DSAI/06_NN/code/nn_6_deployment.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "566e2678",
   "metadata": {},
   "source": [
    "<div style=\"\n",
    "    border: 2px solid #4CAF50; \n",
    "    padding: 15px; \n",
    "    background-color: #f4f4f4; \n",
    "    border-radius: 10px; \n",
    "    align-items: center;\">\n",
    "\n",
    "<h1 style=\"margin: 0; color: #4CAF50;\">Neural Networks: Deployment, Import und Export von Modellen</h1>\n",
    "<h2 style=\"margin: 5px 0; color: #555;\">DSAI</h2>\n",
    "<h3 style=\"margin: 5px 0; color: #555;\">Jakob Eggl</h3>\n",
    "\n",
    "<div style=\"flex-shrink: 0;\">\n",
    "    <img src=\"https://www.htl-grieskirchen.at/wp/wp-content/uploads/2022/11/logo_bildschirm-1024x503.png\" alt=\"Logo\" style=\"width: 250px; height: auto;\"/>\n",
    "</div>\n",
    "<p1> © 2025/26 Jakob Eggl. Nutzung oder Verbreitung nur mit ausdrücklicher Genehmigung des Autors.</p1>\n",
    "</div>\n",
    "<div style=\"flex: 1;\">\n",
    "</div>   "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "405e8e29",
   "metadata": {},
   "source": [
    "Wir wollen uns in diesem Notebook mit 3 wichtigen Konzepten beschäftigen:\n",
    "\n",
    "1) Model Import/Export\n",
    "2) Verwenden und/oder Finetuning von vortrainierten (pretrained) Modellen\n",
    "3) Model Deployment mit Flask"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "13624ed2",
   "metadata": {},
   "source": [
    "Die Motivation für diese Themen ist einfach. In vielen Fällen wäre es praktisch, das fertig trainierte Modell anderen zur Verfügung zu stellen. Diese können dieses dann entweder nur verwenden (online darauf zugreifen), oder, falls das ganze Modell geteilt wird, mit diesem auch weitere Dinge, wie zum Beispiel Finetuning etc., erledigen."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9e220bc5",
   "metadata": {
    "vscode": {
     "languageId": "plaintext"
    }
   },
   "source": [
    "# Model Import/Export"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7af8efb9",
   "metadata": {},
   "source": [
    "In diesem Teil wollen wir unser Modell als Datei exportieren, bzw. im Anschluss wieder importieren."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8800fc68",
   "metadata": {},
   "source": [
    "### PyTorch Export"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e7e28854",
   "metadata": {},
   "source": [
    "Wir starten mit einem normalen Modell, welches wir später exportieren wollen."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "id": "18c53b1b",
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import torch\n",
    "from torch import nn\n",
    "from torch.utils.data import DataLoader, random_split, TensorDataset\n",
    "from sklearn.preprocessing import OrdinalEncoder, StandardScaler\n",
    "import torch.optim as optim\n",
    "import onnx\n",
    "import onnxruntime as ort"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "id": "a63be8e9",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cpu\n"
     ]
    }
   ],
   "source": [
    "device = torch.device('cuda') if torch.cuda.is_available() else torch.device('mps') if torch.backends.mps.is_available() else torch.device('cpu')\n",
    "print(device)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "id": "1aa73a7b",
   "metadata": {},
   "outputs": [],
   "source": [
    "class SimpleRegressor(nn.Module):\n",
    "    def __init__(self):\n",
    "        super(SimpleRegressor, self).__init__()\n",
    "        self.net = nn.Sequential(\n",
    "            nn.Linear(12, 24),\n",
    "            nn.ReLU(),\n",
    "            nn.Linear(24, 12),\n",
    "            nn.ReLU(),\n",
    "            nn.Linear(12, 6),\n",
    "            nn.ReLU(),\n",
    "            nn.Linear(6, 1)\n",
    "            )\n",
    "    def forward(self, x):\n",
    "        return self.net(x)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "id": "4d8d6d67",
   "metadata": {},
   "outputs": [],
   "source": [
    "model = SimpleRegressor().to(device)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "id": "6a112ff7",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Simulation of a train method\n",
    "\n",
    "def train_model(model, epochs):\n",
    "    print(\"Started Dummy Training the model...\")\n",
    "    model.train()\n",
    "    optimizer = optim.Adam(model.parameters(), lr=0.001)\n",
    "    criterion = nn.MSELoss()\n",
    "    for epoch in range(epochs):\n",
    "        # Dummy data\n",
    "        inputs = torch.randn(32, 12).to(device)\n",
    "        targets = torch.randn(32, 1).to(device)\n",
    "        \n",
    "        optimizer.zero_grad()\n",
    "        outputs = model(inputs)\n",
    "        loss = criterion(outputs, targets)\n",
    "        loss.backward()\n",
    "        optimizer.step()\n",
    "        \n",
    "        if (epoch+1) % 1 == 0:\n",
    "            print(f'Epoch [{epoch+1}/{epochs}], Loss: {loss.item():.4f}')\n",
    "\n",
    "    print(\"Finished training the model.\")\n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "id": "63830aed",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Started Dummy Training the model...\n",
      "Epoch [1/100], Loss: 1.2140\n",
      "Epoch [2/100], Loss: 0.8031\n",
      "Epoch [3/100], Loss: 1.2870\n",
      "Epoch [4/100], Loss: 1.4929\n",
      "Epoch [5/100], Loss: 1.7952\n",
      "Epoch [6/100], Loss: 0.7782\n",
      "Epoch [7/100], Loss: 0.6903\n",
      "Epoch [8/100], Loss: 0.6253\n",
      "Epoch [9/100], Loss: 0.8758\n",
      "Epoch [10/100], Loss: 0.7644\n",
      "Epoch [11/100], Loss: 0.7986\n",
      "Epoch [12/100], Loss: 0.8123\n",
      "Epoch [13/100], Loss: 0.9833\n",
      "Epoch [14/100], Loss: 0.9311\n",
      "Epoch [15/100], Loss: 1.1370\n",
      "Epoch [16/100], Loss: 1.0218\n",
      "Epoch [17/100], Loss: 0.7389\n",
      "Epoch [18/100], Loss: 0.6206\n",
      "Epoch [19/100], Loss: 1.3688\n",
      "Epoch [20/100], Loss: 0.8149\n",
      "Epoch [21/100], Loss: 1.1958\n",
      "Epoch [22/100], Loss: 0.9766\n",
      "Epoch [23/100], Loss: 0.9688\n",
      "Epoch [24/100], Loss: 0.9180\n",
      "Epoch [25/100], Loss: 0.9970\n",
      "Epoch [26/100], Loss: 0.8746\n",
      "Epoch [27/100], Loss: 1.3388\n",
      "Epoch [28/100], Loss: 0.7889\n",
      "Epoch [29/100], Loss: 1.2510\n",
      "Epoch [30/100], Loss: 1.2772\n",
      "Epoch [31/100], Loss: 1.0001\n",
      "Epoch [32/100], Loss: 0.9372\n",
      "Epoch [33/100], Loss: 1.3004\n",
      "Epoch [34/100], Loss: 1.6018\n",
      "Epoch [35/100], Loss: 0.6899\n",
      "Epoch [36/100], Loss: 1.0878\n",
      "Epoch [37/100], Loss: 1.0880\n",
      "Epoch [38/100], Loss: 0.3583\n",
      "Epoch [39/100], Loss: 0.8913\n",
      "Epoch [40/100], Loss: 1.0012\n",
      "Epoch [41/100], Loss: 1.6451\n",
      "Epoch [42/100], Loss: 1.1628\n",
      "Epoch [43/100], Loss: 1.2701\n",
      "Epoch [44/100], Loss: 1.0167\n",
      "Epoch [45/100], Loss: 0.7775\n",
      "Epoch [46/100], Loss: 0.9025\n",
      "Epoch [47/100], Loss: 0.8914\n",
      "Epoch [48/100], Loss: 1.2683\n",
      "Epoch [49/100], Loss: 0.9164\n",
      "Epoch [50/100], Loss: 1.2637\n",
      "Epoch [51/100], Loss: 1.0175\n",
      "Epoch [52/100], Loss: 1.2392\n",
      "Epoch [53/100], Loss: 0.8681\n",
      "Epoch [54/100], Loss: 0.7182\n",
      "Epoch [55/100], Loss: 1.0401\n",
      "Epoch [56/100], Loss: 1.3447\n",
      "Epoch [57/100], Loss: 1.2504\n",
      "Epoch [58/100], Loss: 1.0714\n",
      "Epoch [59/100], Loss: 1.2144\n",
      "Epoch [60/100], Loss: 0.6698\n",
      "Epoch [61/100], Loss: 1.0110\n",
      "Epoch [62/100], Loss: 0.9046\n",
      "Epoch [63/100], Loss: 1.0489\n",
      "Epoch [64/100], Loss: 0.8557\n",
      "Epoch [65/100], Loss: 0.9198\n",
      "Epoch [66/100], Loss: 0.8579\n",
      "Epoch [67/100], Loss: 0.9106\n",
      "Epoch [68/100], Loss: 1.3388\n",
      "Epoch [69/100], Loss: 0.7173\n",
      "Epoch [70/100], Loss: 0.8606\n",
      "Epoch [71/100], Loss: 1.2798\n",
      "Epoch [72/100], Loss: 1.1241\n",
      "Epoch [73/100], Loss: 1.0875\n",
      "Epoch [74/100], Loss: 1.3110\n",
      "Epoch [75/100], Loss: 0.6741\n",
      "Epoch [76/100], Loss: 1.0264\n",
      "Epoch [77/100], Loss: 0.6686\n",
      "Epoch [78/100], Loss: 1.2535\n",
      "Epoch [79/100], Loss: 1.0444\n",
      "Epoch [80/100], Loss: 0.7522\n",
      "Epoch [81/100], Loss: 0.9744\n",
      "Epoch [82/100], Loss: 0.9571\n",
      "Epoch [83/100], Loss: 0.9522\n",
      "Epoch [84/100], Loss: 1.3742\n",
      "Epoch [85/100], Loss: 0.8700\n",
      "Epoch [86/100], Loss: 0.8842\n",
      "Epoch [87/100], Loss: 0.8993\n",
      "Epoch [88/100], Loss: 0.9452\n",
      "Epoch [89/100], Loss: 0.7082\n",
      "Epoch [90/100], Loss: 1.2160\n",
      "Epoch [91/100], Loss: 0.9545\n",
      "Epoch [92/100], Loss: 1.1087\n",
      "Epoch [93/100], Loss: 0.6511\n",
      "Epoch [94/100], Loss: 1.0392\n",
      "Epoch [95/100], Loss: 0.8291\n",
      "Epoch [96/100], Loss: 1.3929\n",
      "Epoch [97/100], Loss: 1.1216\n",
      "Epoch [98/100], Loss: 1.0337\n",
      "Epoch [99/100], Loss: 0.9571\n",
      "Epoch [100/100], Loss: 0.8764\n",
      "Finished training the model.\n"
     ]
    }
   ],
   "source": [
    "model = train_model(model, epochs=100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "id": "2860a48b",
   "metadata": {},
   "outputs": [],
   "source": [
    "aux_data = torch.tensor([1.0,2.0,3.0,4.0,5.0,6.0,7.0,8.0,9.0,10.0,11.0,12.0]).to(device).unsqueeze(0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "id": "a42f76c2",
   "metadata": {},
   "outputs": [],
   "source": [
    "output = model(aux_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "id": "553b8caa",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[0.2852]], grad_fn=<AddmmBackward0>)"
      ]
     },
     "execution_count": 89,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "output"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "48d2cfac",
   "metadata": {},
   "source": [
    "Nun wollen wir dieses Model exportieren. Dafür haben wir mehrere Möglichkeiten."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cfb5079d",
   "metadata": {},
   "source": [
    "**Exportieren via State Dictionary**"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f453351f",
   "metadata": {},
   "source": [
    "Ist die am häufigsten verwendete und empfohlene Methode. Dabei wird das `state_dict` gespeichert, welches ein Python-Dictionary ist und alle trainierten Parameter beinhaltet."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "id": "2861e89a",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model state_dict saved to ../models/nn_6_simple_regressor_state_dict.pth\n"
     ]
    }
   ],
   "source": [
    "export_path = os.path.join(\"..\", \"models\", \"nn_6_simple_regressor_state_dict.pth\")\n",
    "\n",
    "torch.save(model.state_dict(), export_path)\n",
    "print(f\"Model state_dict saved to {export_path}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "id": "82118080",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Output before loading state_dict: tensor([[-0.1205]], grad_fn=<AddmmBackward0>)\n"
     ]
    }
   ],
   "source": [
    "model = SimpleRegressor().to(device)\n",
    "model.eval()\n",
    "output_before_import = model(aux_data)\n",
    "print(\"Output before loading state_dict:\", output_before_import)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 92,
   "id": "1a036642",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Output after loading state_dict: tensor([[0.2852]], grad_fn=<AddmmBackward0>)\n"
     ]
    }
   ],
   "source": [
    "# load model\n",
    "model.load_state_dict(torch.load(export_path, map_location=device))\n",
    "model.eval()\n",
    "output_after_import = model(aux_data)\n",
    "print(\"Output after loading state_dict:\", output_after_import)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "eb80d116",
   "metadata": {},
   "source": [
    "**Wichtig:** Es ist in diesem Fall natürlich auch die Struktur vom Modell zu kennen, also es werden nur die Weights importiert/exportiert. Der dazugehörige Computational Graph (sprich die `forward()` Methode) ist hier nicht dabei."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "825088f7",
   "metadata": {},
   "source": [
    "Sollte es nicht möglich sein, die Modellklasse bereit zu halten, bzw. falls wir das Modell wirklich \"komplett\" exportieren wollen, so stehen uns folgende Möglichkeiten zur Verfügung."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2f8ec630",
   "metadata": {},
   "source": [
    "### ONNX Export"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "215da5af",
   "metadata": {},
   "source": [
    "Falls wir uns für den Export via Open Neural Network Exchange entscheiden, so läuft dies folgendermaßen ab:"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "75e2342c",
   "metadata": {},
   "source": [
    "Zuerst müssen wir das Modul `onnx` installieren. Dafür verwenden wir `pip install onnx onnxscript onnxruntime`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "id": "ec27f40b",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/tmp/ipykernel_148960/3099032314.py:3: DeprecationWarning: You are using the legacy TorchScript-based ONNX export. Starting in PyTorch 2.9, the new torch.export-based ONNX exporter will be the default. To switch now, set dynamo=True in torch.onnx.export. This new exporter supports features like exporting LLMs with DynamicCache. We encourage you to try it and share feedback to help improve the experience. Learn more about the new export logic: https://pytorch.org/docs/stable/onnx_dynamo.html. For exporting control flow: https://pytorch.org/tutorials/beginner/onnx/export_control_flow_model_to_onnx_tutorial.html.\n",
      "  torch.onnx.export(model, aux_data, onnx_export_path, input_names=['input'], output_names=['output'])\n"
     ]
    }
   ],
   "source": [
    "onnx_export_path = os.path.join(\"..\", \"models\", \"nn_6_simple_regressor.onnx\")\n",
    "\n",
    "torch.onnx.export(model, aux_data, onnx_export_path, input_names=['input'], output_names=['output'])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "19a8703a",
   "metadata": {},
   "source": [
    "**Hinweis:** Es gibt hier noch weitere Argumente in der onnx.export Methode, jedoch reichen für unsere Anwendung hier die genannten Parameter."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "id": "aabab8fa",
   "metadata": {},
   "outputs": [],
   "source": [
    "del model"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "dc03976a",
   "metadata": {},
   "source": [
    "Wir können nun das Modell wieder importieren. Unter anderem können wir uns auch [hier](https://github.com/lutzroeder/netron) (es gibt auf der Seite unten einen Link zur WebApp) das Modell anzeigen."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b9e8f741",
   "metadata": {},
   "source": [
    "Wir können es aber auch in Python wieder importieren."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "id": "1fe0a837",
   "metadata": {},
   "outputs": [],
   "source": [
    "model = onnx.load(onnx_export_path)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "40d26d8a",
   "metadata": {},
   "source": [
    "**Wichtig:** Das importierte Modell ist nun nicht in PyTorch verfügbar, sondern im ONNX-Format (ONNX-Runtime)."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6b581a80",
   "metadata": {},
   "source": [
    "Zuerst überprüfen wir ob der Import funktioniert hat."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "id": "4a742475",
   "metadata": {},
   "outputs": [],
   "source": [
    "res = onnx.checker.check_model(model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "id": "b4fb9f95",
   "metadata": {},
   "outputs": [],
   "source": [
    "ort_session = ort.InferenceSession(onnx_export_path)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "b85ba62e",
   "metadata": {},
   "outputs": [],
   "source": [
    "output = ort_session.run(None, {'input': aux_data.cpu().numpy()})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 99,
   "id": "71a1b4e1",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[array([[0.28523487]], dtype=float32)]"
      ]
     },
     "execution_count": 99,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "output"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "60e01ff9",
   "metadata": {},
   "source": [
    "Der Output ist wieder gleich!"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7f5af869",
   "metadata": {},
   "source": [
    "### TorchScript Export"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5607d0f3",
   "metadata": {},
   "source": [
    "Die letzte Möglichkeit, die wir uns ansehen werden ist der sogenannte **TorchScript Export**. \n",
    "\n",
    "Es erlaubt uns, das Modell zu importieren, ohne die Modell-Struktur kennen zu müssen."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "id": "9aea4acc",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Started Dummy Training the model...\n",
      "Epoch [1/100], Loss: 0.8839\n",
      "Epoch [2/100], Loss: 1.3471\n",
      "Epoch [3/100], Loss: 1.5558\n",
      "Epoch [4/100], Loss: 1.1475\n",
      "Epoch [5/100], Loss: 0.9933\n",
      "Epoch [6/100], Loss: 1.2930\n",
      "Epoch [7/100], Loss: 1.3009\n",
      "Epoch [8/100], Loss: 1.4032\n",
      "Epoch [9/100], Loss: 1.2986\n",
      "Epoch [10/100], Loss: 1.5628\n",
      "Epoch [11/100], Loss: 1.6245\n",
      "Epoch [12/100], Loss: 0.8749\n",
      "Epoch [13/100], Loss: 1.0093\n",
      "Epoch [14/100], Loss: 0.9486\n",
      "Epoch [15/100], Loss: 1.3173\n",
      "Epoch [16/100], Loss: 1.2826\n",
      "Epoch [17/100], Loss: 1.0709\n",
      "Epoch [18/100], Loss: 0.7309\n",
      "Epoch [19/100], Loss: 0.7271\n",
      "Epoch [20/100], Loss: 0.9565\n",
      "Epoch [21/100], Loss: 0.9427\n",
      "Epoch [22/100], Loss: 0.8496\n",
      "Epoch [23/100], Loss: 1.3563\n",
      "Epoch [24/100], Loss: 1.3468\n",
      "Epoch [25/100], Loss: 0.5225\n",
      "Epoch [26/100], Loss: 1.4669\n",
      "Epoch [27/100], Loss: 1.0838\n",
      "Epoch [28/100], Loss: 1.2212\n",
      "Epoch [29/100], Loss: 1.4901\n",
      "Epoch [30/100], Loss: 0.8856\n",
      "Epoch [31/100], Loss: 0.9297\n",
      "Epoch [32/100], Loss: 1.2879\n",
      "Epoch [33/100], Loss: 0.7851\n",
      "Epoch [34/100], Loss: 0.9544\n",
      "Epoch [35/100], Loss: 0.7390\n",
      "Epoch [36/100], Loss: 0.9176\n",
      "Epoch [37/100], Loss: 0.7545\n",
      "Epoch [38/100], Loss: 0.5950\n",
      "Epoch [39/100], Loss: 1.0645\n",
      "Epoch [40/100], Loss: 0.7515\n",
      "Epoch [41/100], Loss: 1.0809\n",
      "Epoch [42/100], Loss: 1.1745\n",
      "Epoch [43/100], Loss: 1.3536\n",
      "Epoch [44/100], Loss: 1.1044\n",
      "Epoch [45/100], Loss: 1.9122\n",
      "Epoch [46/100], Loss: 1.2601\n",
      "Epoch [47/100], Loss: 0.6494\n",
      "Epoch [48/100], Loss: 1.0749\n",
      "Epoch [49/100], Loss: 0.8888\n",
      "Epoch [50/100], Loss: 1.1439\n",
      "Epoch [51/100], Loss: 1.0482\n",
      "Epoch [52/100], Loss: 1.3640\n",
      "Epoch [53/100], Loss: 0.7824\n",
      "Epoch [54/100], Loss: 1.3926\n",
      "Epoch [55/100], Loss: 0.8370\n",
      "Epoch [56/100], Loss: 0.7971\n",
      "Epoch [57/100], Loss: 0.9532\n",
      "Epoch [58/100], Loss: 0.7365\n",
      "Epoch [59/100], Loss: 0.9876\n",
      "Epoch [60/100], Loss: 0.9812\n",
      "Epoch [61/100], Loss: 1.2391\n",
      "Epoch [62/100], Loss: 1.4859\n",
      "Epoch [63/100], Loss: 1.1096\n",
      "Epoch [64/100], Loss: 0.8135\n",
      "Epoch [65/100], Loss: 1.0660\n",
      "Epoch [66/100], Loss: 0.9529\n",
      "Epoch [67/100], Loss: 0.7976\n",
      "Epoch [68/100], Loss: 1.0494\n",
      "Epoch [69/100], Loss: 0.8920\n",
      "Epoch [70/100], Loss: 0.9967\n",
      "Epoch [71/100], Loss: 0.8961\n",
      "Epoch [72/100], Loss: 0.8740\n",
      "Epoch [73/100], Loss: 0.9696\n",
      "Epoch [74/100], Loss: 0.6680\n",
      "Epoch [75/100], Loss: 0.7864\n",
      "Epoch [76/100], Loss: 1.1901\n",
      "Epoch [77/100], Loss: 0.8527\n",
      "Epoch [78/100], Loss: 0.8770\n",
      "Epoch [79/100], Loss: 1.1248\n",
      "Epoch [80/100], Loss: 0.9718\n",
      "Epoch [81/100], Loss: 1.4774\n",
      "Epoch [82/100], Loss: 1.0423\n",
      "Epoch [83/100], Loss: 0.8642\n",
      "Epoch [84/100], Loss: 1.2729\n",
      "Epoch [85/100], Loss: 1.3168\n",
      "Epoch [86/100], Loss: 1.2866\n",
      "Epoch [87/100], Loss: 0.7739\n",
      "Epoch [88/100], Loss: 1.1269\n",
      "Epoch [89/100], Loss: 0.7487\n",
      "Epoch [90/100], Loss: 1.5029\n",
      "Epoch [91/100], Loss: 1.3421\n",
      "Epoch [92/100], Loss: 1.2918\n",
      "Epoch [93/100], Loss: 0.9249\n",
      "Epoch [94/100], Loss: 1.0778\n",
      "Epoch [95/100], Loss: 1.2862\n",
      "Epoch [96/100], Loss: 0.7971\n",
      "Epoch [97/100], Loss: 0.6628\n",
      "Epoch [98/100], Loss: 0.8082\n",
      "Epoch [99/100], Loss: 0.6520\n",
      "Epoch [100/100], Loss: 1.0457\n",
      "Finished training the model.\n"
     ]
    }
   ],
   "source": [
    "model = SimpleRegressor().to(device)\n",
    "model = train_model(model, epochs=100)\n",
    "_ = model.eval()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 101,
   "id": "e8f24d61",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Output from the original model: tensor([[1.1867]], grad_fn=<AddmmBackward0>)\n"
     ]
    }
   ],
   "source": [
    "output = model(aux_data)\n",
    "print(\"Output from the original model:\", output)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 102,
   "id": "d8092502",
   "metadata": {},
   "outputs": [],
   "source": [
    "model_scripted = torch.jit.script(model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "id": "1a0f45ad",
   "metadata": {},
   "outputs": [],
   "source": [
    "scripted_model_export_path = os.path.join(\"..\", \"models\", \"nn_6_simple_regressor_scripted.pt\")\n",
    "\n",
    "model_scripted.save(scripted_model_export_path)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 104,
   "id": "51187aa4",
   "metadata": {},
   "outputs": [],
   "source": [
    "del model\n",
    "del model_scripted"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "id": "5e19a16f",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Output from the scripted model: tensor([[1.1867]], grad_fn=<AddmmBackward0>)\n"
     ]
    }
   ],
   "source": [
    "model_scripted_loaded = torch.jit.load(scripted_model_export_path).to(device)\n",
    "model_scripted_loaded.eval()\n",
    "output_scripted = model_scripted_loaded(aux_data)\n",
    "print(\"Output from the scripted model:\", output_scripted)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ba55efe4",
   "metadata": {},
   "source": [
    "Im Vergleich zu *ONNX* hat *TorchScript* den Nachteil, dass es nur für Torch Systeme funktioniert (Python bzw. C++). "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9e75eb7e",
   "metadata": {},
   "source": [
    "# Finetuning"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c5626c94",
   "metadata": {},
   "source": [
    "Passend zu unserer bisherigen Thematik mit Model Import und Model Export, wollen wir uns nun über FineTuning unterhalten. Dabei geht es darum, dass wir ein bestehendes Modell etwas anpassen. Dabei wollen wir entweder nochmal auf ein spezielles Dataset zusäztlich trainieren, oder wir wollen sogar einen Teil vom neuronalen Netzwerk mit einem neuen Teil ersetzen."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b2069416",
   "metadata": {},
   "source": [
    "Für die zwei oben genannten Szenarien stelle man sich folgende Beispiele vor:\n",
    "1) Man will einen \"Copilot\" zum Python-Programmieren bauen. Dabei nimmt man sich ein bereits funktionierendes Language Model, welches vielleicht generell auf Text, sprich alles im Internet auffindbare, trainiert wurde. Nun verwendet man nochmal ein explizites Dataset, welches nur aus Python-Code besteht und lässt das Modell nochmal einige Iterationen mit diesen Daten trainieren. Dieses Modell sollte dann das Programmieren schneller lernen, als ein ganz neues Modell, weil unser Modell ja schon vorher \"lesen\" gelernt hat und somit unsere Sprache bereits versteht.\n",
    "2) Man hat ein sehr generelles Modell, welches gelernt hat Bilder in viele (sagen wir 1000) Klassen zu teilen. Nun möchte man selber einen Bildklassifizierer bauen, der aber nur zwischen Autos, Katzen, Obst und Menschen unterscheiden soll. Dann nimmt man das vorgefertigte Modell und ändert am Schluss nur das letzte Layer und ändert es zu einem (in unserem Fall) `nn.Linear(Z, 4)` Layer. Dabei steht $Z$ für die Output Dimension vom vorigen Layer. Danach trainieren wir nochmal mit unseren Daten. Auch hier nutzen wir dann den Vorteil aus, dass das Modell schon ein Grundverständnis von unserer Welt (wie kann ich Objekte erkennen und vom Hintergrund unterscheiden usw.) hat und auch dieses Modell sollte viel schneller trainieren bzw. eine bessere Performance erreichen."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d7c1307e",
   "metadata": {},
   "source": [
    "**Hinweis:** Solche Modelle nennt man dann oft *Foundation Models*."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ba6eab3b",
   "metadata": {},
   "source": [
    "**Hinweis:** Bei Methode 2 können wir auch mehrere Layers austauschen."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "285f4bea",
   "metadata": {},
   "source": [
    "Sehen wir uns nun an, wie wir ein bestehendes Modell finetunen können. "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7611d537",
   "metadata": {},
   "source": [
    "Dazu möchten wir an dieser Stelle auch zeigen, wie man sich Modelle aus PyTorch herunterladen kann, welche für die Öffentlichkeit zur Verfügung gestellt werden. Als Beispiel nehmen wir das Bildklassifizierungsmodell `ResNet18` [Dokumentation](https://docs.pytorch.org/vision/main/models/generated/torchvision.models.resnet18.html)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 106,
   "id": "f8968cad",
   "metadata": {},
   "outputs": [],
   "source": [
    "from torchvision import models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 107,
   "id": "1305fc80",
   "metadata": {},
   "outputs": [],
   "source": [
    "model = models.resnet18(weights = models.ResNet18_Weights.IMAGENET1K_V1)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "62fbbd25",
   "metadata": {},
   "source": [
    "Dieses Modell können wir nun nutzen."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ecf7a7ac",
   "metadata": {},
   "source": [
    "Für das FineTuning können wir nun auf die Architektur zugreifen. Wir erklären hier nur kurz den zweiten Ansatz. Der erste ist, sobald wir das Modell haben, selbsterklärend."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 108,
   "id": "09ec615b",
   "metadata": {},
   "outputs": [],
   "source": [
    "num_ftrs = model.fc.in_features\n",
    "model.fc = nn.Linear(num_ftrs, 4)  # New classes are: Auto, Katze, Obst and Mensch"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1844f9f9",
   "metadata": {},
   "source": [
    "Nun können wir unterscheiden in 2 Teile:\n",
    "* Wir trainieren alle Weights neu\n",
    "* Wir trainieren und ändern nur die Weights für den geänderten Layer\n",
    "\n",
    "Für den ersten Punkt müssen wir nichts machen. Sollten wir aber nur das letzte Layer ändern wollen, so müssen wir bei allen anderen Parametern die \"Differenzierbarkeit\", wegnehmen. Das machen wir mit folgendem Code."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 109,
   "id": "45df698a",
   "metadata": {},
   "outputs": [],
   "source": [
    "for name, param in model.named_parameters():\n",
    "    if \"fc\" in name: # Only the last fully connected layer is changed\n",
    "        param.requires_grad = True \n",
    "    else:\n",
    "        param.requires_grad = False # Those weights should not be updated during training. This is called \"freezing\" the layers."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "356be40c",
   "metadata": {},
   "source": [
    "Das müssen wir dann auch noch dem Optimizer so mitteilen."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 110,
   "id": "b7730b4b",
   "metadata": {},
   "outputs": [],
   "source": [
    "optimizer = torch.optim.SGD(\n",
    "    filter(lambda p: p.requires_grad, model.parameters()), \n",
    "    lr=0.001, \n",
    "    momentum=0.9\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ceaa5a17",
   "metadata": {},
   "source": [
    "Nun können wir dieses Model **finetunen**."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e87de599",
   "metadata": {},
   "source": [
    "# Model Deployment mit Flask"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4245a104",
   "metadata": {},
   "source": [
    "Zuguterletzt wollen wir uns eine Möglichkeit ansehen, wie wir unser Modell selber hosten können. Wir verwenden dazu Flask."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d662da75",
   "metadata": {},
   "source": [
    "**Hinweis:** Es gibt noch viele weitere andere Frameworks, wir werden uns aber nur mit Flask beschäftigen."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cb19e663",
   "metadata": {},
   "source": [
    "**Hinweis:** Natürlich kann wie vorher erwähnt das Modell auch in einem entsprechenden Format (zum Beipsie Open Neural Network Exchange (**ONNX**)) exportiert werden und in einer anderen Programmiersprache eingebunden werden. Etwaiger Hardware-Support (GPU) geht dabei eventuell verloren bzw. muss selber implementiert werden."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "dbba0be9",
   "metadata": {},
   "source": [
    "Wir wechseln nun auf das Python File [nn_6_flask_app](nn_6_flask_app.py). Sobald wir dort den Service gestartet haben, gehen wir wieder hierher zurück."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6297559e",
   "metadata": {},
   "source": [
    "Wir können nun unser Modell testen."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 111,
   "id": "c8a7c549",
   "metadata": {},
   "outputs": [],
   "source": [
    "import requests\n",
    "import os\n",
    "import matplotlib.pyplot as plt\n",
    "from PIL import Image"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 112,
   "id": "f90a2efc",
   "metadata": {},
   "outputs": [],
   "source": [
    "url = \"http://127.0.0.1:5665/predict\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 113,
   "id": "143b4ff0",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "path_of_image = os.path.join(\"..\", \"resources\", \"Columbus.jpeg\")\n",
    "\n",
    "image = Image.open(path_of_image)\n",
    "plt.imshow(image)\n",
    "plt.axis(\"off\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 114,
   "id": "b6474ea0",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Response status code: 200\n"
     ]
    }
   ],
   "source": [
    "with open(path_of_image, \"rb\") as f:\n",
    "    files = {\n",
    "        \"image\": (path_of_image, f, \"image/jpeg\")  # <--- WICHTIG: Filename + MIME-Type\n",
    "    }\n",
    "    response = requests.post(url, files=files)\n",
    "    print(f\"Response status code: {response.status_code}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 115,
   "id": "9f04d32e",
   "metadata": {},
   "outputs": [],
   "source": [
    "class_labels_path = os.path.join(\"..\", \"resources\", \"imagenet_classes.txt\")\n",
    "with open(class_labels_path) as f:\n",
    "    class_labels = [line.strip() for line in f.readlines()]\n",
    "\n",
    "def get_class_name(response, class_labels):\n",
    "    class_label_response = response.json()\n",
    "    class_index = class_label_response[\"predicted_class_index\"]\n",
    "    class_name = class_labels[class_index]\n",
    "    return class_name, class_index"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 116,
   "id": "0e258583",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "('tabby', 281)"
      ]
     },
     "execution_count": 116,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "get_class_name(response, class_labels)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cece33fe",
   "metadata": {},
   "source": [
    "(kann auch nochmal [hier](https://gist.github.com/yrevar/942d3a0ac09ec9e5eb3a) verglichen werden)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e61b39be",
   "metadata": {},
   "source": [
    "Bemerkenswert ist, dass dies nun auch auf der **GPU** läuft, falls verfügbar."
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "dsai",
   "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": "3.9.23"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}