45babc34ecd3e1b814f21fbb' listening on '127.0.0.1:8000'\n",
      "\u001b[2m\u001b[36m(HTTPProxyActor pid=9405)\u001b[0m INFO:     Started server process [9405]\n",
      "\u001b[2m\u001b[36m(ServeController pid=9403)\u001b[0m INFO 2023-01-03 20:16:03,294 controller 9403 deployment_state.py:1310 - Adding 2 replicas to deployment 'cifar-deployment'.\n"
     ]
    }
   ],
   "source": [
    "# Start Ray Serve in detached mode so that if this application terminates, Serve can still process the incoming\n",
    "# requests. \n",
    "serve.start(detached=True)\n",
    "\n",
    "# create the PredictorDeployment and give it a name and number of replicas. Depending on your\n",
    "# request loads, you can scale up the number of replicas\n",
    "deployment = PredictorDeployment.options(name=\"cifar-deployment\", num_replicas=2)\n",
    "deployment.deploy(TorchPredictor, best_checkpoint, batching_params=False, model=cifar_utils.ResNet18(), http_adapter=cifar_utils.json_to_numpy)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Test online prediction from the test data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "prediction: 3; class: cat; matched: True\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\u001b[2m\u001b[36m(HTTPProxyActor pid=9405)\u001b[0m INFO 2023-01-03 20:16:06,490 http_proxy 127.0.0.1 http_proxy.py:361 - POST /cifar-deployment 307 42.0ms\n",
      "\u001b[2m\u001b[36m(HTTPProxyActor pid=9405)\u001b[0m INFO 2023-01-03 20:16:06,512 http_proxy 127.0.0.1 http_proxy.py:361 - POST /cifar-deployment 200 20.8ms\n",
      "\u001b[2m\u001b[36m(ServeReplica:cifar-deployment pid=9416)\u001b[0m INFO 2023-01-03 20:16:06,511 cifar-deployment cifar-deployment#rPepjh replica.py:505 - HANDLE __call__ OK 17.5ms\n",
      "\u001b[2m\u001b[36m(ServeReplica:cifar-deployment pid=9415)\u001b[0m INFO 2023-01-03 20:16:06,489 cifar-deployment cifar-deployment#BZpGjK replica.py:505 - HANDLE __call__ OK 0.2ms\n",
      "\u001b[2m\u001b[36m(HTTPProxyActor pid=9405)\u001b[0m INFO 2023-01-03 20:16:06,607 http_proxy 127.0.0.1 http_proxy.py:361 - POST /cifar-deployment 307 13.5ms\n",
      "\u001b[2m\u001b[36m(HTTPProxyActor pid=9405)\u001b[0m INFO 2023-01-03 20:16:06,623 http_proxy 127.0.0.1 http_proxy.py:361 - POST /cifar-deployment 200 15.3ms\n",
      "\u001b[2m\u001b[36m(ServeReplica:cifar-deployment pid=9416)\u001b[0m INFO 2023-01-03 20:16:06,621 cifar-deployment cifar-deployment#rPepjh replica.py:505 - HANDLE __call__ OK 11.7ms\n",
      "\u001b[2m\u001b[36m(ServeReplica:cifar-deployment pid=9415)\u001b[0m INFO 2023-01-03 20:16:06,605 cifar-deployment cifar-deployment#BZpGjK replica.py:505 - HANDLE __call__ OK 0.2ms\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "prediction: 8; class: ship; matched: True\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\u001b[2m\u001b[36m(HTTPProxyActor pid=9405)\u001b[0m INFO 2023-01-03 20:16:06,774 http_proxy 127.0.0.1 http_proxy.py:361 - POST /cifar-deployment 307 3.7ms\n",
      "\u001b[2m\u001b[36m(HTTPProxyActor pid=9405)\u001b[0m INFO 2023-01-03 20:16:06,792 http_proxy 127.0.0.1 http_proxy.py:361 - POST /cifar-deployment 200 13.4ms\n",
      "\u001b[2m\u001b[36m(ServeReplica:cifar-deployment pid=9416)\u001b[0m INFO 2023-01-03 20:16:06,791 cifar-deployment cifar-deployment#rPepjh replica.py:505 - HANDLE __call__ OK 10.7ms\n",
      "\u001b[2m\u001b[36m(ServeReplica:cifar-deployment pid=9415)\u001b[0m INFO 2023-01-03 20:16:06,773 cifar-deployment cifar-deployment#BZpGjK replica.py:505 - HANDLE __call__ OK 0.1ms\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "prediction: 8; class: ship; matched: True\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "prediction: 2; class: bird; matched: False\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\u001b[2m\u001b[36m(HTTPProxyActor pid=9405)\u001b[0m INFO 2023-01-03 20:16:06,846 http_proxy 127.0.0.1 http_proxy.py:361 - POST /cifar-deployment 307 2.0ms\n",
      "\u001b[2m\u001b[36m(HTTPProxyActor pid=9405)\u001b[0m INFO 2023-01-03 20:16:06,859 http_proxy 127.0.0.1 http_proxy.py:361 - POST /cifar-deployment 200 12.2ms\n",
      "\u001b[2m\u001b[36m(HTTPProxyActor pid=9405)\u001b[0m INFO 2023-01-03 20:16:06,910 http_proxy 127.0.0.1 http_proxy.py:361 - POST /cifar-deployment 307 2.0ms\n",
      "\u001b[2m\u001b[36m(ServeReplica:cifar-deployment pid=9416)\u001b[0m INFO 2023-01-03 20:16:06,858 cifar-deployment cifar-deployment#rPepjh replica.py:505 - HANDLE __call__ OK 10.4ms\n",
      "\u001b[2m\u001b[36m(ServeReplica:cifar-deployment pid=9415)\u001b[0m INFO 2023-01-03 20:16:06,845 cifar-deployment cifar-deployment#BZpGjK replica.py:505 - HANDLE __call__ OK 0.1ms\n",
      "\u001b[2m\u001b[36m(ServeReplica:cifar-deployment pid=9415)\u001b[0m INFO 2023-01-03 20:16:06,909 cifar-deployment cifar-deployment#BZpGjK replica.py:505 - HANDLE __call__ OK 0.1ms\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\u001b[2m\u001b[36m(HTTPProxyActor pid=9405)\u001b[0m INFO 2023-01-03 20:16:06,924 http_proxy 127.0.0.1 http_proxy.py:361 - POST /cifar-deployment 200 12.8ms\n",
      "\u001b[2m\u001b[36m(HTTPProxyActor pid=9405)\u001b[0m INFO 2023-01-03 20:16:06,977 http_proxy 127.0.0.1 http_proxy.py:361 - POST /cifar-deployment 307 1.9ms\n",
      "\u001b[2m\u001b[36m(HTTPProxyActor pid=9405)\u001b[0m INFO 2023-01-03 20:16:06,991 http_proxy 127.0.0.1 http_proxy.py:361 - POST /cifar-deployment 200 12.8ms\n",
      "\u001b[2m\u001b[36m(ServeReplica:cifar-deployment pid=9416)\u001b[0m INFO 2023-01-03 20:16:06,923 cifar-deployment cifar-deployment#rPepjh replica.py:505 - HANDLE __call__ OK 10.6ms\n",
      "\u001b[2m\u001b[36m(ServeReplica:cifar-deployment pid=9416)\u001b[0m INFO 2023-01-03 20:16:06,990 cifar-deployment cifar-deployment#rPepjh replica.py:505 - HANDLE __call__ OK 10.8ms\n",
      "\u001b[2m\u001b[36m(ServeReplica:cifar-deployment pid=9415)\u001b[0m INFO 2023-01-03 20:16:06,977 cifar-deployment cifar-deployment#BZpGjK replica.py:505 - HANDLE __call__ OK 0.1ms\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "prediction: 6; class: frog; matched: True\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAaAAAAGdCAYAAABU0qcqAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjYuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/av/WaAAAACXBIWXMAAA9hAAAPYQGoP6dpAAAwd0lEQVR4nO3de3Cc9Xn3/8/uak86rSzLOuED5hwCdiYuOPqRUAIutjvDA8HTgSQzNSkDA7WZgpsmcSeBJG3HKZlJSDKO+aMUNzMBEjoBHpgGCiYWk9SmxcF1gOBix2AbW/JRZ2mP9/MHP9Qo2HBdtuSvJd4vZmeQ9vKl7733fe+l1e5+NhZFUSQAAE6xeOgFAAA+nBhAAIAgGEAAgCAYQACAIBhAAIAgGEAAgCAYQACAIBhAAIAgqkIv4A9VKhXt27dPdXV1isVioZcDAHCKokj9/f1qb29XPH78xzmn3QDat2+fZs2aFXoZAICTtGfPHs2cOfO410/YAFq7dq2+/e1vq6urS/Pnz9cPfvADXXrppR/47+rq6iRJD962VNXppOln/c/BA+Z1/eo3u8y1klSTrTbXfvyc49/Qx5KL7H8BjYaGXL2LUdlcm6zJuHrH4wlXfX9/v7k2lU65eitmX0vf8IirdV8+b64tV/nWnamd7qo/Olgw13Y7zgdJ0kjJXFqX9B0rHiVVXPVDJfv+SWeyvrWUfGuplOznW03Kdxs2NdqPlb0Hul29BwtFc63n71GlclnPvf670fvz45mQAfSTn/xEq1at0v3336+FCxfqvvvu0+LFi7V9+3Y1Nze/7799989u1emkeQBlUvbNSCZ8T3slq+z1nnVIUtYzgIq+3lWOhL+kc93eAVRM2uvTSech6VhLoeRbd75iry9X+Xp7tzOVtN8hJp1rUcJ+sKQSzt6Oe6246y5OKkb2tXhvk5hzLRXH+ZaawGMl6dw/yYT9uDqRJ0Q+6GmUCXkRwne+8x3dcsst+sIXvqALL7xQ999/v6qrq/XP//zPE/HjAACT0LgPoEKhoC1btmjRokX/+0PicS1atEibNm16T30+n1dfX9+YCwBg6hv3AXTo0CGVy2W1tLSM+X5LS4u6urreU79mzRrlcrnRCy9AAIAPh+DvA1q9erV6e3tHL3v27Am9JADAKTDuL0JoampSIpFQd/fYV2N0d3ertbX1PfXpdFrpdHq8lwEAOM2N+yOgVCqlBQsWaMOGDaPfq1Qq2rBhgzo6Osb7xwEAJqkJeRn2qlWrtHz5cv3RH/2RLr30Ut13330aHBzUF77whYn4cQCASWhCBtANN9yggwcP6u6771ZXV5c+9rGP6emnn37PCxMAAB9eE5aEsHLlSq1cufKE/31VTZ2SGdu7ywu73zL3XfCRM13raGx4/3fy/r4653v0NGB/93SUrXG1bqixv/O7Uh529S6X7euWpGzafpjFYr53oJdG7O+Gr0/a3tg8yrGdg3l7UoEkJRIDrvrYiD3FIeX8w/qI7O+itL9v/h2uWzxyvJtTUtLxDMLA0V5X70rZdxzmPuAd/7+v2ni/9q5YZF9LTcb3fHqV45yIHOsoGpMhgr8KDgDw4cQAAgAEwQACAATBAAIABMEAAgAEwQACAATBAAIABMEAAgAEwQACAATBAAIABDFhUTwnqxxPqBS3ZdtMb5hu7tvaNsO1jkJ+0F7b1+/qPeDonUjVunqX4/ZPcK8USq7eGffHZ9gjbcol31ocm6li3hc5VC1786oq3+9yqYQvdqZYZY9BOZgfcvUeHLHvn0TMF2eUTNvrs0l7fJQk1VXZs6/q8r4QoYxj3ZIUjzkORGfkUH7EcT/hWIYkxSv28y0esx/j1i3kERAAIAgGEAAgCAYQACAIBhAAIAgGEAAgCAYQACAIBhAAIAgGEAAgCAYQACAIBhAAIAgGEAAgiNM2C65YLqlYts3H5pZWc99M2jdzk4mMubYyNOLqrZg93yub9WVTRVHBXFvlDJDKZnxZcOWSPYcrVeU7JFNZe37YQL8vq69ctmd2JVO+HLP+vh5XfV3cvpZYOe/q3TdoP25jzruMpOzHeMyZkVYVpcy1DTXVrt41aft5L0llR6ZaKbJn70lST2+fvXfR17uhrs5cG4/b7zvzxkxHHgEBAIJgAAEAgmAAAQCCYAABAIJgAAEAgmAAAQCCYAABAIJgAAEAgmAAAQCCYAABAII4baN4FJWkyDof7TE1R3t9cSzJVMJcW/ClYCibtcd91Fbb1yFJkeyxJomyL0YmcsSOSFJtjb1/zJcKpFLRHiOTyvoO95Ehe5yRIt9t0pyrddUni/Z4nTkz21y9D+UPmmsLzqgXefanM4qnv8ceUVMZ9sUTpXP2CCFJSlTZf5d3JNq8s5a0/bh13oRyLFsJR23FWMsjIABAEAwgAEAQDCAAQBAMIABAEAwgAEAQDCAAQBAMIABAEAwgAEAQDCAAQBAMIABAEAwgAEAQp20WXCqbVSqTMtXmC/bAqe5uXxZce0ujuTbtyHaTpHLZkavlzXhK2G+TmDeALW7P3pMkRfbtjJV9mV2e8LhUypd5Nzxsz4LrGxly9Z7WPN1VP72SNtdG9fZaSSrFbOeZJB06YM9fk6RZTfbtTCV9x9XhA0fNtUnHNkpSqVh01Vccv8tHnlA1Sdm0/X4lk/Kdy5WK/Y4lVWW/DWPGHE8eAQEAghj3AfT1r39dsVhszOWCCy4Y7x8DAJjkJuRPcB/96Ef13HPP/e8PqTpt/9IHAAhkQiZDVVWVWltbJ6I1AGCKmJDngN544w21t7frrLPO0uc//3nt3r37uLX5fF59fX1jLgCAqW/cB9DChQu1fv16Pf3001q3bp127dqlT33qU+rvP/arz9asWaNcLjd6mTVr1ngvCQBwGhr3AbR06VL92Z/9mebNm6fFixfr3/7t39TT06Of/vSnx6xfvXq1ent7Ry979uwZ7yUBAE5DE/7qgIaGBp133nnasWPHMa9Pp9NKp33vWwAATH4T/j6ggYEB7dy5U21tbRP9owAAk8i4D6AvfvGL6uzs1Jtvvqn/+I//0Gc+8xklEgl99rOfHe8fBQCYxMb9T3B79+7VZz/7WR0+fFgzZszQJz/5SW3evFkzZsxw9amub1RN1vanuf273jL3LVR8MzeTqTbXlov26BZJimrsvVXxRWyUHGvJVjtjSuK++lSsYq6tDAz4eqcct2FVwtXbkSCkwlCvq3dvYcRVn47ZY2oaM75jfMEc+7l5tM4XZxQV7VEvUZUvb2oobd9BhYJjZ0pS3He+DQ4M2lsnfMdh1nE/4e2dcJwTnnivkvH+atwH0COPPDLeLQEAUxBZcACAIBhAAIAgGEAAgCAYQACAIBhAAIAgGEAAgCAYQACAIBhAAIAgGEAAgCAYQACAICb84xhOVL5YVlWVLb/prd32LLg5c+b61jFsz+yKV+yZZ5IUj9mzlaLI1ztbbc/sqko7M9IKvsyutGM7YwlfzlxRjnyqki8PrCZl/5iQfMWRSSepEvOdelHCvpak8/fKRKlkr3XmBu56e7+5NlXry5lzxONpZHjY1TtRcTSX1D80ZK71fvxMKmOvr0S+czOZtG9nuezIdDTW8ggIABAEAwgAEAQDCAAQBAMIABAEAwgAEAQDCAAQBAMIABAEAwgAEAQDCAAQBAMIABDEaRvFs/ftLmXTttiP1uZ2c19f6Iw0OGCP2KhN+uZ5pWKPQEkm7JEzklRy9E44D4OE7L0lKd9vvw2TzjijSsq+9qGCL46lXCiaawuOmBJJKjgihCSpv2iPhMplfDEy1Y6Toi7ri8tpbGo019ZMz7l6D8UPmWuPDPa4epcd8USS1DB9mrnWG8UTOeJ1quLOWC1ndM944xEQACAIBhAAIAgGEAAgCAYQACAIBhAAIAgGEAAgCAYQACAIBhAAIAgGEAAgCAYQACAIBhAAIIjTNgsuiiUVxWyZVom4PbdpoLfXtY7mXL25NlXly/dSwp41loyVXa37BwbMtSVnHlRt0pc3VV1fY64tlnxr6S/bc8/yKW9WX8Fcm623Z55JUrngyxrrO2TPPSv2+jLvWurtx3ii7DvGk8mMvTZT7eqdqW8w1w5HR129s1W+PL2kMbdSkhT33YaViv2ciCV8x3gxbz/GE4nxHxc8AgIABMEAAgAEwQACAATBAAIABMEAAgAEwQACAATBAAIABMEAAgAEwQACAATBAAIABMEAAgAEcdpmwR0+0qtMypbHdGDv78x95194vmsdmZQ9y6pU8GVwVacdeVNlX3ZYQ67OXhxLu3qn4nlXfT6yr7234mqtw7LnzCWqHbeJpGyN/fezxtYWV+9k/2FX/VBhxFzbf8jXOzlizxkcjuz5hZJUitvvYnr67NsoSUcH7Mfhwd5BV++ZDY5sN0kDg/b+5YrvIE+m7GuJ+aIUlUra74PiMXuGnbWWR0AAgCDcA+iFF17QNddco/b2dsViMT3++ONjro+iSHfffbfa2tqUzWa1aNEivfHGG+O1XgDAFOEeQIODg5o/f77Wrl17zOvvvfdeff/739f999+vF198UTU1NVq8eLFGRnwPrwEAU5v7OaClS5dq6dKlx7wuiiLdd999+upXv6prr71WkvSjH/1ILS0tevzxx3XjjTee3GoBAFPGuD4HtGvXLnV1dWnRokWj38vlclq4cKE2bdp0zH+Tz+fV19c35gIAmPrGdQB1dXVJklpaxr4iqKWlZfS6P7RmzRrlcrnRy6xZs8ZzSQCA01TwV8GtXr1avb29o5c9e/aEXhIA4BQY1wHU2toqSeru7h7z/e7u7tHr/lA6nVZ9ff2YCwBg6hvXATR37ly1trZqw4YNo9/r6+vTiy++qI6OjvH8UQCASc79KriBgQHt2LFj9Otdu3Zp69atamxs1OzZs3XnnXfq7//+73Xuuedq7ty5+trXvqb29nZdd91147luAMAk5x5AL730kj796U+Pfr1q1SpJ0vLly7V+/Xp96Utf0uDgoG699Vb19PTok5/8pJ5++mllMvZIG0n6ReevVBW3PUBrb7THseTqfHEshw4cMNcODfS7es+e1Wyura/23X6RI5KjUvEdBkf67LeJJJUciUNVTe2u3rPaP2auHer1RQjt27nLXFsa9EXU1FXbj1lJStdkzbV9/b5jpZK1nxMjke+PJuWi/XY5cqDX1fuV/7FHcI2U7DEyklSs+DJtYnFHf0ekjSSVyvaopFLJF9mViMY/XkeSKhXbmt0D6IorrlD0PvdusVhM3/zmN/XNb37T2xoA8CES/FVwAIAPJwYQACAIBhAAIAgGEAAgCAYQACAIBhAAIAgGEAAgCAYQACAIBhAAIAgGEAAgCHcUz6ny6p6jssYrnTFntrnvtJwvCy5RseeH1Zw919W7vr7WXNvfd9TVOz9iX3e5UnH1PjSScNVnM/btbGg49sd2HE9trf3jO4YOv+nqXZUYMde+/Outrt6HD/vy9M48o8lcmy/79k9Vwn43UF9j35eS1H/YftweHfblr1Vkz8erRL6svq7+QVd9Q8Z+G2a9v/ZHjrvplDOrr2w/9z23Yd6YSccjIABAEAwgAEAQDCAAQBAMIABAEAwgAEAQDCAAQBAMIABAEAwgAEAQDCAAQBAMIABAEKdtFE/z9GlKxG3zMZ2pNvftPtTrWkfSGAckSbUNOVfvfMEebREl0q7eyWzKXHu03xcLk/dEg0hqbWo316aq7PEqktT79m5zbeHIflfvhqw90uaCc8529f5vx76XpOltM821UeSLtMkX7LFNyVr7uSZJwwcPmWv7hu3RR5JUKNm3c6Rgi4YZZbzveVe1MXpGktJVjjsVSfF40lybd25nsWSP4klU2c+HfKlsquMREAAgCAYQACAIBhAAIAgGEAAgCAYQACAIBhAAIAgGEAAgCAYQACAIBhAAIAgGEAAgCAYQACCI0zYL7rKLzlWqyra8uuoac98tW193rePC8+aYa1sK9lwlSSoWbXlJkjQyXHD1TmftmWqZ2jpX79a6eld9Y2OTubZY9GWk9e2zZ8GVB305gLnpzebappZZrt5N7S2u+rqcfX/29fW5eqdS9tzAw90HXb1jCfvvuMm0L+9QcXumWo0zwy4es5+bklSVtG9nbV3G1Xt42L6WQsV3H1R2ZNglHRmDBbLgAACnMwYQACAIBhAAIAgGEAAgCAYQACAIBhAAIAgGEAAgCAYQACAIBhAAIAgGEAAgiNM2imfOtHplkklT7f4D9niQ4bwvYqMie0xJPJ5w9U4l7dEjQxp29T585Ki5traxwdW7ptYefSRJyZQ9eiRdZb+9JWna7Jnm2sPdvv2TdEQ8VWV9vatqal31xZI9iilX5+sdj9t/Dx3M+PZP2xlnmGt7h30xTJlqezxRpWCPnJGkwsiIqz47LWeuPeMM+zErSb19Q+ba3W93u3p7xGSPPioZa3kEBAAIggEEAAjCPYBeeOEFXXPNNWpvb1csFtPjjz8+5vqbbrpJsVhszGXJkiXjtV4AwBThHkCDg4OaP3++1q5de9yaJUuWaP/+/aOXhx9++KQWCQCYetwvQli6dKmWLl36vjXpdFqtra0nvCgAwNQ3Ic8Bbdy4Uc3NzTr//PN1++236/Dhw8etzefz6uvrG3MBAEx94z6AlixZoh/96EfasGGD/vEf/1GdnZ1aunSpyuVjv/x5zZo1yuVyo5dZs3yfLAkAmJzG/X1AN9544+j/X3zxxZo3b57OPvtsbdy4UVddddV76levXq1Vq1aNft3X18cQAoAPgQl/GfZZZ52lpqYm7dix45jXp9Np1dfXj7kAAKa+CR9Ae/fu1eHDh9XW1jbRPwoAMIm4/wQ3MDAw5tHMrl27tHXrVjU2NqqxsVHf+MY3tGzZMrW2tmrnzp360pe+pHPOOUeLFy8e14UDACY39wB66aWX9OlPf3r063efv1m+fLnWrVunbdu26V/+5V/U09Oj9vZ2XX311fq7v/s7pdP23DNJqolJ2Vhkqm2rqzP37Y58GU9DQ/b6kZG8q3e5XDHXloq+DLsjR3vNtYl6X3bY9GpffSZjz+zqd2TYSVIqYT+uEnH7OiSpMGzPD0s3+HLMImfWWFSw9y9HtvPmXUlj5qIkNU9rdPWuVOx/ZOkf7Hf1Hhqx5yN2He5x9c4m7blnklRdY/8LTyZjz0aUpPqGJnPt3oO+88dzP9FUZ88BtGbBuQfQFVdcoeh9DvBnnnnG2xIA8CFEFhwAIAgGEAAgCAYQACAIBhAAIAgGEAAgCAYQACAIBhAAIAgGEAAgCAYQACAIBhAAIIhx/zyg8ZKsREpWbJlW07L2PLBMtsG1jsZ6e30U+fKjkin7unMN9rwuSXqra7+5tndw0NX7fOdHZry27Tfm2kP7D7h6f/TcC8y18aRv3QNHD5lrD/zPq67esSpfNmJtdYO5dtC5P4/3YZHH0p/35R2+sc++P3e9udvVe/8R+6cnDzuzFOPVvv1TqdhzHeWL6lPacT9R32TPjZOk3Qfsx3hqcMhcWzQeUzwCAgAEwQACAATBAAIABMEAAgAEwQACAATBAAIABMEAAgAEwQACAATBAAIABMEAAgAEcdpG8WSTaVWnbPEzZUe2xdHeftc6YvHp5tp0nS/qpVC2z//SiC9eZSRfNNfu2bHX1fviCz/mqh842muubaqvc/VubGo01+793R5X71//93+ba3Mt9nVI0mFHBIoktcw4w1x7aMAemSJJuw/a19I7OOLq/fbb9iie4SFfzE+mOmsvjidcvRtqfOdyrGSP+qnP1bh6yxELNK1phqt1ofxbc21voWCuLZVt0UQ8AgIABMEAAgAEwQACAATBAAIABMEAAgAEwQACAATBAAIABMEAAgAEwQACAATBAAIABMEAAgAEcdpmwVXF46qK2+Zj79Cwue+RI4dd62gaaTLXFhRz9Vb1NHOp9bZ4V26aPZvs/z7Z6ep97pkXuOrPPvMcc215sM/Vu7fniLn26JGDrt4Ntfb9c/n/9yeu3nt2/I+r/vXX7fX7Dtmz9yTpjQNHzbUF+TLVSmVbnqMktU1rcPXO1mbMtft67ceJJFUn7b0lKenIo0z4bkI1tLeba3tLvrt0Y2SbJKlnxH4/W66QBQcAOI0xgAAAQTCAAABBMIAAAEEwgAAAQTCAAABBMIAAAEEwgAAAQTCAAABBMIAAAEGctlE8sURcsYRtPlZnq819Z8+e7VpHpiptri0Viq7e8VTZXFspl3y94/YIlL37fBE169Y/7Kr/P0uuMNc2NdS5emcPDJhre9/ucfVWv31/9r2539X6jPrprvqDNfbb5fXfve3qHRsYMtdOb25x9VZNjbk0a0+zkSQlY474m0LB1bu/t8dVX27OmmtTSft9iiTVZu29285odvWe3myPmzrQdcBcGzNGE/EICAAQhGsArVmzRpdcconq6urU3Nys6667Ttu3bx9TMzIyohUrVmj69Omqra3VsmXL1N3dPa6LBgBMfq4B1NnZqRUrVmjz5s169tlnVSwWdfXVV2twcHC05q677tKTTz6pRx99VJ2dndq3b5+uv/76cV84AGBycz0H9PTTT4/5ev369WpubtaWLVt0+eWXq7e3Vw888IAeeughXXnllZKkBx98UB/5yEe0efNmfeITnxi/lQMAJrWTeg6ot/edzx1pbHzns2e2bNmiYrGoRYsWjdZccMEFmj17tjZt2nTMHvl8Xn19fWMuAICp74QHUKVS0Z133qnLLrtMF110kSSpq6tLqVRKDQ0NY2pbWlrU1dV1zD5r1qxRLpcbvcyaNetElwQAmEROeACtWLFCr7zyih555JGTWsDq1avV29s7etmzZ89J9QMATA4n9D6glStX6qmnntILL7ygmTNnjn6/tbVVhUJBPT09Yx4FdXd3q7W19Zi90um00mnf6+IBAJOf6xFQFEVauXKlHnvsMT3//POaO3fumOsXLFigZDKpDRs2jH5v+/bt2r17tzo6OsZnxQCAKcH1CGjFihV66KGH9MQTT6iurm70eZ1cLqdsNqtcLqebb75Zq1atUmNjo+rr63XHHXeoo6ODV8ABAMZwDaB169ZJkq644oox33/wwQd10003SZK++93vKh6Pa9myZcrn81q8eLF++MMfjstiAQBTh2sARdEH5/tkMhmtXbtWa9euPeFFSVImm1EmlTLVxir2vsNHfS/zHuq1Z40Vh/Ou3mX1mmt7Dx77VYTHs3v3XnNtVdz3WpRDR+zrlqSfPPGMuTaX82XBtU5rNNfOSNjz8SQp3mPfzqGBwQ8u+j31zTlX/cFB+3FbSfue2s1H9sy7oaO+3MAokTDXZqOYq3f7NPttOMN5XEXOY6VYtOc69vcPu3rPyNtz7Kozvn0/bXq9ufbIfnuiTcwwKySy4AAAgTCAAABBMIAAAEEwgAAAQTCAAABBMIAAAEEwgAAAQTCAAABBMIAAAEEwgAAAQZzQxzGcColkUomUMQ5jpGTuWxwZca0jZk8S0cCRHlfvSr09AsX7SbGHD9pjMy46s83VO9c0w1W/9217jNDBo76YnzeHhsy1+ZpaV+8ZKfvHhAxlHAeKpNd3v+mq39F9yFwbS2dcvfscSy/kfedPZE+o0cG8L8qqWLafPzMb7ZFNkj+eqliyRc9I0u9+t9vVu6m53Vwbq/ft+2l11eZaTziRNVSJR0AAgCAYQACAIBhAAIAgGEAAgCAYQACAIBhAAIAgGEAAgCAYQACAIBhAAIAgGEAAgCAYQACAIE7bLLhSpaxS2RYk1dvTY+5bW+3LA0umUubafmcWXJW9tSLZs6Yk6cyZM821583x9d6/z55LJkmZ+npz7UeaWly9E2lr6pQUFe3ZYZLUUJ8z1x7oOerq/cpee1afJO3uGTDXRlGPq3ciaU/5SiYcB62kqri9d1/JERwnafDwEXPtwIgvZ64l49vO6jPseYqHDtnXLUm7Xt9urp174Vmu3mc0TjPXbq+yhwaWKrbzkkdAAIAgGEAAgCAYQACAIBhAAIAgGEAAgCAYQACAIBhAAIAgGEAAgCAYQACAIBhAAIAgTtsoniNHj2jYGBFy9Ig9BmVm+yzXOnIN9qiKt3oOuHr37N9vrp0z92xX7xlnzjHXHtr9W1fvt1/31c9psMfrJCr26BZJqk7b64sJe5SIJPX1D5prK3lfzM/0XJOrfihKm2uLBd9a8o76qGiPPpKkwbK9d6nKt+9jSfvvz92DQ67erXU1vrU4crUOdne5ekf518y1mWrfXXrLtOnm2vPOtd8HFUoldb75wdvJIyAAQBAMIABAEAwgAEAQDCAAQBAMIABAEAwgAEAQDCAAQBAMIABAEAwgAEAQDCAAQBAMIABAEKdtFlz8///Poq3ZnmeUjhdc6xjsO2zvHfPN894jPeba7thuV+/UrDZzbW1bu6v3nI/Pc9U3T5thrj3y9kFX76499vraZNbVO5e111dqfBlp8awv96w2bj+2eou+Y/zQkD3zbqhQcvXWiCOXruxbdzaeMdcmM/ZaSSqm7NlukrS/r99c232419W7ULEfWyMvv+7qPXuuPTNyzix7juaIMV+QR0AAgCBcA2jNmjW65JJLVFdXp+bmZl133XXavn37mJorrrhCsVhszOW2224b10UDACY/1wDq7OzUihUrtHnzZj377LMqFou6+uqrNTg49iH8Lbfcov37949e7r333nFdNABg8nM9B/T000+P+Xr9+vVqbm7Wli1bdPnll49+v7q6Wq2treOzQgDAlHRSzwH19r7zZFpjY+OY7//4xz9WU1OTLrroIq1evVpDQ8f/MKh8Pq++vr4xFwDA1HfCr4KrVCq68847ddlll+miiy4a/f7nPvc5zZkzR+3t7dq2bZu+/OUva/v27frZz352zD5r1qzRN77xjRNdBgBgkjrhAbRixQq98sor+uUvfznm+7feeuvo/1988cVqa2vTVVddpZ07d+rss9/7ka6rV6/WqlWrRr/u6+vTLMfL/QAAk9MJDaCVK1fqqaee0gsvvKCZM2e+b+3ChQslSTt27DjmAEqn00qn7Z93DwCYGlwDKIoi3XHHHXrssce0ceNGzZ079wP/zdatWyVJbW32N0YCAKY+1wBasWKFHnroIT3xxBOqq6tTV1eXJCmXyymbzWrnzp166KGH9Kd/+qeaPn26tm3bprvuukuXX3655s3zvXseADC1uQbQunXrJL3zZtPf9+CDD+qmm25SKpXSc889p/vuu0+Dg4OaNWuWli1bpq9+9avjtmAAwNTg/hPc+5k1a5Y6OztPakH/Ky7rq8QjR05WPvJldin2/tv8+6Y3NLhaV9fXmGv3Hjrg6r3pP94y1y74xCWu3qWEL1dryyuvmmtrY76nJUsJ+76f1mzPpJOk6ip770Sv/TiRpMh5HMYj+1rqnVlwDXXV5tqK41yTpKGhYXPtH76h/YPU1tjPn0Qi4epdLNjXLUn5wby5trWpwdX7DEdWY0u7L9fxtVdfMde2TZ9mri2UbJmBZMEBAIJgAAEAgmAAAQCCYAABAIJgAAEAgmAAAQCCYAABAIJgAAEAgmAAAQCCYAABAII44c8DmmilUlmlmDGKJ2mPhuk+2u9aR9oxoufm7FEVkhSv2ONb6tJZV++jtiQMSdKbr7/p6j2tpdlVv3ewbK4tOZOSMlVJc208ctwokuJle3zLtCrf/jlSHnDV19fYP7KkMZlz9S5X7Df6yMjxP934mPVp+/6JNda7etfn7PXliv0YlKTBYV8sUOQ4l5POWKC6mpS5tsYRHyVJNSl778qwPZ6oUrLd3jwCAgAEwQACAATBAAIABMEAAgAEwQACAATBAAIABMEAAgAEwQACAATBAAIABMEAAgAEwQACAARx2mbBpTNZZVK2HKlCzJ6tdLTfl2XVkLVnWeVHRly9+3p7zLUDA74Mu2mZWnNtrGjPsZKkna++7qrPpe1rmdPc6uo9NNhjro0qBVfvSmTf96m471SaVl3jqi8k7f2TMfu6JWmw155L50u8k6rq7DmNyaRv3dXV1ebaYsmXA1jI2rP3JKlcqZhrK5Evl85z7v/utQOu3i2NTebaM1tbzLXDhaL0X69+YB2PgAAAQTCAAABBMIAAAEEwgAAAQTCAAABBMIAAAEEwgAAAQTCAAABBMIAAAEEwgAAAQZy2UTyHDnQpY4wfSdfkzH1n1PvCRFqbpptrCyN5V++kI0JoWnWdq7cS9t8t0vW+3o7W7/R3xNRkYr5YIM+vUFHMHpciSSMqmmurnL/LZZ1RL7GyfS0jA72u3sXBYXNtfb09VkmSMln77RKL+27DTJX9/IkZY73eNZy3396SVInZa4sVXyzQsOxRSdMb7PeFktQ0rdFcW5tKmWsTst0gPAICAATBAAIABMEAAgAEwQACAATBAAIABMEAAgAEwQACAATBAAIABMEAAgAEwQACAATBAAIABHHaZsFlq6uVNWbB1dfaM4rqqn1ZcKl0xlx75GjB17vKfvMnkvZtlKRKZM89i8q+DLumhhpXfbbKvvZksezq7fkVaqDs2z+HRkbMtaUR37rrMr4suErJvo8SVc5cupz9nIgSjtAzSfGE/RiPxX29o5i9PuM4jyWp7ItrU9mxllLZl3dYXWPP36tEvrv0pOxrKQwN2WuLtiw9HgEBAIJwDaB169Zp3rx5qq+vV319vTo6OvTzn/989PqRkRGtWLFC06dPV21trZYtW6bu7u5xXzQAYPJzDaCZM2fqW9/6lrZs2aKXXnpJV155pa699lq9+uqrkqS77rpLTz75pB599FF1dnZq3759uv766ydk4QCAyc31B8NrrrlmzNf/8A//oHXr1mnz5s2aOXOmHnjgAT300EO68sorJUkPPvigPvKRj2jz5s36xCc+MX6rBgBMeif8HFC5XNYjjzyiwcFBdXR0aMuWLSoWi1q0aNFozQUXXKDZs2dr06ZNx+2Tz+fV19c35gIAmPrcA+g3v/mNamtrlU6nddttt+mxxx7ThRdeqK6uLqVSKTU0NIypb2lpUVdX13H7rVmzRrlcbvQya9Ys90YAACYf9wA6//zztXXrVr344ou6/fbbtXz5cr322msnvIDVq1ert7d39LJnz54T7gUAmDzc7wNKpVI655xzJEkLFizQf/3Xf+l73/uebrjhBhUKBfX09Ix5FNTd3a3W1tbj9kun00qnfe+JAABMfif9PqBKpaJ8Pq8FCxYomUxqw4YNo9dt375du3fvVkdHx8n+GADAFON6BLR69WotXbpUs2fPVn9/vx566CFt3LhRzzzzjHK5nG6++WatWrVKjY2Nqq+v1x133KGOjg5eAQcAeA/XADpw4ID+/M//XPv371cul9O8efP0zDPP6E/+5E8kSd/97ncVj8e1bNky5fN5LV68WD/84Q9PaGHpTEbpVNJUW+uIqqhK+R709Q0Pmmv39vX4evfYX/HXVFPv6l2fs8flJPK+26S777Crvrra/ifWtC+lRPFKwlxbTPjijApFe/RIT1+/q3dU8mW9VDv+TJ3JVrt6F0v22KZYzLeDUmn7bR5Fvt5VjiirmCMqR5ISCd85MVK0xzzVOp9yqM3YY4QKFV8kVCJm386oZN/GqGyL4nENoAceeOB9r89kMlq7dq3Wrl3raQsA+BAiCw4AEAQDCAAQBAMIABAEAwgAEAQDCAAQBAMIABAEAwgAEAQDCAAQBAMIABCEOw17or0bxzFctEU5SNJQwR4RURX5Zu5Qwb6OkaIvXiXviGPx9k461u2PHfGtJVa0x+VU3FE89hiZQmXitjNf8kWgjDijeOJx+21YrvL1Lk1gFE+V49iK5OtdcpSXIl8UT95x/kjScMle70wckmRfeyHyRvHY933k2MZ3z50PileKRd4Apgm2d+9ePpQOAKaAPXv2aObMmce9/rQbQJVKRfv27VNdXd2YAMG+vj7NmjVLe/bsUX29L5hzMmE7p44PwzZKbOdUMx7bGUWR+vv71d7ernj8+I+CT7s/wcXj8fedmPX19VN657+L7Zw6PgzbKLGdU83Jbmcul/vAGl6EAAAIggEEAAhi0gygdDqte+65R2nnhzlNNmzn1PFh2EaJ7ZxqTuV2nnYvQgAAfDhMmkdAAICphQEEAAiCAQQACIIBBAAIYtIMoLVr1+rMM89UJpPRwoUL9Z//+Z+hlzSuvv71rysWi425XHDBBaGXdVJeeOEFXXPNNWpvb1csFtPjjz8+5vooinT33Xerra1N2WxWixYt0htvvBFmsSfhg7bzpptues++XbJkSZjFnqA1a9bokksuUV1dnZqbm3Xddddp+/btY2pGRka0YsUKTZ8+XbW1tVq2bJm6u7sDrfjEWLbziiuueM/+vO222wKt+MSsW7dO8+bNG32zaUdHh37+85+PXn+q9uWkGEA/+clPtGrVKt1zzz369a9/rfnz52vx4sU6cOBA6KWNq49+9KPav3//6OWXv/xl6CWdlMHBQc2fP19r16495vX33nuvvv/97+v+++/Xiy++qJqaGi1evFgjIyOneKUn54O2U5KWLFkyZt8+/PDDp3CFJ6+zs1MrVqzQ5s2b9eyzz6pYLOrqq6/W4ODgaM1dd92lJ598Uo8++qg6Ozu1b98+XX/99QFX7WfZTkm65ZZbxuzPe++9N9CKT8zMmTP1rW99S1u2bNFLL72kK6+8Utdee61effVVSadwX0aTwKWXXhqtWLFi9OtyuRy1t7dHa9asCbiq8XXPPfdE8+fPD72MCSMpeuyxx0a/rlQqUWtra/Ttb3979Hs9PT1ROp2OHn744QArHB9/uJ1RFEXLly+Prr322iDrmSgHDhyIJEWdnZ1RFL2z75LJZPToo4+O1vz2t7+NJEWbNm0KtcyT9ofbGUVR9Md//MfRX/3VX4Vb1ASZNm1a9E//9E+ndF+e9o+ACoWCtmzZokWLFo1+Lx6Pa9GiRdq0aVPAlY2/N954Q+3t7TrrrLP0+c9/Xrt37w69pAmza9cudXV1jdmvuVxOCxcunHL7VZI2btyo5uZmnX/++br99tt1+PDh0Es6Kb29vZKkxsZGSdKWLVtULBbH7M8LLrhAs2fPntT78w+3810//vGP1dTUpIsuukirV6/W0NBQiOWNi3K5rEceeUSDg4Pq6Og4pfvytAsj/UOHDh1SuVxWS0vLmO+3tLTo9ddfD7Sq8bdw4UKtX79e559/vvbv369vfOMb+tSnPqVXXnlFdXV1oZc37rq6uiTpmPv13eumiiVLluj666/X3LlztXPnTv3t3/6tli5dqk2bNimRsH/Oz+miUqnozjvv1GWXXaaLLrpI0jv7M5VKqaGhYUztZN6fx9pOSfrc5z6nOXPmqL29Xdu2bdOXv/xlbd++XT/72c8CrtbvN7/5jTo6OjQyMqLa2lo99thjuvDCC7V169ZTti9P+wH0YbF06dLR/583b54WLlyoOXPm6Kc//aluvvnmgCvDybrxxhtH///iiy/WvHnzdPbZZ2vjxo266qqrAq7sxKxYsUKvvPLKpH+O8oMcbztvvfXW0f+/+OKL1dbWpquuuko7d+7U2WeffaqXecLOP/98bd26Vb29vfrXf/1XLV++XJ2dnad0Daf9n+CampqUSCTe8wqM7u5utba2BlrVxGtoaNB5552nHTt2hF7KhHh3333Y9qsknXXWWWpqapqU+3blypV66qmn9Itf/GLMx6a0traqUCiop6dnTP1k3Z/H285jWbhwoSRNuv2ZSqV0zjnnaMGCBVqzZo3mz5+v733ve6d0X572AyiVSmnBggXasGHD6PcqlYo2bNigjo6OgCubWAMDA9q5c6fa2tpCL2VCzJ07V62trWP2a19fn1588cUpvV+ldz719/Dhw5Nq30ZRpJUrV+qxxx7T888/r7lz5465fsGCBUomk2P25/bt27V79+5JtT8/aDuPZevWrZI0qfbnsVQqFeXz+VO7L8f1JQ0T5JFHHonS6XS0fv366LXXXotuvfXWqKGhIerq6gq9tHHz13/919HGjRujXbt2Rb/61a+iRYsWRU1NTdGBAwdCL+2E9ff3Ry+//HL08ssvR5Ki73znO9HLL78cvfXWW1EURdG3vvWtqKGhIXriiSeibdu2Rddee200d+7caHh4OPDKfd5vO/v7+6MvfvGL0aZNm6Jdu3ZFzz33XPTxj388Ovfcc6ORkZHQSze7/fbbo1wuF23cuDHav3//6GVoaGi05rbbbotmz54dPf/889FLL70UdXR0RB0dHQFX7fdB27ljx47om9/8ZvTSSy9Fu3btip544onorLPOii6//PLAK/f5yle+EnV2dka7du2Ktm3bFn3lK1+JYrFY9O///u9RFJ26fTkpBlAURdEPfvCDaPbs2VEqlYouvfTSaPPmzaGXNK5uuOGGqK2tLUqlUtEZZ5wR3XDDDdGOHTtCL+uk/OIXv4gkveeyfPnyKIreeSn21772tailpSVKp9PRVVddFW3fvj3sok/A+23n0NBQdPXVV0czZsyIkslkNGfOnOiWW26ZdL88HWv7JEUPPvjgaM3w8HD0l3/5l9G0adOi6urq6DOf+Uy0f//+cIs+AR+0nbt3744uv/zyqLGxMUqn09E555wT/c3f/E3U29sbduFOf/EXfxHNmTMnSqVS0YwZM6KrrrpqdPhE0anbl3wcAwAgiNP+OSAAwNTEAAIABMEAAgAEwQACAATBAAIABMEAAgAEwQACAATBAAIABMEAAgAEwQACAATBAAIABMEAAgAE8f8ANErDuB4fQxgAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "prediction: 6; class: frog; matched: True\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "prediction: 1; class: car; matched: True\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\u001b[2m\u001b[36m(HTTPProxyActor pid=9405)\u001b[0m INFO 2023-01-03 20:16:07,046 http_proxy 127.0.0.1 http_proxy.py:361 - POST /cifar-deployment 307 2.2ms\n",
      "\u001b[2m\u001b[36m(HTTPProxyActor pid=9405)\u001b[0m INFO 2023-01-03 20:16:07,060 http_proxy 127.0.0.1 http_proxy.py:361 - POST /cifar-deployment 200 12.2ms\n",
      "\u001b[2m\u001b[36m(HTTPProxyActor pid=9405)\u001b[0m INFO 2023-01-03 20:16:07,109 http_proxy 127.0.0.1 http_proxy.py:361 - POST /cifar-deployment 307 1.7ms\n",
      "\u001b[2m\u001b[36m(HTTPProxyActor pid=9405)\u001b[0m INFO 2023-01-03 20:16:07,122 http_proxy 127.0.0.1 http_proxy.py:361 - POST /cifar-deployment 200 12.1ms\n",
      "\u001b[2m\u001b[36m(ServeReplica:cifar-deployment pid=9416)\u001b[0m INFO 2023-01-03 20:16:07,059 cifar-deployment cifar-deployment#rPepjh replica.py:505 - HANDLE __call__ OK 10.2ms\n",
      "\u001b[2m\u001b[36m(ServeReplica:cifar-deployment pid=9416)\u001b[0m INFO 2023-01-03 20:16:07,121 cifar-deployment cifar-deployment#rPepjh replica.py:505 - HANDLE __call__ OK 10.3ms\n",
      "\u001b[2m\u001b[36m(ServeReplica:cifar-deployment pid=9415)\u001b[0m INFO 2023-01-03 20:16:07,045 cifar-deployment cifar-deployment#BZpGjK replica.py:505 - HANDLE __call__ OK 0.1ms\n",
      "\u001b[2m\u001b[36m(ServeReplica:cifar-deployment pid=9415)\u001b[0m INFO 2023-01-03 20:16:07,108 cifar-deployment cifar-deployment#BZpGjK replica.py:505 - HANDLE __call__ OK 0.1ms\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "prediction: 6; class: frog; matched: True\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\u001b[2m\u001b[36m(HTTPProxyActor pid=9405)\u001b[0m INFO 2023-01-03 20:16:07,177 http_proxy 127.0.0.1 http_proxy.py:361 - POST /cifar-deployment 307 1.9ms\n",
      "\u001b[2m\u001b[36m(HTTPProxyActor pid=9405)\u001b[0m INFO 2023-01-03 20:16:07,190 http_proxy 127.0.0.1 http_proxy.py:361 - POST /cifar-deployment 200 12.2ms\n",
      "\u001b[2m\u001b[36m(ServeReplica:cifar-deployment pid=9416)\u001b[0m INFO 2023-01-03 20:16:07,189 cifar-deployment cifar-deployment#rPepjh replica.py:505 - HANDLE __call__ OK 10.2ms\n",
      "\u001b[2m\u001b[36m(ServeReplica:cifar-deployment pid=9415)\u001b[0m INFO 2023-01-03 20:16:07,176 cifar-deployment cifar-deployment#BZpGjK replica.py:505 - HANDLE __call__ OK 0.1ms\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "prediction: 3; class: cat; matched: True\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "prediction: 1; class: car; matched: True\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\u001b[2m\u001b[36m(HTTPProxyActor pid=9405)\u001b[0m INFO 2023-01-03 20:16:07,242 http_proxy 127.0.0.1 http_proxy.py:361 - POST /cifar-deployment 307 1.8ms\n",
      "\u001b[2m\u001b[36m(HTTPProxyActor pid=9405)\u001b[0m INFO 2023-01-03 20:16:07,255 http_proxy 127.0.0.1 http_proxy.py:361 - POST /cifar-deployment 200 11.9ms\n",
      "\u001b[2m\u001b[36m(ServeReplica:cifar-deployment pid=9416)\u001b[0m INFO 2023-01-03 20:16:07,254 cifar-deployment cifar-deployment#rPepjh replica.py:505 - HANDLE __call__ OK 9.8ms\n",
      "\u001b[2m\u001b[36m(ServeReplica:cifar-deployment pid=9415)\u001b[0m INFO 2023-01-03 20:16:07,241 cifar-deployment cifar-deployment#BZpGjK replica.py:505 - HANDLE __call__ OK 0.1ms\n"
     ]
    }
   ],
   "source": [
    "# Test online deployment\n",
    "batch = test_dataset.take(10)\n",
    "for i in range(10):\n",
    "    array = np.expand_dims(np.array(batch[i][\"image\"]), axis=0)\n",
    "    label = np.array(batch[i][\"label\"])\n",
    "    \n",
    "    # send request and fetch prediction\n",
    "    payload  = {\"array\": array.tolist()}\n",
    "    response = requests.post(deployment.url, json=payload).json()\n",
    "    result = response[\"predictions\"][0]\n",
    "    idx, cls = cifar_utils.to_prediction_cls(result)\n",
    "    matched = idx == label\n",
    "    cifar_utils.img_show(batch[i][\"image\"])\n",
    "    print(f\"prediction: {idx}; class: {cls}; matched: {matched}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\u001b[2m\u001b[36m(ServeController pid=9403)\u001b[0m INFO 2023-01-03 20:17:02,217 controller 9403 deployment_state.py:1336 - Removing 2 replicas from deployment 'cifar-deployment'.\n"
     ]
    }
   ],
   "source": [
    "# Shutdown our server\n",
    "serve.shutdown()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Summary\n",
    "\n",
    "In this end-to-end ML application, you walked through a step-by-step guide on how to use Ray AIR library and its various components at each stage of an ML workflow and its respective workload: from data preprocessing and ingestion, training and tuning, to batch inference, and finaly to deploying for online inference. \n",
    "\n",
    "For this image classificaiton use case, we used PyTorch and its relevant AIR integrations such as `TorchTrainer`, `TorchCheckpoint`, `TorchPredictor`, etc. AIR supports other common frameworks such as TensorFlow, XGBoost, Scikit-learn, HuggingFace etc.\n",
    "\n",
    "While we used a fixed set of hyperparameters, try training and tuning with some different parameter space in the exercise below, keeping the caveat in mind.\n",
    "\n",
    "The takeaway here is that Ray AIR makes data distributed parallel for distributed training and HPO seamlessly easier with a unifed set of high-level and domain specific APIs while abstracting and delegating low-level functionality to Ray Core and the respective integtrated native ML library.\n",
    "\n",
    "**Caveat**: On single machine, an M1 Mac laptop with 10 cores, with the above HPO parameters configuration took over 3+ hours to train and tune with 15 epochs. However,on an Anyscale Ray cluster with 4 nodes, 80 CPUs, and 4 GPUs, it took only 10 minutes!\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Exercises\n",
    "\n",
    " * Try scaling your Trainer and Tunner\n",
    "  * increase the `num_workers` 7 if you have 10 cores more\n",
    "  * extend the tuning space \n",
    "    *  `lr=[0.001, 0.01]`\n",
    "    * `epochs=[20, 25]`\n",
    "    * `batch_size=[64, 128]`\n",
    "    \n",
    "With lower learning rate, larger batch size, and an extra workder, does the loss converge close to 0.01 and does the overall batch accuracy improve?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Homework \n",
    " * Read the resources below\n",
    "\n",
    "\n",
    "### Resources\n",
    "\n",
    "Here some extra reading and resources for you:\n",
    " * Watch the Ray Summit 2022 on [Introduction to Ray AIR](https://www.anyscale.com/ray-summit-2022/agenda/sessions/226)\n",
    " * Ray AIR [documentation](https://docs.ray.io/en/latest/ray-air/getting-started.html)\n",
    " * Understand its [Components and APIs](https://docs.ray.io/en/latest/ray-air/package-ref.html)\n",
    " * Ray AIR [user guides](https://docs.ray.io/en/latest/ray-air/user-guides.html) and [examples](https://docs.ray.io/en/latest/ray-air/examples/index.html) \n",
    " * Read the [Ray AIR whitepaper](https://docs.google.com/document/d/1bYL-638GN6EeJ45dPuLiPImA8msojEDDKiBx3YzB4_s/edit#heading=h.ru1taexewu7i)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.13"
  },
  "vscode": {
   "interpreter": {
    "hash": "2706dab362c94031065b1e6b4ba35ae246b1d7f08f9a1709d5e0cd568b23fcbf"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}