PyTorch Implementation Causal Neural Network.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "widgets": {
      "application/vnd.jupyter.widget-state+json": {
        "21f580a934244ab7a7dc81f7c3a96056": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HBoxModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HBoxView",
            "box_style": "",
            "children": [
              "IPY_MODEL_50447b13c0c74893827782b8f43b08ac",
              "IPY_MODEL_aac9cd8d6dc34800978158a61142cc12",
              "IPY_MODEL_eb7514eaef82491b876c3428398d36e8"
            ],
            "layout": "IPY_MODEL_0ed15cb565a04b2cb2e623396d188215"
          }
        },
        "50447b13c0c74893827782b8f43b08ac": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HTMLModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HTMLView",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_69f92189a487448aa767238307afa650",
            "placeholder": "​",
            "style": "IPY_MODEL_5118328b79b547fba4494e6c27a11653",
            "value": "i: 551 | loss: 1.85329592 | CATE: 0.472:   6%"
          }
        },
        "aac9cd8d6dc34800978158a61142cc12": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "FloatProgressModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "ProgressView",
            "bar_style": "danger",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_53d3bbac346340b28da726ea2fa627e5",
            "max": 10000,
            "min": 0,
            "orientation": "horizontal",
            "style": "IPY_MODEL_f56deae597074424a12415307fa643b1",
            "value": 551
          }
        },
        "eb7514eaef82491b876c3428398d36e8": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HTMLModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HTMLView",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_852fb823cba84da8b030b089deeecc5c",
            "placeholder": "​",
            "style": "IPY_MODEL_e5427b39c9f8478d988d0be4a8dbd40f",
            "value": " 551/10000 [00:35&lt;05:30, 28.63it/s]"
          }
        },
        "0ed15cb565a04b2cb2e623396d188215": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "69f92189a487448aa767238307afa650": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "5118328b79b547fba4494e6c27a11653": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "DescriptionStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "description_width": ""
          }
        },
        "53d3bbac346340b28da726ea2fa627e5": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "f56deae597074424a12415307fa643b1": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "ProgressStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "bar_color": null,
            "description_width": ""
          }
        },
        "852fb823cba84da8b030b089deeecc5c": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "e5427b39c9f8478d988d0be4a8dbd40f": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "DescriptionStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "description_width": ""
          }
        }
      }
    }
  },
  "cells": [
    {
      "cell_type": "code",
      "source": [
        "%%bash\n",
        "pip install torchview"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "FNLYB-fAAwq8",
        "outputId": "9cd77188-a4ca-4af8-983f-6a879c97e11a"
      },
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Requirement already satisfied: torchview in /usr/local/lib/python3.10/dist-packages (0.2.6)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Computing the Conditional Average Treatment Effect (CATE) with Neural Networks"
      ],
      "metadata": {
        "id": "9h9_l9faEweK"
      }
    },
    {
      "cell_type": "code",
      "execution_count": 2,
      "metadata": {
        "id": "8F09549GEiMy"
      },
      "outputs": [],
      "source": [
        "import torch\n",
        "import torch.nn as nn\n",
        "import torch.nn.functional as F\n",
        "import numpy as np\n",
        "import seaborn as sns\n",
        "import matplotlib.pyplot as graph\n",
        "from scipy import stats\n",
        "from scipy.special import expit\n",
        "from sklearn.datasets import make_circles, make_moons\n",
        "from sklearn.model_selection import train_test_split\n",
        "from sklearn.metrics import accuracy_score\n",
        "from tqdm.auto import tqdm, trange\n",
        "from torchview import draw_graph\n",
        "from IPython.display import Markdown\n",
        "graph.style.use('fivethirtyeight')\n",
        "graph.rcParams['lines.linewidth'] = 1"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Setup quick fake data\n",
        "n, p = int(1e4), 20\n",
        "treatment_effect = 0.5\n",
        "\n",
        "m = stats.norm().rvs(p)\n",
        "x = stats.norm().rvs((n, p))\n",
        "# treatment_labels = stats.bernoulli(0.2).rvs(n)  # Random assignment (Unconfounded)\n",
        "treatment_labels = expit((x[:, :5] @ m[:5]) + 2*stats.norm().rvs(n)).round()  # Conditional assignment (Confounding)\n",
        "\n",
        "y = x@m + treatment_effect*treatment_labels + stats.norm().rvs(n)\n",
        "\n",
        "print(x.shape, y.shape, treatment_labels.shape)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "d0fHNoSiFzmu",
        "outputId": "972d36ec-df8e-4cb5-bd05-de81ea800fc6"
      },
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "(10000, 20) (10000,) (10000,)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Naive Estimate\n",
        "sns.histplot(y[treatment_labels.astype(bool)], label='Treatment')\n",
        "sns.histplot(y[~treatment_labels.astype(bool)], label='Control')\n",
        "graph.legend()\n",
        "graph.show()\n",
        "\n",
        "control_mean = y[~treatment_labels.astype(bool)].mean()\n",
        "treatment_mean = y[treatment_labels.astype(bool)].mean()\n",
        "naive_causal_estimate = treatment_mean - control_mean\n",
        "\n",
        "print(f'True Treatment Effect = {treatment_effect}')\n",
        "print(f'Naive Estimate of the Treatment Effect = {naive_causal_estimate:2f}')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 484
        },
        "id": "TsydFGiRF9o-",
        "outputId": "88b1df9d-4d03-46d0-a57c-463d91f18d57"
      },
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "True Treatment Effect = 0.5\n",
            "Naive Estimate of the Treatment Effect = 4.522126\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "$$CATE = {1 \\over N} \\sum [ f_{treatment}(x) - f_{control}(x) ] $$\n",
        "\n",
        "$$J(\\theta) = {1 \\over N} \\sum [ (f_{t_i}(x_i) - y_i)^2 + \\mathcal{H}_{cross}(g(x_i), t_i) ] $$\n",
        "\n",
        "By optimising $J(\\theta)$ we are sure that we are optimising the CATE directly. This is both the model _must_ learn a covariate set in $Z$ that contains the information for estimating the propensity score. This should be _equivalent_ to adjusting for the propensity score."
      ],
      "metadata": {
        "id": "uPc5PYlhFZhk"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Specify the models\n",
        "class LatentModel(nn.Module):\n",
        "    def __init__(self, input_dim, output_dim):\n",
        "        super().__init__()\n",
        "        self.input_dim = input_dim\n",
        "        self.output_dim = output_dim\n",
        "\n",
        "        self.dnn_z = nn.Sequential(\n",
        "            nn.Linear(input_dim, 75),\n",
        "            nn.ELU(),\n",
        "            nn.BatchNorm1d(75),\n",
        "            nn.LazyLinear(output_dim),\n",
        "            nn.ELU(),\n",
        "            nn.BatchNorm1d(output_dim)\n",
        "        )\n",
        "\n",
        "    def forward(self, x):\n",
        "        return self.dnn_z(x)\n",
        "\n",
        "\n",
        "class PropensityModel(nn.Module):\n",
        "    def __init__(self):\n",
        "        super().__init__()\n",
        "        self.propensity = nn.Sequential(\n",
        "            nn.LazyLinear(1),\n",
        "            nn.Sigmoid(),\n",
        "        )\n",
        "\n",
        "    def forward(self, x):\n",
        "        return self.propensity(x)\n",
        "\n",
        "\n",
        "class OutcomeModel(nn.Module):\n",
        "    def __init__(self):\n",
        "        super().__init__()\n",
        "        self.outcome_dnn = nn.Sequential(\n",
        "            nn.LazyLinear(75),\n",
        "            nn.ELU(),\n",
        "            nn.LazyLinear(1),\n",
        "        )\n",
        "\n",
        "    def forward(self, x):\n",
        "        return self.outcome_dnn(x)\n",
        "\n",
        "# Contruct the model\n",
        "class CausalModel(nn.Module):\n",
        "    def __init__(self, input_dim):\n",
        "        super().__init__()\n",
        "        self.input_dim = input_dim\n",
        "\n",
        "        # Required modules\n",
        "        self.z_model = LatentModel(input_dim, 1)\n",
        "        self.propensity_model = PropensityModel()\n",
        "        self.q_given_treamtent = OutcomeModel()\n",
        "        self.q_given_control = OutcomeModel()\n",
        "\n",
        "    def forward(self, x):\n",
        "        z = self.z_model(x)\n",
        "        g = self.propensity_model(z)\n",
        "        q_t = self.q_given_treamtent(z)\n",
        "        q_c = self.q_given_control(z)\n",
        "        return g, q_t, q_c\n",
        "\n",
        "    def propensity(self, x):\n",
        "        z = self.z_model(x)\n",
        "        g = self.propensity_model(x)\n",
        "        return g\n",
        "\n",
        "    def q(self, x, given):\n",
        "        z = self.z_model(x)\n",
        "        if given.lower() == 't':\n",
        "            return self.q_given_treamtent(x)\n",
        "        elif given.lower() == 'c':\n",
        "            return self.q_given_control(x)\n",
        "\n",
        "    def ite(self, x):\n",
        "        z = self.z_model(x)\n",
        "        return self.q_given_treamtent(z) - self.q_given_control(z)\n",
        "\n",
        "class CausalLoss(nn.modules.loss._Loss):\n",
        "    def __init__(self):\n",
        "        super().__init__()\n",
        "\n",
        "    def forward(self, g_hat, is_treatment, qt_hat, qc_hat, y_true):\n",
        "        # Q MSE terms\n",
        "        treatment_mse = (is_treatment) * (qt_hat - y_true)**2\n",
        "        control_mse = (1-is_treatment) * (qc_hat - y_true)**2\n",
        "        q_mse = torch.mean(treatment_mse) + torch.mean(control_mse)\n",
        "\n",
        "        # Binary cross entropy term\n",
        "        g_entropy = F.binary_cross_entropy(g_hat, is_treatment)\n",
        "\n",
        "        return q_mse + g_entropy"
      ],
      "metadata": {
        "id": "aiRKNmHIE6R6"
      },
      "execution_count": 5,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# Contruct the model\n",
        "causal_model = CausalModel(p)\n",
        "optimizer = torch.optim.Adam(causal_model.parameters(), lr=1e-3)\n",
        "loss_func = CausalLoss()\n",
        "\n",
        "# Prep model data\n",
        "x_train, x_test, y_train, y_test, is_treatment_train, is_treatment_test = train_test_split(\n",
        "    x, y, treatment_labels,\n",
        "    test_size=0.2\n",
        ")\n",
        "x_train = torch.from_numpy(x_train).float()\n",
        "y_train = torch.from_numpy(y_train.reshape(-1, 1)).float()\n",
        "x_test = torch.from_numpy(x_test).float()\n",
        "y_test = torch.from_numpy(y_test.reshape(-1, 1)).float()\n",
        "is_treatment_train = torch.from_numpy(is_treatment_train.reshape(-1, 1)).float()\n",
        "is_treatment_test = torch.from_numpy(is_treatment_test.reshape(-1, 1)).float()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "DFEv0xncHpvj",
        "outputId": "b142ccc9-7e7e-4881-daa0-e7b6f69bdbd5"
      },
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.10/dist-packages/torch/nn/modules/lazy.py:180: UserWarning: Lazy modules are a new feature under heavy development so changes to the API or functionality can happen at any moment.\n",
            "  warnings.warn('Lazy modules are a new feature under heavy development '\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "display(Markdown('## Model Summary'))\n",
        "display(draw_graph(causal_model, input_size=x_train.shape, depth=1).visual_graph)\n",
        "\n",
        "display(Markdown('## Model'))\n",
        "display(draw_graph(causal_model, input_size=x_train.shape, expand_nested=True).visual_graph)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "f98YW6XcAldz",
        "outputId": "b2cb193c-4b66-45c5-c018-f198e260349e"
      },
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<IPython.core.display.Markdown object>"
            ],
            "text/markdown": "## Model Summary"
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "image/svg+xml": "<?xml version=\"1.0\" encoding=\"UTF-8\" standalone=\"no\"?>\n<!DOCTYPE svg PUBLIC \"-//W3C//DTD SVG 1.1//EN\"\n \"http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd\">\n<!-- Generated by graphviz version 2.43.0 (0)\n -->\n<!-- Title: model Pages: 1 -->\n<svg width=\"701pt\" height=\"264pt\"\n viewBox=\"0.00 0.00 701.00 264.00\" xmlns=\"http://www.w3.org/2000/svg\" xmlns:xlink=\"http://www.w3.org/1999/xlink\">\n<g id=\"graph0\" class=\"graph\" transform=\"scale(1 1) rotate(0) translate(4 260)\">\n<title>model</title>\n<polygon fill=\"white\" stroke=\"transparent\" points=\"-4,4 -4,-260 697,-260 697,4 -4,4\"/>\n<!-- 0 -->\n<g id=\"node1\" class=\"node\">\n<title>0</title>\n<polygon fill=\"lightyellow\" stroke=\"transparent\" points=\"432.5,-256 278.5,-256 278.5,-224 432.5,-224 432.5,-256\"/>\n<polygon fill=\"none\" stroke=\"black\" points=\"278.5,-224 278.5,-256 361.5,-256 361.5,-224 278.5,-224\"/>\n<text text-anchor=\"start\" x=\"283.5\" y=\"-243\" font-family=\"Linux libertine\" font-size=\"10.00\">input&#45;tensor</text>\n<text text-anchor=\"start\" x=\"298.5\" y=\"-232\" font-family=\"Linux libertine\" font-size=\"10.00\">depth:0</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"361.5,-224 361.5,-256 432.5,-256 432.5,-224 361.5,-224\"/>\n<text text-anchor=\"start\" x=\"366.5\" y=\"-237.5\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 20)</text>\n</g>\n<!-- 1 -->\n<g id=\"node2\" class=\"node\">\n<title>1</title>\n<polygon fill=\"#c1ffc1\" stroke=\"transparent\" points=\"462,-188 249,-188 249,-146 462,-146 462,-188\"/>\n<polygon fill=\"none\" stroke=\"black\" points=\"249.5,-146 249.5,-188 326.5,-188 326.5,-146 249.5,-146\"/>\n<text text-anchor=\"start\" x=\"254.5\" y=\"-170\" font-family=\"Linux libertine\" font-size=\"10.00\">LatentModel</text>\n<text text-anchor=\"start\" x=\"266.5\" y=\"-159\" font-family=\"Linux libertine\" font-size=\"10.00\">depth:1</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"326.5,-167 326.5,-188 385.5,-188 385.5,-167 326.5,-167\"/>\n<text text-anchor=\"start\" x=\"337.5\" y=\"-175\" font-family=\"Linux libertine\" font-size=\"10.00\">input:</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"385.5,-167 385.5,-188 462.5,-188 462.5,-167 385.5,-167\"/>\n<text text-anchor=\"start\" x=\"390.5\" y=\"-175\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 20) </text>\n<polygon fill=\"none\" stroke=\"black\" points=\"326.5,-146 326.5,-167 385.5,-167 385.5,-146 326.5,-146\"/>\n<text text-anchor=\"start\" x=\"331.5\" y=\"-154\" font-family=\"Linux libertine\" font-size=\"10.00\">output: </text>\n<polygon fill=\"none\" stroke=\"black\" points=\"385.5,-146 385.5,-167 462.5,-167 462.5,-146 385.5,-146\"/>\n<text text-anchor=\"start\" x=\"393.5\" y=\"-154\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 1) </text>\n</g>\n<!-- 0&#45;&gt;1 -->\n<g id=\"edge1\" class=\"edge\">\n<title>0&#45;&gt;1</title>\n<path fill=\"none\" stroke=\"black\" d=\"M355.5,-223.94C355.5,-216.45 355.5,-207.12 355.5,-198.24\"/>\n<polygon fill=\"black\" stroke=\"black\" points=\"359,-198.16 355.5,-188.16 352,-198.16 359,-198.16\"/>\n</g>\n<!-- 2 -->\n<g id=\"node3\" class=\"node\">\n<title>2</title>\n<polygon fill=\"#c1ffc1\" stroke=\"transparent\" points=\"231,-110 0,-110 0,-68 231,-68 231,-110\"/>\n<polygon fill=\"none\" stroke=\"black\" points=\"0.5,-68 0.5,-110 101.5,-110 101.5,-68 0.5,-68\"/>\n<text text-anchor=\"start\" x=\"5.5\" y=\"-92\" font-family=\"Linux libertine\" font-size=\"10.00\">PropensityModel</text>\n<text text-anchor=\"start\" x=\"29.5\" y=\"-81\" font-family=\"Linux libertine\" font-size=\"10.00\">depth:1</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"101.5,-89 101.5,-110 160.5,-110 160.5,-89 101.5,-89\"/>\n<text text-anchor=\"start\" x=\"112.5\" y=\"-97\" font-family=\"Linux libertine\" font-size=\"10.00\">input:</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"160.5,-89 160.5,-110 231.5,-110 231.5,-89 160.5,-89\"/>\n<text text-anchor=\"start\" x=\"165.5\" y=\"-97\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 1) </text>\n<polygon fill=\"none\" stroke=\"black\" points=\"101.5,-68 101.5,-89 160.5,-89 160.5,-68 101.5,-68\"/>\n<text text-anchor=\"start\" x=\"106.5\" y=\"-76\" font-family=\"Linux libertine\" font-size=\"10.00\">output: </text>\n<polygon fill=\"none\" stroke=\"black\" points=\"160.5,-68 160.5,-89 231.5,-89 231.5,-68 160.5,-68\"/>\n<text text-anchor=\"start\" x=\"165.5\" y=\"-76\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 1) </text>\n</g>\n<!-- 1&#45;&gt;2 -->\n<g id=\"edge2\" class=\"edge\">\n<title>1&#45;&gt;2</title>\n<path fill=\"none\" stroke=\"black\" d=\"M292.13,-145.93C260.28,-135.85 221.52,-123.57 188.43,-113.09\"/>\n<polygon fill=\"black\" stroke=\"black\" points=\"189.4,-109.73 178.81,-110.05 187.29,-116.4 189.4,-109.73\"/>\n</g>\n<!-- 4 -->\n<g id=\"node5\" class=\"node\">\n<title>4</title>\n<polygon fill=\"#c1ffc1\" stroke=\"transparent\" points=\"462,-110 249,-110 249,-68 462,-68 462,-110\"/>\n<polygon fill=\"none\" stroke=\"black\" points=\"249.5,-68 249.5,-110 332.5,-110 332.5,-68 249.5,-68\"/>\n<text text-anchor=\"start\" x=\"254.5\" y=\"-92\" font-family=\"Linux libertine\" font-size=\"10.00\">OutcomeModel</text>\n<text text-anchor=\"start\" x=\"269.5\" y=\"-81\" font-family=\"Linux libertine\" font-size=\"10.00\">depth:1</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"332.5,-89 332.5,-110 391.5,-110 391.5,-89 332.5,-89\"/>\n<text text-anchor=\"start\" x=\"343.5\" y=\"-97\" font-family=\"Linux libertine\" font-size=\"10.00\">input:</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"391.5,-89 391.5,-110 462.5,-110 462.5,-89 391.5,-89\"/>\n<text text-anchor=\"start\" x=\"396.5\" y=\"-97\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 1) </text>\n<polygon fill=\"none\" stroke=\"black\" points=\"332.5,-68 332.5,-89 391.5,-89 391.5,-68 332.5,-68\"/>\n<text text-anchor=\"start\" x=\"337.5\" y=\"-76\" font-family=\"Linux libertine\" font-size=\"10.00\">output: </text>\n<polygon fill=\"none\" stroke=\"black\" points=\"391.5,-68 391.5,-89 462.5,-89 462.5,-68 391.5,-68\"/>\n<text text-anchor=\"start\" x=\"396.5\" y=\"-76\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 1) </text>\n</g>\n<!-- 1&#45;&gt;4 -->\n<g id=\"edge3\" class=\"edge\">\n<title>1&#45;&gt;4</title>\n<path fill=\"none\" stroke=\"black\" d=\"M355.5,-145.63C355.5,-137.82 355.5,-128.73 355.5,-120.18\"/>\n<polygon fill=\"black\" stroke=\"black\" points=\"359,-120.16 355.5,-110.16 352,-120.16 359,-120.16\"/>\n</g>\n<!-- 6 -->\n<g id=\"node7\" class=\"node\">\n<title>6</title>\n<polygon fill=\"#c1ffc1\" stroke=\"transparent\" points=\"693,-110 480,-110 480,-68 693,-68 693,-110\"/>\n<polygon fill=\"none\" stroke=\"black\" points=\"480.5,-68 480.5,-110 563.5,-110 563.5,-68 480.5,-68\"/>\n<text text-anchor=\"start\" x=\"485.5\" y=\"-92\" font-family=\"Linux libertine\" font-size=\"10.00\">OutcomeModel</text>\n<text text-anchor=\"start\" x=\"500.5\" y=\"-81\" font-family=\"Linux libertine\" font-size=\"10.00\">depth:1</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"563.5,-89 563.5,-110 622.5,-110 622.5,-89 563.5,-89\"/>\n<text text-anchor=\"start\" x=\"574.5\" y=\"-97\" font-family=\"Linux libertine\" font-size=\"10.00\">input:</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"622.5,-89 622.5,-110 693.5,-110 693.5,-89 622.5,-89\"/>\n<text text-anchor=\"start\" x=\"627.5\" y=\"-97\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 1) </text>\n<polygon fill=\"none\" stroke=\"black\" points=\"563.5,-68 563.5,-89 622.5,-89 622.5,-68 563.5,-68\"/>\n<text text-anchor=\"start\" x=\"568.5\" y=\"-76\" font-family=\"Linux libertine\" font-size=\"10.00\">output: </text>\n<polygon fill=\"none\" stroke=\"black\" points=\"622.5,-68 622.5,-89 693.5,-89 693.5,-68 622.5,-68\"/>\n<text text-anchor=\"start\" x=\"627.5\" y=\"-76\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 1) </text>\n</g>\n<!-- 1&#45;&gt;6 -->\n<g id=\"edge4\" class=\"edge\">\n<title>1&#45;&gt;6</title>\n<path fill=\"none\" stroke=\"black\" d=\"M416.49,-145.93C447.02,-135.89 484.14,-123.68 515.89,-113.23\"/>\n<polygon fill=\"black\" stroke=\"black\" points=\"517.16,-116.5 525.56,-110.05 514.97,-109.85 517.16,-116.5\"/>\n</g>\n<!-- 3 -->\n<g id=\"node4\" class=\"node\">\n<title>3</title>\n<polygon fill=\"lightyellow\" stroke=\"transparent\" points=\"192.5,-32 38.5,-32 38.5,0 192.5,0 192.5,-32\"/>\n<polygon fill=\"none\" stroke=\"black\" points=\"38.5,0 38.5,-32 127.5,-32 127.5,0 38.5,0\"/>\n<text text-anchor=\"start\" x=\"43.5\" y=\"-19\" font-family=\"Linux libertine\" font-size=\"10.00\">output&#45;tensor</text>\n<text text-anchor=\"start\" x=\"61.5\" y=\"-8\" font-family=\"Linux libertine\" font-size=\"10.00\">depth:0</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"127.5,0 127.5,-32 192.5,-32 192.5,0 127.5,0\"/>\n<text text-anchor=\"start\" x=\"132.5\" y=\"-13.5\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 1)</text>\n</g>\n<!-- 2&#45;&gt;3 -->\n<g id=\"edge5\" class=\"edge\">\n<title>2&#45;&gt;3</title>\n<path fill=\"none\" stroke=\"black\" d=\"M115.5,-67.84C115.5,-59.89 115.5,-50.66 115.5,-42.26\"/>\n<polygon fill=\"black\" stroke=\"black\" points=\"119,-42.24 115.5,-32.24 112,-42.24 119,-42.24\"/>\n</g>\n<!-- 5 -->\n<g id=\"node6\" class=\"node\">\n<title>5</title>\n<polygon fill=\"lightyellow\" stroke=\"transparent\" points=\"432.5,-32 278.5,-32 278.5,0 432.5,0 432.5,-32\"/>\n<polygon fill=\"none\" stroke=\"black\" points=\"278.5,0 278.5,-32 367.5,-32 367.5,0 278.5,0\"/>\n<text text-anchor=\"start\" x=\"283.5\" y=\"-19\" font-family=\"Linux libertine\" font-size=\"10.00\">output&#45;tensor</text>\n<text text-anchor=\"start\" x=\"301.5\" y=\"-8\" font-family=\"Linux libertine\" font-size=\"10.00\">depth:0</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"367.5,0 367.5,-32 432.5,-32 432.5,0 367.5,0\"/>\n<text text-anchor=\"start\" x=\"372.5\" y=\"-13.5\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 1)</text>\n</g>\n<!-- 4&#45;&gt;5 -->\n<g id=\"edge6\" class=\"edge\">\n<title>4&#45;&gt;5</title>\n<path fill=\"none\" stroke=\"black\" d=\"M355.5,-67.84C355.5,-59.89 355.5,-50.66 355.5,-42.26\"/>\n<polygon fill=\"black\" stroke=\"black\" points=\"359,-42.24 355.5,-32.24 352,-42.24 359,-42.24\"/>\n</g>\n<!-- 7 -->\n<g id=\"node8\" class=\"node\">\n<title>7</title>\n<polygon fill=\"lightyellow\" stroke=\"transparent\" points=\"663.5,-32 509.5,-32 509.5,0 663.5,0 663.5,-32\"/>\n<polygon fill=\"none\" stroke=\"black\" points=\"509.5,0 509.5,-32 598.5,-32 598.5,0 509.5,0\"/>\n<text text-anchor=\"start\" x=\"514.5\" y=\"-19\" font-family=\"Linux libertine\" font-size=\"10.00\">output&#45;tensor</text>\n<text text-anchor=\"start\" x=\"532.5\" y=\"-8\" font-family=\"Linux libertine\" font-size=\"10.00\">depth:0</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"598.5,0 598.5,-32 663.5,-32 663.5,0 598.5,0\"/>\n<text text-anchor=\"start\" x=\"603.5\" y=\"-13.5\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 1)</text>\n</g>\n<!-- 6&#45;&gt;7 -->\n<g id=\"edge7\" class=\"edge\">\n<title>6&#45;&gt;7</title>\n<path fill=\"none\" stroke=\"black\" d=\"M586.5,-67.84C586.5,-59.89 586.5,-50.66 586.5,-42.26\"/>\n<polygon fill=\"black\" stroke=\"black\" points=\"590,-42.24 586.5,-32.24 583,-42.24 590,-42.24\"/>\n</g>\n</g>\n</svg>\n",
            "text/plain": [
              "<graphviz.graphs.Digraph at 0x7fbafe568850>"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<IPython.core.display.Markdown object>"
            ],
            "text/markdown": "## Model"
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "image/svg+xml": "<?xml version=\"1.0\" encoding=\"UTF-8\" standalone=\"no\"?>\n<!DOCTYPE svg PUBLIC \"-//W3C//DTD SVG 1.1//EN\"\n \"http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd\">\n<!-- Generated by graphviz version 2.43.0 (0)\n -->\n<!-- Title: model Pages: 1 -->\n<svg width=\"700pt\" height=\"910pt\"\n viewBox=\"0.00 0.00 700.00 910.00\" xmlns=\"http://www.w3.org/2000/svg\" xmlns:xlink=\"http://www.w3.org/1999/xlink\">\n<g id=\"graph0\" class=\"graph\" transform=\"scale(1 1) rotate(0) translate(4 906)\">\n<title>model</title>\n<polygon fill=\"white\" stroke=\"transparent\" points=\"-4,4 -4,-906 696,-906 696,4 -4,4\"/>\n<g id=\"clust1\" class=\"cluster\">\n<title>cluster_2</title>\n<polygon fill=\"none\" stroke=\"black\" stroke-dasharray=\"5,2\" points=\"221,-356 221,-862 465,-862 465,-356 221,-356\"/>\n<text text-anchor=\"middle\" x=\"259.5\" y=\"-848.4\" font-family=\"Times,serif\" font-size=\"12.00\">LatentModel</text>\n</g>\n<g id=\"clust2\" class=\"cluster\">\n<title>cluster_3</title>\n<polygon fill=\"none\" stroke=\"black\" stroke-dasharray=\"5,2\" points=\"229,-364 229,-833 457,-833 457,-364 229,-364\"/>\n<text text-anchor=\"middle\" x=\"262\" y=\"-819.4\" font-family=\"Times,serif\" font-size=\"12.00\">Sequential</text>\n</g>\n<g id=\"clust3\" class=\"cluster\">\n<title>cluster_4</title>\n<polygon fill=\"none\" stroke=\"black\" stroke-dasharray=\"5,2\" points=\"8,-81 8,-348 224,-348 224,-81 8,-81\"/>\n<text text-anchor=\"middle\" x=\"57\" y=\"-334.4\" font-family=\"Times,serif\" font-size=\"12.00\">PropensityModel</text>\n</g>\n<g id=\"clust4\" class=\"cluster\">\n<title>cluster_5</title>\n<polygon fill=\"none\" stroke=\"black\" stroke-dasharray=\"5,2\" points=\"16,-89 16,-319 216,-319 216,-89 16,-89\"/>\n<text text-anchor=\"middle\" x=\"49\" y=\"-305.4\" font-family=\"Times,serif\" font-size=\"12.00\">Sequential</text>\n</g>\n<g id=\"clust5\" class=\"cluster\">\n<title>cluster_6</title>\n<polygon fill=\"none\" stroke=\"black\" stroke-dasharray=\"5,2\" points=\"232,-8 232,-348 454,-348 454,-8 232,-8\"/>\n<text text-anchor=\"middle\" x=\"277.5\" y=\"-334.4\" font-family=\"Times,serif\" font-size=\"12.00\">OutcomeModel</text>\n</g>\n<g id=\"clust6\" class=\"cluster\">\n<title>cluster_7</title>\n<polygon fill=\"none\" stroke=\"black\" stroke-dasharray=\"5,2\" points=\"240,-16 240,-319 446,-319 446,-16 240,-16\"/>\n<text text-anchor=\"middle\" x=\"273\" y=\"-305.4\" font-family=\"Times,serif\" font-size=\"12.00\">Sequential</text>\n</g>\n<g id=\"clust7\" class=\"cluster\">\n<title>cluster_8</title>\n<polygon fill=\"none\" stroke=\"black\" stroke-dasharray=\"5,2\" points=\"462,-8 462,-348 684,-348 684,-8 462,-8\"/>\n<text text-anchor=\"middle\" x=\"507.5\" y=\"-334.4\" font-family=\"Times,serif\" font-size=\"12.00\">OutcomeModel</text>\n</g>\n<g id=\"clust8\" class=\"cluster\">\n<title>cluster_9</title>\n<polygon fill=\"none\" stroke=\"black\" stroke-dasharray=\"5,2\" points=\"470,-16 470,-319 676,-319 676,-16 470,-16\"/>\n<text text-anchor=\"middle\" x=\"503\" y=\"-305.4\" font-family=\"Times,serif\" font-size=\"12.00\">Sequential</text>\n</g>\n<!-- 0 -->\n<g id=\"node1\" class=\"node\">\n<title>0</title>\n<polygon fill=\"lightyellow\" stroke=\"transparent\" points=\"420,-902 266,-902 266,-870 420,-870 420,-902\"/>\n<polygon fill=\"none\" stroke=\"black\" points=\"266,-870 266,-902 349,-902 349,-870 266,-870\"/>\n<text text-anchor=\"start\" x=\"271\" y=\"-889\" font-family=\"Linux libertine\" font-size=\"10.00\">input&#45;tensor</text>\n<text text-anchor=\"start\" x=\"286\" y=\"-878\" font-family=\"Linux libertine\" font-size=\"10.00\">depth:0</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"349,-870 349,-902 420,-902 420,-870 349,-870\"/>\n<text text-anchor=\"start\" x=\"354\" y=\"-883.5\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 20)</text>\n</g>\n<!-- 1 -->\n<g id=\"node2\" class=\"node\">\n<title>1</title>\n<polygon fill=\"#c1ffc1\" stroke=\"transparent\" points=\"437.5,-804 248.5,-804 248.5,-762 437.5,-762 437.5,-804\"/>\n<polygon fill=\"none\" stroke=\"black\" points=\"249,-762 249,-804 302,-804 302,-762 249,-762\"/>\n<text text-anchor=\"start\" x=\"257\" y=\"-786\" font-family=\"Linux libertine\" font-size=\"10.00\">Linear</text>\n<text text-anchor=\"start\" x=\"254\" y=\"-775\" font-family=\"Linux libertine\" font-size=\"10.00\">depth:3</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"302,-783 302,-804 361,-804 361,-783 302,-783\"/>\n<text text-anchor=\"start\" x=\"313\" y=\"-791\" font-family=\"Linux libertine\" font-size=\"10.00\">input:</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"361,-783 361,-804 438,-804 438,-783 361,-783\"/>\n<text text-anchor=\"start\" x=\"366\" y=\"-791\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 20) </text>\n<polygon fill=\"none\" stroke=\"black\" points=\"302,-762 302,-783 361,-783 361,-762 302,-762\"/>\n<text text-anchor=\"start\" x=\"307\" y=\"-770\" font-family=\"Linux libertine\" font-size=\"10.00\">output: </text>\n<polygon fill=\"none\" stroke=\"black\" points=\"361,-762 361,-783 438,-783 438,-762 361,-762\"/>\n<text text-anchor=\"start\" x=\"366\" y=\"-770\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 75) </text>\n</g>\n<!-- 0&#45;&gt;1 -->\n<g id=\"edge1\" class=\"edge\">\n<title>0&#45;&gt;1</title>\n<path fill=\"none\" stroke=\"black\" d=\"M343,-869.66C343,-854.99 343,-832.46 343,-814.19\"/>\n<polygon fill=\"black\" stroke=\"black\" points=\"346.5,-814.02 343,-804.02 339.5,-814.02 346.5,-814.02\"/>\n</g>\n<!-- 2 -->\n<g id=\"node3\" class=\"node\">\n<title>2</title>\n<polygon fill=\"#c1ffc1\" stroke=\"transparent\" points=\"437.5,-726 248.5,-726 248.5,-684 437.5,-684 437.5,-726\"/>\n<polygon fill=\"none\" stroke=\"black\" points=\"249,-684 249,-726 302,-726 302,-684 249,-684\"/>\n<text text-anchor=\"start\" x=\"266\" y=\"-708\" font-family=\"Linux libertine\" font-size=\"10.00\">ELU</text>\n<text text-anchor=\"start\" x=\"254\" y=\"-697\" font-family=\"Linux libertine\" font-size=\"10.00\">depth:3</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"302,-705 302,-726 361,-726 361,-705 302,-705\"/>\n<text text-anchor=\"start\" x=\"313\" y=\"-713\" font-family=\"Linux libertine\" font-size=\"10.00\">input:</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"361,-705 361,-726 438,-726 438,-705 361,-705\"/>\n<text text-anchor=\"start\" x=\"366\" y=\"-713\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 75) </text>\n<polygon fill=\"none\" stroke=\"black\" points=\"302,-684 302,-705 361,-705 361,-684 302,-684\"/>\n<text text-anchor=\"start\" x=\"307\" y=\"-692\" font-family=\"Linux libertine\" font-size=\"10.00\">output: </text>\n<polygon fill=\"none\" stroke=\"black\" points=\"361,-684 361,-705 438,-705 438,-684 361,-684\"/>\n<text text-anchor=\"start\" x=\"366\" y=\"-692\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 75) </text>\n</g>\n<!-- 1&#45;&gt;2 -->\n<g id=\"edge2\" class=\"edge\">\n<title>1&#45;&gt;2</title>\n<path fill=\"none\" stroke=\"black\" d=\"M343,-761.63C343,-753.82 343,-744.73 343,-736.18\"/>\n<polygon fill=\"black\" stroke=\"black\" points=\"346.5,-736.16 343,-726.16 339.5,-736.16 346.5,-736.16\"/>\n</g>\n<!-- 3 -->\n<g id=\"node4\" class=\"node\">\n<title>3</title>\n<polygon fill=\"#c1ffc1\" stroke=\"transparent\" points=\"449.5,-648 236.5,-648 236.5,-606 449.5,-606 449.5,-648\"/>\n<polygon fill=\"none\" stroke=\"black\" points=\"237,-606 237,-648 314,-648 314,-606 237,-606\"/>\n<text text-anchor=\"start\" x=\"242\" y=\"-630\" font-family=\"Linux libertine\" font-size=\"10.00\">BatchNorm1d</text>\n<text text-anchor=\"start\" x=\"254\" y=\"-619\" font-family=\"Linux libertine\" font-size=\"10.00\">depth:3</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"314,-627 314,-648 373,-648 373,-627 314,-627\"/>\n<text text-anchor=\"start\" x=\"325\" y=\"-635\" font-family=\"Linux libertine\" font-size=\"10.00\">input:</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"373,-627 373,-648 450,-648 450,-627 373,-627\"/>\n<text text-anchor=\"start\" x=\"378\" y=\"-635\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 75) </text>\n<polygon fill=\"none\" stroke=\"black\" points=\"314,-606 314,-627 373,-627 373,-606 314,-606\"/>\n<text text-anchor=\"start\" x=\"319\" y=\"-614\" font-family=\"Linux libertine\" font-size=\"10.00\">output: </text>\n<polygon fill=\"none\" stroke=\"black\" points=\"373,-606 373,-627 450,-627 450,-606 373,-606\"/>\n<text text-anchor=\"start\" x=\"378\" y=\"-614\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 75) </text>\n</g>\n<!-- 2&#45;&gt;3 -->\n<g id=\"edge3\" class=\"edge\">\n<title>2&#45;&gt;3</title>\n<path fill=\"none\" stroke=\"black\" d=\"M343,-683.63C343,-675.82 343,-666.73 343,-658.18\"/>\n<polygon fill=\"black\" stroke=\"black\" points=\"346.5,-658.16 343,-648.16 339.5,-658.16 346.5,-658.16\"/>\n</g>\n<!-- 4 -->\n<g id=\"node5\" class=\"node\">\n<title>4</title>\n<polygon fill=\"#c1ffc1\" stroke=\"transparent\" points=\"437.5,-570 248.5,-570 248.5,-528 437.5,-528 437.5,-570\"/>\n<polygon fill=\"none\" stroke=\"black\" points=\"249,-528 249,-570 302,-570 302,-528 249,-528\"/>\n<text text-anchor=\"start\" x=\"257\" y=\"-552\" font-family=\"Linux libertine\" font-size=\"10.00\">Linear</text>\n<text text-anchor=\"start\" x=\"254\" y=\"-541\" font-family=\"Linux libertine\" font-size=\"10.00\">depth:3</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"302,-549 302,-570 361,-570 361,-549 302,-549\"/>\n<text text-anchor=\"start\" x=\"313\" y=\"-557\" font-family=\"Linux libertine\" font-size=\"10.00\">input:</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"361,-549 361,-570 438,-570 438,-549 361,-549\"/>\n<text text-anchor=\"start\" x=\"366\" y=\"-557\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 75) </text>\n<polygon fill=\"none\" stroke=\"black\" points=\"302,-528 302,-549 361,-549 361,-528 302,-528\"/>\n<text text-anchor=\"start\" x=\"307\" y=\"-536\" font-family=\"Linux libertine\" font-size=\"10.00\">output: </text>\n<polygon fill=\"none\" stroke=\"black\" points=\"361,-528 361,-549 438,-549 438,-528 361,-528\"/>\n<text text-anchor=\"start\" x=\"369\" y=\"-536\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 1) </text>\n</g>\n<!-- 3&#45;&gt;4 -->\n<g id=\"edge4\" class=\"edge\">\n<title>3&#45;&gt;4</title>\n<path fill=\"none\" stroke=\"black\" d=\"M343,-605.63C343,-597.82 343,-588.73 343,-580.18\"/>\n<polygon fill=\"black\" stroke=\"black\" points=\"346.5,-580.16 343,-570.16 339.5,-580.16 346.5,-580.16\"/>\n</g>\n<!-- 5 -->\n<g id=\"node6\" class=\"node\">\n<title>5</title>\n<polygon fill=\"#c1ffc1\" stroke=\"transparent\" points=\"434.5,-492 251.5,-492 251.5,-450 434.5,-450 434.5,-492\"/>\n<polygon fill=\"none\" stroke=\"black\" points=\"252,-450 252,-492 305,-492 305,-450 252,-450\"/>\n<text text-anchor=\"start\" x=\"269\" y=\"-474\" font-family=\"Linux libertine\" font-size=\"10.00\">ELU</text>\n<text text-anchor=\"start\" x=\"257\" y=\"-463\" font-family=\"Linux libertine\" font-size=\"10.00\">depth:3</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"305,-471 305,-492 364,-492 364,-471 305,-471\"/>\n<text text-anchor=\"start\" x=\"316\" y=\"-479\" font-family=\"Linux libertine\" font-size=\"10.00\">input:</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"364,-471 364,-492 435,-492 435,-471 364,-471\"/>\n<text text-anchor=\"start\" x=\"369\" y=\"-479\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 1) </text>\n<polygon fill=\"none\" stroke=\"black\" points=\"305,-450 305,-471 364,-471 364,-450 305,-450\"/>\n<text text-anchor=\"start\" x=\"310\" y=\"-458\" font-family=\"Linux libertine\" font-size=\"10.00\">output: </text>\n<polygon fill=\"none\" stroke=\"black\" points=\"364,-450 364,-471 435,-471 435,-450 364,-450\"/>\n<text text-anchor=\"start\" x=\"369\" y=\"-458\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 1) </text>\n</g>\n<!-- 4&#45;&gt;5 -->\n<g id=\"edge5\" class=\"edge\">\n<title>4&#45;&gt;5</title>\n<path fill=\"none\" stroke=\"black\" d=\"M343,-527.63C343,-519.82 343,-510.73 343,-502.18\"/>\n<polygon fill=\"black\" stroke=\"black\" points=\"346.5,-502.16 343,-492.16 339.5,-502.16 346.5,-502.16\"/>\n</g>\n<!-- 6 -->\n<g id=\"node7\" class=\"node\">\n<title>6</title>\n<polygon fill=\"#c1ffc1\" stroke=\"transparent\" points=\"446.5,-414 239.5,-414 239.5,-372 446.5,-372 446.5,-414\"/>\n<polygon fill=\"none\" stroke=\"black\" points=\"240,-372 240,-414 317,-414 317,-372 240,-372\"/>\n<text text-anchor=\"start\" x=\"245\" y=\"-396\" font-family=\"Linux libertine\" font-size=\"10.00\">BatchNorm1d</text>\n<text text-anchor=\"start\" x=\"257\" y=\"-385\" font-family=\"Linux libertine\" font-size=\"10.00\">depth:3</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"317,-393 317,-414 376,-414 376,-393 317,-393\"/>\n<text text-anchor=\"start\" x=\"328\" y=\"-401\" font-family=\"Linux libertine\" font-size=\"10.00\">input:</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"376,-393 376,-414 447,-414 447,-393 376,-393\"/>\n<text text-anchor=\"start\" x=\"381\" y=\"-401\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 1) </text>\n<polygon fill=\"none\" stroke=\"black\" points=\"317,-372 317,-393 376,-393 376,-372 317,-372\"/>\n<text text-anchor=\"start\" x=\"322\" y=\"-380\" font-family=\"Linux libertine\" font-size=\"10.00\">output: </text>\n<polygon fill=\"none\" stroke=\"black\" points=\"376,-372 376,-393 447,-393 447,-372 376,-372\"/>\n<text text-anchor=\"start\" x=\"381\" y=\"-380\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 1) </text>\n</g>\n<!-- 5&#45;&gt;6 -->\n<g id=\"edge6\" class=\"edge\">\n<title>5&#45;&gt;6</title>\n<path fill=\"none\" stroke=\"black\" d=\"M343,-449.63C343,-441.82 343,-432.73 343,-424.18\"/>\n<polygon fill=\"black\" stroke=\"black\" points=\"346.5,-424.16 343,-414.16 339.5,-424.16 346.5,-424.16\"/>\n</g>\n<!-- 7 -->\n<g id=\"node8\" class=\"node\">\n<title>7</title>\n<polygon fill=\"#c1ffc1\" stroke=\"transparent\" points=\"207.5,-290 24.5,-290 24.5,-248 207.5,-248 207.5,-290\"/>\n<polygon fill=\"none\" stroke=\"black\" points=\"25,-248 25,-290 78,-290 78,-248 25,-248\"/>\n<text text-anchor=\"start\" x=\"33\" y=\"-272\" font-family=\"Linux libertine\" font-size=\"10.00\">Linear</text>\n<text text-anchor=\"start\" x=\"30\" y=\"-261\" font-family=\"Linux libertine\" font-size=\"10.00\">depth:3</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"78,-269 78,-290 137,-290 137,-269 78,-269\"/>\n<text text-anchor=\"start\" x=\"89\" y=\"-277\" font-family=\"Linux libertine\" font-size=\"10.00\">input:</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"137,-269 137,-290 208,-290 208,-269 137,-269\"/>\n<text text-anchor=\"start\" x=\"142\" y=\"-277\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 1) </text>\n<polygon fill=\"none\" stroke=\"black\" points=\"78,-248 78,-269 137,-269 137,-248 78,-248\"/>\n<text text-anchor=\"start\" x=\"83\" y=\"-256\" font-family=\"Linux libertine\" font-size=\"10.00\">output: </text>\n<polygon fill=\"none\" stroke=\"black\" points=\"137,-248 137,-269 208,-269 208,-248 137,-248\"/>\n<text text-anchor=\"start\" x=\"142\" y=\"-256\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 1) </text>\n</g>\n<!-- 6&#45;&gt;7 -->\n<g id=\"edge7\" class=\"edge\">\n<title>6&#45;&gt;7</title>\n<path fill=\"none\" stroke=\"black\" d=\"M288.77,-371.98C271.59,-365.04 252.72,-356.77 236,-348 206.61,-332.58 175.21,-312.01 151.91,-295.87\"/>\n<polygon fill=\"black\" stroke=\"black\" points=\"153.91,-292.99 143.71,-290.14 149.9,-298.73 153.91,-292.99\"/>\n</g>\n<!-- 10 -->\n<g id=\"node11\" class=\"node\">\n<title>10</title>\n<polygon fill=\"#c1ffc1\" stroke=\"transparent\" points=\"437.5,-290 248.5,-290 248.5,-248 437.5,-248 437.5,-290\"/>\n<polygon fill=\"none\" stroke=\"black\" points=\"249,-248 249,-290 302,-290 302,-248 249,-248\"/>\n<text text-anchor=\"start\" x=\"257\" y=\"-272\" font-family=\"Linux libertine\" font-size=\"10.00\">Linear</text>\n<text text-anchor=\"start\" x=\"254\" y=\"-261\" font-family=\"Linux libertine\" font-size=\"10.00\">depth:3</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"302,-269 302,-290 361,-290 361,-269 302,-269\"/>\n<text text-anchor=\"start\" x=\"313\" y=\"-277\" font-family=\"Linux libertine\" font-size=\"10.00\">input:</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"361,-269 361,-290 438,-290 438,-269 361,-269\"/>\n<text text-anchor=\"start\" x=\"369\" y=\"-277\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 1) </text>\n<polygon fill=\"none\" stroke=\"black\" points=\"302,-248 302,-269 361,-269 361,-248 302,-248\"/>\n<text text-anchor=\"start\" x=\"307\" y=\"-256\" font-family=\"Linux libertine\" font-size=\"10.00\">output: </text>\n<polygon fill=\"none\" stroke=\"black\" points=\"361,-248 361,-269 438,-269 438,-248 361,-248\"/>\n<text text-anchor=\"start\" x=\"366\" y=\"-256\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 75) </text>\n</g>\n<!-- 6&#45;&gt;10 -->\n<g id=\"edge10\" class=\"edge\">\n<title>6&#45;&gt;10</title>\n<path fill=\"none\" stroke=\"black\" d=\"M343,-371.85C343,-352.43 343,-322.77 343,-300.46\"/>\n<polygon fill=\"black\" stroke=\"black\" points=\"346.5,-300.18 343,-290.18 339.5,-300.18 346.5,-300.18\"/>\n</g>\n<!-- 14 -->\n<g id=\"node15\" class=\"node\">\n<title>14</title>\n<polygon fill=\"#c1ffc1\" stroke=\"transparent\" points=\"667.5,-290 478.5,-290 478.5,-248 667.5,-248 667.5,-290\"/>\n<polygon fill=\"none\" stroke=\"black\" points=\"479,-248 479,-290 532,-290 532,-248 479,-248\"/>\n<text text-anchor=\"start\" x=\"487\" y=\"-272\" font-family=\"Linux libertine\" font-size=\"10.00\">Linear</text>\n<text text-anchor=\"start\" x=\"484\" y=\"-261\" font-family=\"Linux libertine\" font-size=\"10.00\">depth:3</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"532,-269 532,-290 591,-290 591,-269 532,-269\"/>\n<text text-anchor=\"start\" x=\"543\" y=\"-277\" font-family=\"Linux libertine\" font-size=\"10.00\">input:</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"591,-269 591,-290 668,-290 668,-269 591,-269\"/>\n<text text-anchor=\"start\" x=\"599\" y=\"-277\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 1) </text>\n<polygon fill=\"none\" stroke=\"black\" points=\"532,-248 532,-269 591,-269 591,-248 532,-248\"/>\n<text text-anchor=\"start\" x=\"537\" y=\"-256\" font-family=\"Linux libertine\" font-size=\"10.00\">output: </text>\n<polygon fill=\"none\" stroke=\"black\" points=\"591,-248 591,-269 668,-269 668,-248 591,-248\"/>\n<text text-anchor=\"start\" x=\"596\" y=\"-256\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 75) </text>\n</g>\n<!-- 6&#45;&gt;14 -->\n<g id=\"edge14\" class=\"edge\">\n<title>6&#45;&gt;14</title>\n<path fill=\"none\" stroke=\"black\" d=\"M396.9,-371.98C414.15,-365 433.14,-356.73 450,-348 480,-332.47 512.19,-311.91 536.11,-295.79\"/>\n<polygon fill=\"black\" stroke=\"black\" points=\"538.22,-298.59 544.52,-290.08 534.29,-292.8 538.22,-298.59\"/>\n</g>\n<!-- 8 -->\n<g id=\"node9\" class=\"node\">\n<title>8</title>\n<polygon fill=\"#c1ffc1\" stroke=\"transparent\" points=\"207.5,-212 24.5,-212 24.5,-170 207.5,-170 207.5,-212\"/>\n<polygon fill=\"none\" stroke=\"black\" points=\"25,-170 25,-212 78,-212 78,-170 25,-170\"/>\n<text text-anchor=\"start\" x=\"30\" y=\"-194\" font-family=\"Linux libertine\" font-size=\"10.00\">Sigmoid</text>\n<text text-anchor=\"start\" x=\"30\" y=\"-183\" font-family=\"Linux libertine\" font-size=\"10.00\">depth:3</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"78,-191 78,-212 137,-212 137,-191 78,-191\"/>\n<text text-anchor=\"start\" x=\"89\" y=\"-199\" font-family=\"Linux libertine\" font-size=\"10.00\">input:</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"137,-191 137,-212 208,-212 208,-191 137,-191\"/>\n<text text-anchor=\"start\" x=\"142\" y=\"-199\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 1) </text>\n<polygon fill=\"none\" stroke=\"black\" points=\"78,-170 78,-191 137,-191 137,-170 78,-170\"/>\n<text text-anchor=\"start\" x=\"83\" y=\"-178\" font-family=\"Linux libertine\" font-size=\"10.00\">output: </text>\n<polygon fill=\"none\" stroke=\"black\" points=\"137,-170 137,-191 208,-191 208,-170 137,-170\"/>\n<text text-anchor=\"start\" x=\"142\" y=\"-178\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 1) </text>\n</g>\n<!-- 7&#45;&gt;8 -->\n<g id=\"edge8\" class=\"edge\">\n<title>7&#45;&gt;8</title>\n<path fill=\"none\" stroke=\"black\" d=\"M116,-247.63C116,-239.82 116,-230.73 116,-222.18\"/>\n<polygon fill=\"black\" stroke=\"black\" points=\"119.5,-222.16 116,-212.16 112.5,-222.16 119.5,-222.16\"/>\n</g>\n<!-- 9 -->\n<g id=\"node10\" class=\"node\">\n<title>9</title>\n<polygon fill=\"lightyellow\" stroke=\"transparent\" points=\"193,-129 39,-129 39,-97 193,-97 193,-129\"/>\n<polygon fill=\"none\" stroke=\"black\" points=\"39,-97 39,-129 128,-129 128,-97 39,-97\"/>\n<text text-anchor=\"start\" x=\"44\" y=\"-116\" font-family=\"Linux libertine\" font-size=\"10.00\">output&#45;tensor</text>\n<text text-anchor=\"start\" x=\"62\" y=\"-105\" font-family=\"Linux libertine\" font-size=\"10.00\">depth:0</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"128,-97 128,-129 193,-129 193,-97 128,-97\"/>\n<text text-anchor=\"start\" x=\"133\" y=\"-110.5\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 1)</text>\n</g>\n<!-- 8&#45;&gt;9 -->\n<g id=\"edge9\" class=\"edge\">\n<title>8&#45;&gt;9</title>\n<path fill=\"none\" stroke=\"black\" d=\"M116,-169.63C116,-160.33 116,-149.22 116,-139.39\"/>\n<polygon fill=\"black\" stroke=\"black\" points=\"119.5,-139.36 116,-129.36 112.5,-139.36 119.5,-139.36\"/>\n</g>\n<!-- 11 -->\n<g id=\"node12\" class=\"node\">\n<title>11</title>\n<polygon fill=\"#c1ffc1\" stroke=\"transparent\" points=\"437.5,-212 248.5,-212 248.5,-170 437.5,-170 437.5,-212\"/>\n<polygon fill=\"none\" stroke=\"black\" points=\"249,-170 249,-212 302,-212 302,-170 249,-170\"/>\n<text text-anchor=\"start\" x=\"266\" y=\"-194\" font-family=\"Linux libertine\" font-size=\"10.00\">ELU</text>\n<text text-anchor=\"start\" x=\"254\" y=\"-183\" font-family=\"Linux libertine\" font-size=\"10.00\">depth:3</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"302,-191 302,-212 361,-212 361,-191 302,-191\"/>\n<text text-anchor=\"start\" x=\"313\" y=\"-199\" font-family=\"Linux libertine\" font-size=\"10.00\">input:</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"361,-191 361,-212 438,-212 438,-191 361,-191\"/>\n<text text-anchor=\"start\" x=\"366\" y=\"-199\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 75) </text>\n<polygon fill=\"none\" stroke=\"black\" points=\"302,-170 302,-191 361,-191 361,-170 302,-170\"/>\n<text text-anchor=\"start\" x=\"307\" y=\"-178\" font-family=\"Linux libertine\" font-size=\"10.00\">output: </text>\n<polygon fill=\"none\" stroke=\"black\" points=\"361,-170 361,-191 438,-191 438,-170 361,-170\"/>\n<text text-anchor=\"start\" x=\"366\" y=\"-178\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 75) </text>\n</g>\n<!-- 10&#45;&gt;11 -->\n<g id=\"edge11\" class=\"edge\">\n<title>10&#45;&gt;11</title>\n<path fill=\"none\" stroke=\"black\" d=\"M343,-247.63C343,-239.82 343,-230.73 343,-222.18\"/>\n<polygon fill=\"black\" stroke=\"black\" points=\"346.5,-222.16 343,-212.16 339.5,-222.16 346.5,-222.16\"/>\n</g>\n<!-- 12 -->\n<g id=\"node13\" class=\"node\">\n<title>12</title>\n<polygon fill=\"#c1ffc1\" stroke=\"transparent\" points=\"437.5,-134 248.5,-134 248.5,-92 437.5,-92 437.5,-134\"/>\n<polygon fill=\"none\" stroke=\"black\" points=\"249,-92 249,-134 302,-134 302,-92 249,-92\"/>\n<text text-anchor=\"start\" x=\"257\" y=\"-116\" font-family=\"Linux libertine\" font-size=\"10.00\">Linear</text>\n<text text-anchor=\"start\" x=\"254\" y=\"-105\" font-family=\"Linux libertine\" font-size=\"10.00\">depth:3</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"302,-113 302,-134 361,-134 361,-113 302,-113\"/>\n<text text-anchor=\"start\" x=\"313\" y=\"-121\" font-family=\"Linux libertine\" font-size=\"10.00\">input:</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"361,-113 361,-134 438,-134 438,-113 361,-113\"/>\n<text text-anchor=\"start\" x=\"366\" y=\"-121\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 75) </text>\n<polygon fill=\"none\" stroke=\"black\" points=\"302,-92 302,-113 361,-113 361,-92 302,-92\"/>\n<text text-anchor=\"start\" x=\"307\" y=\"-100\" font-family=\"Linux libertine\" font-size=\"10.00\">output: </text>\n<polygon fill=\"none\" stroke=\"black\" points=\"361,-92 361,-113 438,-113 438,-92 361,-92\"/>\n<text text-anchor=\"start\" x=\"369\" y=\"-100\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 1) </text>\n</g>\n<!-- 11&#45;&gt;12 -->\n<g id=\"edge12\" class=\"edge\">\n<title>11&#45;&gt;12</title>\n<path fill=\"none\" stroke=\"black\" d=\"M343,-169.63C343,-161.82 343,-152.73 343,-144.18\"/>\n<polygon fill=\"black\" stroke=\"black\" points=\"346.5,-144.16 343,-134.16 339.5,-144.16 346.5,-144.16\"/>\n</g>\n<!-- 13 -->\n<g id=\"node14\" class=\"node\">\n<title>13</title>\n<polygon fill=\"lightyellow\" stroke=\"transparent\" points=\"420,-56 266,-56 266,-24 420,-24 420,-56\"/>\n<polygon fill=\"none\" stroke=\"black\" points=\"266,-24 266,-56 355,-56 355,-24 266,-24\"/>\n<text text-anchor=\"start\" x=\"271\" y=\"-43\" font-family=\"Linux libertine\" font-size=\"10.00\">output&#45;tensor</text>\n<text text-anchor=\"start\" x=\"289\" y=\"-32\" font-family=\"Linux libertine\" font-size=\"10.00\">depth:0</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"355,-24 355,-56 420,-56 420,-24 355,-24\"/>\n<text text-anchor=\"start\" x=\"360\" y=\"-37.5\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 1)</text>\n</g>\n<!-- 12&#45;&gt;13 -->\n<g id=\"edge13\" class=\"edge\">\n<title>12&#45;&gt;13</title>\n<path fill=\"none\" stroke=\"black\" d=\"M343,-91.84C343,-83.89 343,-74.66 343,-66.26\"/>\n<polygon fill=\"black\" stroke=\"black\" points=\"346.5,-66.24 343,-56.24 339.5,-66.24 346.5,-66.24\"/>\n</g>\n<!-- 15 -->\n<g id=\"node16\" class=\"node\">\n<title>15</title>\n<polygon fill=\"#c1ffc1\" stroke=\"transparent\" points=\"667.5,-212 478.5,-212 478.5,-170 667.5,-170 667.5,-212\"/>\n<polygon fill=\"none\" stroke=\"black\" points=\"479,-170 479,-212 532,-212 532,-170 479,-170\"/>\n<text text-anchor=\"start\" x=\"496\" y=\"-194\" font-family=\"Linux libertine\" font-size=\"10.00\">ELU</text>\n<text text-anchor=\"start\" x=\"484\" y=\"-183\" font-family=\"Linux libertine\" font-size=\"10.00\">depth:3</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"532,-191 532,-212 591,-212 591,-191 532,-191\"/>\n<text text-anchor=\"start\" x=\"543\" y=\"-199\" font-family=\"Linux libertine\" font-size=\"10.00\">input:</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"591,-191 591,-212 668,-212 668,-191 591,-191\"/>\n<text text-anchor=\"start\" x=\"596\" y=\"-199\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 75) </text>\n<polygon fill=\"none\" stroke=\"black\" points=\"532,-170 532,-191 591,-191 591,-170 532,-170\"/>\n<text text-anchor=\"start\" x=\"537\" y=\"-178\" font-family=\"Linux libertine\" font-size=\"10.00\">output: </text>\n<polygon fill=\"none\" stroke=\"black\" points=\"591,-170 591,-191 668,-191 668,-170 591,-170\"/>\n<text text-anchor=\"start\" x=\"596\" y=\"-178\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 75) </text>\n</g>\n<!-- 14&#45;&gt;15 -->\n<g id=\"edge15\" class=\"edge\">\n<title>14&#45;&gt;15</title>\n<path fill=\"none\" stroke=\"black\" d=\"M573,-247.63C573,-239.82 573,-230.73 573,-222.18\"/>\n<polygon fill=\"black\" stroke=\"black\" points=\"576.5,-222.16 573,-212.16 569.5,-222.16 576.5,-222.16\"/>\n</g>\n<!-- 16 -->\n<g id=\"node17\" class=\"node\">\n<title>16</title>\n<polygon fill=\"#c1ffc1\" stroke=\"transparent\" points=\"667.5,-134 478.5,-134 478.5,-92 667.5,-92 667.5,-134\"/>\n<polygon fill=\"none\" stroke=\"black\" points=\"479,-92 479,-134 532,-134 532,-92 479,-92\"/>\n<text text-anchor=\"start\" x=\"487\" y=\"-116\" font-family=\"Linux libertine\" font-size=\"10.00\">Linear</text>\n<text text-anchor=\"start\" x=\"484\" y=\"-105\" font-family=\"Linux libertine\" font-size=\"10.00\">depth:3</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"532,-113 532,-134 591,-134 591,-113 532,-113\"/>\n<text text-anchor=\"start\" x=\"543\" y=\"-121\" font-family=\"Linux libertine\" font-size=\"10.00\">input:</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"591,-113 591,-134 668,-134 668,-113 591,-113\"/>\n<text text-anchor=\"start\" x=\"596\" y=\"-121\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 75) </text>\n<polygon fill=\"none\" stroke=\"black\" points=\"532,-92 532,-113 591,-113 591,-92 532,-92\"/>\n<text text-anchor=\"start\" x=\"537\" y=\"-100\" font-family=\"Linux libertine\" font-size=\"10.00\">output: </text>\n<polygon fill=\"none\" stroke=\"black\" points=\"591,-92 591,-113 668,-113 668,-92 591,-92\"/>\n<text text-anchor=\"start\" x=\"599\" y=\"-100\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 1) </text>\n</g>\n<!-- 15&#45;&gt;16 -->\n<g id=\"edge16\" class=\"edge\">\n<title>15&#45;&gt;16</title>\n<path fill=\"none\" stroke=\"black\" d=\"M573,-169.63C573,-161.82 573,-152.73 573,-144.18\"/>\n<polygon fill=\"black\" stroke=\"black\" points=\"576.5,-144.16 573,-134.16 569.5,-144.16 576.5,-144.16\"/>\n</g>\n<!-- 17 -->\n<g id=\"node18\" class=\"node\">\n<title>17</title>\n<polygon fill=\"lightyellow\" stroke=\"transparent\" points=\"650,-56 496,-56 496,-24 650,-24 650,-56\"/>\n<polygon fill=\"none\" stroke=\"black\" points=\"496,-24 496,-56 585,-56 585,-24 496,-24\"/>\n<text text-anchor=\"start\" x=\"501\" y=\"-43\" font-family=\"Linux libertine\" font-size=\"10.00\">output&#45;tensor</text>\n<text text-anchor=\"start\" x=\"519\" y=\"-32\" font-family=\"Linux libertine\" font-size=\"10.00\">depth:0</text>\n<polygon fill=\"none\" stroke=\"black\" points=\"585,-24 585,-56 650,-56 650,-24 585,-24\"/>\n<text text-anchor=\"start\" x=\"590\" y=\"-37.5\" font-family=\"Linux libertine\" font-size=\"10.00\">(8000, 1)</text>\n</g>\n<!-- 16&#45;&gt;17 -->\n<g id=\"edge17\" class=\"edge\">\n<title>16&#45;&gt;17</title>\n<path fill=\"none\" stroke=\"black\" d=\"M573,-91.84C573,-83.89 573,-74.66 573,-66.26\"/>\n<polygon fill=\"black\" stroke=\"black\" points=\"576.5,-66.24 573,-56.24 569.5,-66.24 576.5,-66.24\"/>\n</g>\n</g>\n</svg>\n",
            "text/plain": [
              "<graphviz.graphs.Digraph at 0x7fbafdf57460>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Training loop\n",
        "n_epochs = int(10e3)\n",
        "hist = []\n",
        "best_epoch = 0\n",
        "best_loss = float('inf')\n",
        "iterator = trange(n_epochs)\n",
        "for epoch in iterator:\n",
        "    # Forward pass\n",
        "    g_hat, qt_hat, qc_hat = causal_model(x_train)\n",
        "    loss_i = loss_func(g_hat, is_treatment_train, qt_hat, qc_hat, y_train)\n",
        "\n",
        "    # Backward pass\n",
        "    optimizer.zero_grad()\n",
        "    loss_i.backward()\n",
        "    optimizer.step()\n",
        "\n",
        "    # Evaluate\n",
        "    with torch.inference_mode():\n",
        "        g_hat_test, qt_hat_test, qc_hat_test = causal_model(x_test)\n",
        "        hist.append((\n",
        "            loss_func(\n",
        "                g_hat_test,\n",
        "                is_treatment_test,\n",
        "                qt_hat_test,\n",
        "                qc_hat_test,\n",
        "                y_test\n",
        "            ),\n",
        "            torch.mean(causal_model.ite(x_test))\n",
        "        ))\n",
        "        iterator.set_description(f'i: {epoch:,} | loss: {hist[-1][0].numpy():.8f} | CATE: {hist[-1][1].numpy():.3f}')\n",
        "        if hist[-1][0] < best_loss:\n",
        "            best_loss = hist[-1][0]\n",
        "            best_epoch = epoch\n",
        "        elif (epoch - best_epoch) > 10:\n",
        "            print(f'Early stopping at epoch {epoch:,}')\n",
        "            break"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 66,
          "referenced_widgets": [
            "21f580a934244ab7a7dc81f7c3a96056",
            "50447b13c0c74893827782b8f43b08ac",
            "aac9cd8d6dc34800978158a61142cc12",
            "eb7514eaef82491b876c3428398d36e8",
            "0ed15cb565a04b2cb2e623396d188215",
            "69f92189a487448aa767238307afa650",
            "5118328b79b547fba4494e6c27a11653",
            "53d3bbac346340b28da726ea2fa627e5",
            "f56deae597074424a12415307fa643b1",
            "852fb823cba84da8b030b089deeecc5c",
            "e5427b39c9f8478d988d0be4a8dbd40f"
          ]
        },
        "id": "AyMt9_BmOgBQ",
        "outputId": "dc5d2415-087a-4297-c866-64d0c04df60f"
      },
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "  0%|          | 0/10000 [00:00<?, ?it/s]"
            ],
            "application/vnd.jupyter.widget-view+json": {
              "version_major": 2,
              "version_minor": 0,
              "model_id": "21f580a934244ab7a7dc81f7c3a96056"
            }
          },
          "metadata": {}
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Early stopping at epoch 551\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "graph.plot(np.array([t[0].numpy() for t in hist]), label='Causal Model')\n",
        "# graph.yscale('log')\n",
        "graph.ylabel('Causal Loss')\n",
        "graph.legend()\n",
        "graph.show()\n",
        "\n",
        "graph.plot(np.array([t[1].numpy() for t in hist]), label='Causal Model')\n",
        "graph.axhline(treatment_effect, linestyle='--', color='black', lw=1.5, label='True Treatment Effect')\n",
        "graph.axhline(naive_causal_estimate, linestyle='--', color='red', lw=1.5, label='Naive Treatment Effect')\n",
        "graph.ylabel('CATE')\n",
        "graph.legend()\n",
        "graph.show()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 881
        },
        "id": "0yu1ZRaQQ2ba",
        "outputId": "69e54e95-1070-4844-f2b5-5f7c3c398a17"
      },
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "_Stephen Anthony Rose_"
      ],
      "metadata": {
        "id": "o5QUJhbyZAW5"
      }
    }
  ]
}