{

  "nbformat": 4,

  "nbformat_minor": 0,

  "metadata": {

    "colab": {

      "provenance": [],

      "machine_shape": "hm"

    },

    "kernelspec": {

      "name": "python3",

      "display_name": "Python 3"

    }

  },

  "cells": [

    {

      "cell_type": "markdown",

      "metadata": {

        "id": "8XnVMPBXmtRa"

      },

      "source": [

        "# TensorNetworks in Neural Networks.\n",

        "\n",

        "Here, we have a small toy example of how to use a TN inside of a fully connected neural network.\n",

        "\n",

        "First off, let's install tensornetwork"

      ]

    },

    {

      "cell_type": "code",

      "metadata": {

        "id": "7HGRsYNAFxME"

      },

      "source": [

        "# !pip install tensornetwork\n",

        "\n",

        "import numpy as np\n",

        "import matplotlib.pyplot as plt\n",

        "import tensorflow as tf\n",

        "# Import tensornetwork\n",

        "import tensornetwork as tn\n",

        "import random\n",

        "import time\n",

        "# Set the backend to tesorflow\n",

        "# (default is numpy)\n",

        "tn.set_default_backend(\"tensorflow\")\n",

        "np.random.seed(42)\n",

        "random.seed(42)\n",

        "tf.random.set_seed(42)"

      ],

      "execution_count": 50,

      "outputs": []

    },

    {

      "cell_type": "markdown",

      "metadata": {

        "id": "g1OMCo5XmrYu"

      },

      "source": [

        "# TensorNetwork layer definition\n",

        "\n",

        "Here, we define the TensorNetwork layer we wish to use to replace the fully connected layer. Here, we simply use a 2 node Matrix Product Operator network to replace the normal dense weight matrix.\n",

        "\n",

        "We TensorNetwork's NCon API to keep the code short."

      ]

    },

    {

      "cell_type": "code",

      "metadata": {

        "id": "wvSMKtPufnLp"

      },

      "source": [

        "class TNLayer(tf.keras.layers.Layer):\n",

        "\n",

        "  def __init__(self):\n",

        "    super(TNLayer, self).__init__()\n",

        "    # Create the variables for the layer.\n",

        "    self.a_var = tf.Variable(tf.random.normal(shape=(32, 32, 2),\n",

        "                                              stddev=1.0/32.0),\n",

        "                             name=\"a\", trainable=True)\n",

        "    self.b_var = tf.Variable(tf.random.normal(shape=(32, 32, 2),\n",

        "                                              stddev=1.0/32.0),\n",

        "                             name=\"b\", trainable=True)\n",

        "    self.bias = tf.Variable(tf.zeros(shape=(32, 32)),\n",

        "                            name=\"bias\", trainable=True)\n",

        "\n",

        "  def call(self, inputs):\n",

        "    # Define the contraction.\n",

        "    # We break it out so we can parallelize a batch using\n",

        "    # tf.vectorized_map (see below).\n",

        "    def f(input_vec, a_var, b_var, bias_var):\n",

        "      # Reshape to a matrix instead of a vector.\n",

        "      input_vec = tf.reshape(input_vec, (32, 32))\n",

        "\n",

        "      # Now we create the network.\n",

        "      a = tn.Node(a_var)\n",

        "      b = tn.Node(b_var)\n",

        "      x_node = tn.Node(input_vec)\n",

        "      a[1] ^ x_node[0]\n",

        "      b[1] ^ x_node[1]\n",

        "      a[2] ^ b[2]\n",

        "\n",

        "      # The TN should now look like this\n",

        "      #   |     |\n",

        "      #   a --- b\n",

        "      #    \\   /\n",

        "      #      x\n",

        "\n",

        "      # Now we begin the contraction.\n",

        "      c = a @ x_node\n",

        "      result = (c @ b).tensor\n",

        "\n",

        "      # To make the code shorter, we also could've used Ncon.\n",

        "      # The above few lines of code is the same as this:\n",

        "      # result = tn.ncon([x, a_var, b_var], [[1, 2], [-1, 1, 3], [-2, 2, 3]])\n",

        "\n",

        "      # Finally, add bias.\n",

        "      return result + bias_var\n",

        "\n",

        "    # To deal with a batch of items, we can use the tf.vectorized_map\n",

        "    # function.\n",

        "    # https://www.tensorflow.org/api_docs/python/tf/vectorized_map\n",

        "    result = tf.vectorized_map(\n",

        "        lambda vec: f(vec, self.a_var, self.b_var, self.bias), inputs)\n",

        "    return tf.nn.relu(tf.reshape(result, (-1, 1024)))"

      ],

      "execution_count": 51,

      "outputs": []

    },

    {

      "cell_type": "markdown",

      "metadata": {

        "id": "V-CVqIhPnhY_"

      },

      "source": [

        "# Smaller model\n",

        "These two models are effectively the same, but notice how the TN layer has nearly 10x fewer parameters."

      ]

    },

    {

      "cell_type": "code",

      "metadata": {

        "id": "XPBvnB95jg4b",

        "outputId": "f6ab50e3-d97d-4d0f-b04b-a06c0e7ace08",

        "colab": {

          "base_uri": "https://localhost:8080/",

          "height": 0

        }

      },

      "source": [

        "Dense = tf.keras.layers.Dense\n",

        "fc_model = tf.keras.Sequential(\n",

        "    [\n",

        "     tf.keras.Input(shape=(2,)),\n",

        "     Dense(1024, activation=tf.nn.relu),\n",

        "     Dense(1024, activation=tf.nn.relu),\n",

        "     Dense(1, activation=None)])\n",

        "fc_model.summary()"

      ],

      "execution_count": 52,

      "outputs": [

        {

          "output_type": "stream",

          "name": "stdout",

          "text": [

            "Model: \"sequential_8\"\n",

            "_________________________________________________________________\n",

            " Layer (type)                Output Shape              Param #   \n",

            "=================================================================\n",

            " dense_20 (Dense)            (None, 1024)              3072      \n",

            "                                                                 \n",

            " dense_21 (Dense)            (None, 1024)              1049600   \n",

            "                                                                 \n",

            " dense_22 (Dense)            (None, 1)                 1025      \n",

            "                                                                 \n",

            "=================================================================\n",

            "Total params: 1053697 (4.02 MB)\n",

            "Trainable params: 1053697 (4.02 MB)\n",

            "Non-trainable params: 0 (0.00 Byte)\n",

            "_________________________________________________________________\n"

          ]

        }

      ]

    },

    {

      "cell_type": "code",

      "metadata": {

        "id": "bbKsmK8wIFTp",

        "outputId": "9114f08d-f095-457c-d26e-c50708c0210f",

        "colab": {

          "base_uri": "https://localhost:8080/",

          "height": 0

        }

      },

      "source": [

        "tn_model = tf.keras.Sequential(\n",

        "    [\n",

        "     tf.keras.Input(shape=(2,)),\n",

        "     Dense(1024, activation=tf.nn.relu),\n",

        "     # Here, we replace the dense layer with our MPS.\n",

        "     TNLayer(),\n",

        "     Dense(1, activation=None)])\n",

        "tn_model.summary()"

      ],

      "execution_count": 53,

      "outputs": [

        {

          "output_type": "stream",

          "name": "stdout",

          "text": [

            "Model: \"sequential_9\"\n",

            "_________________________________________________________________\n",

            " Layer (type)                Output Shape              Param #   \n",

            "=================================================================\n",

            " dense_23 (Dense)            (None, 1024)              3072      \n",

            "                                                                 \n",

            " tn_layer_10 (TNLayer)       (None, 1024)              5120      \n",

            "                                                                 \n",

            " dense_24 (Dense)            (None, 1)                 1025      \n",

            "                                                                 \n",

            "=================================================================\n",

            "Total params: 9217 (36.00 KB)\n",

            "Trainable params: 9217 (36.00 KB)\n",

            "Non-trainable params: 0 (0.00 Byte)\n",

            "_________________________________________________________________\n"

          ]

        }

      ]

    },

    {

      "cell_type": "markdown",

      "metadata": {

        "id": "GWwoYp0WnsLA"

      },

      "source": [

        "# Training a model\n",

        "\n",

        "You can train the TN model just as you would a normal neural network model! Here, we give an example of how to do it in Keras."

      ]

    },

    {

      "cell_type": "code",

      "metadata": {

        "id": "qDFzOC7sDBJ-"

      },

      "source": [

        "X = np.concatenate([np.random.randn(20, 2) + np.array([3, 3]),\n",

        "                    np.random.randn(20, 2) + np.array([-3, -3]),\n",

        "                    np.random.randn(20, 2) + np.array([-3, 3]),\n",

        "                    np.random.randn(20, 2) + np.array([3, -3])])\n",

        "\n",

        "Y = np.concatenate([np.ones((40)), -np.ones((40))])"

      ],

      "execution_count": 54,

      "outputs": []

    },

    {

      "cell_type": "code",

      "source": [

        "seconds = time.time()\n",

        "print(\"Time in seconds since beginning of run:\", seconds)\n",

        "local_time = time.ctime(seconds)\n",

        "print(local_time)"

      ],

      "metadata": {

        "colab": {

          "base_uri": "https://localhost:8080/",

          "height": 0

        },

        "id": "19TWP-1eKURB",

        "outputId": "21bed910-b9bc-4673-e150-4b4590ece06e"

      },

      "execution_count": 55,

      "outputs": [

        {

          "output_type": "stream",

          "name": "stdout",

          "text": [

            "Time in seconds since beginning of run: 1709531165.8950815\n",

            "Mon Mar  4 05:46:05 2024\n"

          ]

        }

      ]

    },

    {

      "cell_type": "code",

      "metadata": {

        "id": "crc0q1vbIyTj",

        "outputId": "a344d443-4472-4cd1-894f-9a8bbbcd219c",

        "colab": {

          "base_uri": "https://localhost:8080/",

          "height": 0

        }

      },

      "source": [

        "tn_model.compile(optimizer=\"adam\", loss=\"mean_squared_error\")\n",

        "tn_model.fit(X, Y, epochs=300, verbose=2)"

      ],

      "execution_count": 56,

      "outputs": [

        {

          "output_type": "stream",

          "name": "stdout",

          "text": [

            "Epoch 1/300\n",

            "3/3 - 1s - loss: 0.9824 - 989ms/epoch - 330ms/step\n",

            "Epoch 2/300\n",

            "3/3 - 0s - loss: 0.9310 - 12ms/epoch - 4ms/step\n",

            "Epoch 3/300\n",

            "3/3 - 0s - loss: 0.8843 - 11ms/epoch - 4ms/step\n",

            "Epoch 4/300\n",

            "3/3 - 0s - loss: 0.8317 - 10ms/epoch - 3ms/step\n",

            "Epoch 5/300\n",

            "3/3 - 0s - loss: 0.7705 - 10ms/epoch - 3ms/step\n",

            "Epoch 6/300\n",

            "3/3 - 0s - loss: 0.6941 - 11ms/epoch - 4ms/step\n",

            "Epoch 7/300\n",

            "3/3 - 0s - loss: 0.6033 - 12ms/epoch - 4ms/step\n",

            "Epoch 8/300\n",

            "3/3 - 0s - loss: 0.4972 - 12ms/epoch - 4ms/step\n",

            "Epoch 9/300\n",

            "3/3 - 0s - loss: 0.3789 - 11ms/epoch - 4ms/step\n",

            "Epoch 10/300\n",

            "3/3 - 0s - loss: 0.2568 - 11ms/epoch - 4ms/step\n",

            "Epoch 11/300\n",

            "3/3 - 0s - loss: 0.1478 - 11ms/epoch - 4ms/step\n",

            "Epoch 12/300\n",

            "3/3 - 0s - loss: 0.0716 - 10ms/epoch - 3ms/step\n",

            "Epoch 13/300\n",

            "3/3 - 0s - loss: 0.0546 - 10ms/epoch - 3ms/step\n",

            "Epoch 14/300\n",

            "3/3 - 0s - loss: 0.0772 - 10ms/epoch - 3ms/step\n",

            "Epoch 15/300\n",

            "3/3 - 0s - loss: 0.0885 - 10ms/epoch - 3ms/step\n",

            "Epoch 16/300\n",

            "3/3 - 0s - loss: 0.0685 - 11ms/epoch - 4ms/step\n",

            "Epoch 17/300\n",

            "3/3 - 0s - loss: 0.0493 - 11ms/epoch - 4ms/step\n",

            "Epoch 18/300\n",

            "3/3 - 0s - loss: 0.0414 - 10ms/epoch - 3ms/step\n",

            "Epoch 19/300\n",

            "3/3 - 0s - loss: 0.0453 - 9ms/epoch - 3ms/step\n",

            "Epoch 20/300\n",

            "3/3 - 0s - loss: 0.0485 - 10ms/epoch - 3ms/step\n",

            "Epoch 21/300\n",

            "3/3 - 0s - loss: 0.0463 - 11ms/epoch - 4ms/step\n",

            "Epoch 22/300\n",

            "3/3 - 0s - loss: 0.0407 - 10ms/epoch - 3ms/step\n",

            "Epoch 23/300\n",

            "3/3 - 0s - loss: 0.0373 - 11ms/epoch - 4ms/step\n",

            "Epoch 24/300\n",

            "3/3 - 0s - loss: 0.0360 - 10ms/epoch - 3ms/step\n",

            "Epoch 25/300\n",

            "3/3 - 0s - loss: 0.0361 - 12ms/epoch - 4ms/step\n",

            "Epoch 26/300\n",

            "3/3 - 0s - loss: 0.0354 - 10ms/epoch - 3ms/step\n",

            "Epoch 27/300\n",

            "3/3 - 0s - loss: 0.0337 - 11ms/epoch - 4ms/step\n",

            "Epoch 28/300\n",

            "3/3 - 0s - loss: 0.0323 - 10ms/epoch - 3ms/step\n",

            "Epoch 29/300\n",

            "3/3 - 0s - loss: 0.0310 - 10ms/epoch - 3ms/step\n",

            "Epoch 30/300\n",

            "3/3 - 0s - loss: 0.0310 - 10ms/epoch - 3ms/step\n",

            "Epoch 31/300\n",

            "3/3 - 0s - loss: 0.0302 - 12ms/epoch - 4ms/step\n",

            "Epoch 32/300\n",

            "3/3 - 0s - loss: 0.0287 - 10ms/epoch - 3ms/step\n",

            "Epoch 33/300\n",

            "3/3 - 0s - loss: 0.0281 - 10ms/epoch - 3ms/step\n",

            "Epoch 34/300\n",

            "3/3 - 0s - loss: 0.0275 - 11ms/epoch - 4ms/step\n",

            "Epoch 35/300\n",

            "3/3 - 0s - loss: 0.0264 - 10ms/epoch - 3ms/step\n",

            "Epoch 36/300\n",

            "3/3 - 0s - loss: 0.0256 - 10ms/epoch - 3ms/step\n",

            "Epoch 37/300\n",

            "3/3 - 0s - loss: 0.0250 - 11ms/epoch - 4ms/step\n",

            "Epoch 38/300\n",

            "3/3 - 0s - loss: 0.0245 - 10ms/epoch - 3ms/step\n",

            "Epoch 39/300\n",

            "3/3 - 0s - loss: 0.0235 - 11ms/epoch - 4ms/step\n",

            "Epoch 40/300\n",

            "3/3 - 0s - loss: 0.0228 - 10ms/epoch - 3ms/step\n",

            "Epoch 41/300\n",

            "3/3 - 0s - loss: 0.0220 - 10ms/epoch - 3ms/step\n",

            "Epoch 42/300\n",

            "3/3 - 0s - loss: 0.0213 - 10ms/epoch - 3ms/step\n",

            "Epoch 43/300\n",

            "3/3 - 0s - loss: 0.0205 - 11ms/epoch - 4ms/step\n",

            "Epoch 44/300\n",

            "3/3 - 0s - loss: 0.0199 - 10ms/epoch - 3ms/step\n",

            "Epoch 45/300\n",

            "3/3 - 0s - loss: 0.0191 - 10ms/epoch - 3ms/step\n",

            "Epoch 46/300\n",

            "3/3 - 0s - loss: 0.0187 - 10ms/epoch - 3ms/step\n",

            "Epoch 47/300\n",

            "3/3 - 0s - loss: 0.0181 - 10ms/epoch - 3ms/step\n",

            "Epoch 48/300\n",

            "3/3 - 0s - loss: 0.0175 - 11ms/epoch - 4ms/step\n",

            "Epoch 49/300\n",

            "3/3 - 0s - loss: 0.0164 - 10ms/epoch - 3ms/step\n",

            "Epoch 50/300\n",

            "3/3 - 0s - loss: 0.0157 - 10ms/epoch - 3ms/step\n",

            "Epoch 51/300\n",

            "3/3 - 0s - loss: 0.0152 - 10ms/epoch - 3ms/step\n",

            "Epoch 52/300\n",

            "3/3 - 0s - loss: 0.0145 - 10ms/epoch - 3ms/step\n",

            "Epoch 53/300\n",

            "3/3 - 0s - loss: 0.0138 - 12ms/epoch - 4ms/step\n",

            "Epoch 54/300\n",

            "3/3 - 0s - loss: 0.0133 - 9ms/epoch - 3ms/step\n",

            "Epoch 55/300\n",

            "3/3 - 0s - loss: 0.0125 - 10ms/epoch - 3ms/step\n",

            "Epoch 56/300\n",

            "3/3 - 0s - loss: 0.0122 - 10ms/epoch - 3ms/step\n",

            "Epoch 57/300\n",

            "3/3 - 0s - loss: 0.0113 - 10ms/epoch - 3ms/step\n",

            "Epoch 58/300\n",

            "3/3 - 0s - loss: 0.0108 - 10ms/epoch - 3ms/step\n",

            "Epoch 59/300\n",

            "3/3 - 0s - loss: 0.0102 - 9ms/epoch - 3ms/step\n",

            "Epoch 60/300\n",

            "3/3 - 0s - loss: 0.0096 - 10ms/epoch - 3ms/step\n",

            "Epoch 61/300\n",

            "3/3 - 0s - loss: 0.0090 - 10ms/epoch - 3ms/step\n",

            "Epoch 62/300\n",

            "3/3 - 0s - loss: 0.0084 - 10ms/epoch - 3ms/step\n",

            "Epoch 63/300\n",

            "3/3 - 0s - loss: 0.0077 - 9ms/epoch - 3ms/step\n",

            "Epoch 64/300\n",

            "3/3 - 0s - loss: 0.0070 - 10ms/epoch - 3ms/step\n",

            "Epoch 65/300\n",

            "3/3 - 0s - loss: 0.0070 - 10ms/epoch - 3ms/step\n",

            "Epoch 66/300\n",

            "3/3 - 0s - loss: 0.0062 - 10ms/epoch - 3ms/step\n",

            "Epoch 67/300\n",

            "3/3 - 0s - loss: 0.0057 - 10ms/epoch - 3ms/step\n",

            "Epoch 68/300\n",

            "3/3 - 0s - loss: 0.0054 - 9ms/epoch - 3ms/step\n",

            "Epoch 69/300\n",

            "3/3 - 0s - loss: 0.0050 - 10ms/epoch - 3ms/step\n",

            "Epoch 70/300\n",

            "3/3 - 0s - loss: 0.0046 - 11ms/epoch - 4ms/step\n",

            "Epoch 71/300\n",

            "3/3 - 0s - loss: 0.0041 - 10ms/epoch - 3ms/step\n",

            "Epoch 72/300\n",

            "3/3 - 0s - loss: 0.0037 - 9ms/epoch - 3ms/step\n",

            "Epoch 73/300\n",

            "3/3 - 0s - loss: 0.0033 - 10ms/epoch - 3ms/step\n",

            "Epoch 74/300\n",

            "3/3 - 0s - loss: 0.0030 - 11ms/epoch - 4ms/step\n",

            "Epoch 75/300\n",

            "3/3 - 0s - loss: 0.0027 - 11ms/epoch - 4ms/step\n",

            "Epoch 76/300\n",

            "3/3 - 0s - loss: 0.0024 - 11ms/epoch - 4ms/step\n",

            "Epoch 77/300\n",

            "3/3 - 0s - loss: 0.0021 - 11ms/epoch - 4ms/step\n",

            "Epoch 78/300\n",

            "3/3 - 0s - loss: 0.0019 - 10ms/epoch - 3ms/step\n",

            "Epoch 79/300\n",

            "3/3 - 0s - loss: 0.0016 - 10ms/epoch - 3ms/step\n",

            "Epoch 80/300\n",

            "3/3 - 0s - loss: 0.0015 - 10ms/epoch - 3ms/step\n",

            "Epoch 81/300\n",

            "3/3 - 0s - loss: 0.0013 - 11ms/epoch - 4ms/step\n",

            "Epoch 82/300\n",

            "3/3 - 0s - loss: 0.0012 - 11ms/epoch - 4ms/step\n",

            "Epoch 83/300\n",

            "3/3 - 0s - loss: 0.0011 - 10ms/epoch - 3ms/step\n",

            "Epoch 84/300\n",

            "3/3 - 0s - loss: 9.0545e-04 - 10ms/epoch - 3ms/step\n",

            "Epoch 85/300\n",

            "3/3 - 0s - loss: 7.9152e-04 - 10ms/epoch - 3ms/step\n",

            "Epoch 86/300\n",

            "3/3 - 0s - loss: 7.0412e-04 - 10ms/epoch - 3ms/step\n",

            "Epoch 87/300\n",

            "3/3 - 0s - loss: 6.1978e-04 - 10ms/epoch - 3ms/step\n",

            "Epoch 88/300\n",

            "3/3 - 0s - loss: 5.2338e-04 - 10ms/epoch - 3ms/step\n",

            "Epoch 89/300\n",

            "3/3 - 0s - loss: 5.0727e-04 - 10ms/epoch - 3ms/step\n",

            "Epoch 90/300\n",

            "3/3 - 0s - loss: 4.4617e-04 - 11ms/epoch - 4ms/step\n",

            "Epoch 91/300\n",

            "3/3 - 0s - loss: 4.1021e-04 - 11ms/epoch - 4ms/step\n",

            "Epoch 92/300\n",

            "3/3 - 0s - loss: 3.7321e-04 - 10ms/epoch - 3ms/step\n",

            "Epoch 93/300\n",

            "3/3 - 0s - loss: 3.3965e-04 - 11ms/epoch - 4ms/step\n",

            "Epoch 94/300\n",

            "3/3 - 0s - loss: 2.8217e-04 - 11ms/epoch - 4ms/step\n",

            "Epoch 95/300\n",

            "3/3 - 0s - loss: 2.8671e-04 - 10ms/epoch - 3ms/step\n",

            "Epoch 96/300\n",

            "3/3 - 0s - loss: 2.6762e-04 - 11ms/epoch - 4ms/step\n",

            "Epoch 97/300\n",

            "3/3 - 0s - loss: 2.3497e-04 - 10ms/epoch - 3ms/step\n",

            "Epoch 98/300\n",

            "3/3 - 0s - loss: 2.1839e-04 - 10ms/epoch - 3ms/step\n",

            "Epoch 99/300\n",

            "3/3 - 0s - loss: 1.9666e-04 - 11ms/epoch - 4ms/step\n",

            "Epoch 100/300\n",

            "3/3 - 0s - loss: 1.8738e-04 - 10ms/epoch - 3ms/step\n",

            "Epoch 101/300\n",

            "3/3 - 0s - loss: 1.5480e-04 - 10ms/epoch - 3ms/step\n",

            "Epoch 102/300\n",

            "3/3 - 0s - loss: 1.7105e-04 - 10ms/epoch - 3ms/step\n",

            "Epoch 103/300\n",

            "3/3 - 0s - loss: 1.4210e-04 - 11ms/epoch - 4ms/step\n",

            "Epoch 104/300\n",

            "3/3 - 0s - loss: 1.4973e-04 - 11ms/epoch - 4ms/step\n",

            "Epoch 105/300\n",

            "3/3 - 0s - loss: 1.3315e-04 - 12ms/epoch - 4ms/step\n",

            "Epoch 106/300\n",

            "3/3 - 0s - loss: 1.2269e-04 - 12ms/epoch - 4ms/step\n",

            "Epoch 107/300\n",

            "3/3 - 0s - loss: 1.2132e-04 - 11ms/epoch - 4ms/step\n",

            "Epoch 108/300\n",

            "3/3 - 0s - loss: 1.0276e-04 - 11ms/epoch - 4ms/step\n",

            "Epoch 109/300\n",

            "3/3 - 0s - loss: 1.0838e-04 - 10ms/epoch - 3ms/step\n",

            "Epoch 110/300\n",

            "3/3 - 0s - loss: 1.0152e-04 - 10ms/epoch - 3ms/step\n",

            "Epoch 111/300\n",

            "3/3 - 0s - loss: 9.3861e-05 - 10ms/epoch - 3ms/step\n",

            "Epoch 112/300\n",

            "3/3 - 0s - loss: 8.2561e-05 - 9ms/epoch - 3ms/step\n",

            "Epoch 113/300\n",

            "3/3 - 0s - loss: 8.2601e-05 - 11ms/epoch - 4ms/step\n",

            "Epoch 114/300\n",

            "3/3 - 0s - loss: 7.4005e-05 - 9ms/epoch - 3ms/step\n",

            "Epoch 115/300\n",

            "3/3 - 0s - loss: 7.7541e-05 - 10ms/epoch - 3ms/step\n",

            "Epoch 116/300\n",

            "3/3 - 0s - loss: 6.6006e-05 - 12ms/epoch - 4ms/step\n",

            "Epoch 117/300\n",

            "3/3 - 0s - loss: 6.4523e-05 - 10ms/epoch - 3ms/step\n",

            "Epoch 118/300\n",

            "3/3 - 0s - loss: 6.4468e-05 - 10ms/epoch - 3ms/step\n",

            "Epoch 119/300\n",

            "3/3 - 0s - loss: 6.0156e-05 - 10ms/epoch - 3ms/step\n",

            "Epoch 120/300\n",

            "3/3 - 0s - loss: 5.5749e-05 - 9ms/epoch - 3ms/step\n",

            "Epoch 121/300\n",

            "3/3 - 0s - loss: 5.0933e-05 - 11ms/epoch - 4ms/step\n",

            "Epoch 122/300\n",

            "3/3 - 0s - loss: 5.1243e-05 - 11ms/epoch - 4ms/step\n",

            "Epoch 123/300\n",

            "3/3 - 0s - loss: 4.6262e-05 - 12ms/epoch - 4ms/step\n",

            "Epoch 124/300\n",

            "3/3 - 0s - loss: 4.6049e-05 - 10ms/epoch - 3ms/step\n",

            "Epoch 125/300\n",

            "3/3 - 0s - loss: 4.2880e-05 - 10ms/epoch - 3ms/step\n",

            "Epoch 126/300\n",

            "3/3 - 0s - loss: 4.0773e-05 - 11ms/epoch - 4ms/step\n",

            "Epoch 127/300\n",

            "3/3 - 0s - loss: 4.0393e-05 - 10ms/epoch - 3ms/step\n",

            "Epoch 128/300\n",

            "3/3 - 0s - loss: 4.4048e-05 - 10ms/epoch - 3ms/step\n",

            "Epoch 129/300\n",

            "3/3 - 0s - loss: 3.5081e-05 - 10ms/epoch - 3ms/step\n",

            "Epoch 130/300\n",

            "3/3 - 0s - loss: 3.8248e-05 - 10ms/epoch - 3ms/step\n",

            "Epoch 131/300\n",

            "3/3 - 0s - loss: 3.2229e-05 - 10ms/epoch - 3ms/step\n",

            "Epoch 132/300\n",

            "3/3 - 0s - loss: 3.0671e-05 - 9ms/epoch - 3ms/step\n",

            "Epoch 133/300\n",

            "3/3 - 0s - loss: 3.0762e-05 - 9ms/epoch - 3ms/step\n",

            "Epoch 134/300\n",

            "3/3 - 0s - loss: 2.7715e-05 - 10ms/epoch - 3ms/step\n",

            "Epoch 135/300\n",

            "3/3 - 0s - loss: 2.9620e-05 - 10ms/epoch - 3ms/step\n",

            "Epoch 136/300\n",

            "3/3 - 0s - loss: 2.6592e-05 - 12ms/epoch - 4ms/step\n",

            "Epoch 137/300\n",

            "3/3 - 0s - loss: 3.0199e-05 - 12ms/epoch - 4ms/step\n",

            "Epoch 138/300\n",

            "3/3 - 0s - loss: 2.5135e-05 - 11ms/epoch - 4ms/step\n",

            "Epoch 139/300\n",

            "3/3 - 0s - loss: 2.4707e-05 - 10ms/epoch - 3ms/step\n",

            "Epoch 140/300\n",

            "3/3 - 0s - loss: 2.3645e-05 - 10ms/epoch - 3ms/step\n",

            "Epoch 141/300\n",

            "3/3 - 0s - loss: 2.3964e-05 - 10ms/epoch - 3ms/step\n",

            "Epoch 142/300\n",

            "3/3 - 0s - loss: 2.2884e-05 - 10ms/epoch - 3ms/step\n",

            "Epoch 143/300\n",

            "3/3 - 0s - loss: 2.0532e-05 - 10ms/epoch - 3ms/step\n",

            "Epoch 144/300\n",

            "3/3 - 0s - loss: 2.1781e-05 - 11ms/epoch - 4ms/step\n",

            "Epoch 145/300\n",

            "3/3 - 0s - loss: 2.0647e-05 - 11ms/epoch - 4ms/step\n",

            "Epoch 146/300\n",

            "3/3 - 0s - loss: 1.9495e-05 - 12ms/epoch - 4ms/step\n",

            "Epoch 147/300\n",

            "3/3 - 0s - loss: 1.8582e-05 - 11ms/epoch - 4ms/step\n",

            "Epoch 148/300\n",

            "3/3 - 0s - loss: 1.8290e-05 - 11ms/epoch - 4ms/step\n",

            "Epoch 149/300\n",

            "3/3 - 0s - loss: 1.7448e-05 - 11ms/epoch - 4ms/step\n",

            "Epoch 150/300\n",

            "3/3 - 0s - loss: 1.7882e-05 - 10ms/epoch - 3ms/step\n",

            "Epoch 151/300\n",

            "3/3 - 0s - loss: 1.6310e-05 - 11ms/epoch - 4ms/step\n",

            "Epoch 152/300\n",

            "3/3 - 0s - loss: 1.6778e-05 - 10ms/epoch - 3ms/step\n",

            "Epoch 153/300\n",

            "3/3 - 0s - loss: 1.5487e-05 - 11ms/epoch - 4ms/step\n",

            "Epoch 154/300\n",

            "3/3 - 0s - loss: 1.5726e-05 - 10ms/epoch - 3ms/step\n",

            "Epoch 155/300\n",

            "3/3 - 0s - loss: 1.5094e-05 - 11ms/epoch - 4ms/step\n",

            "Epoch 156/300\n",

            "3/3 - 0s - loss: 1.4633e-05 - 10ms/epoch - 3ms/step\n",

            "Epoch 157/300\n",

            "3/3 - 0s - loss: 1.4032e-05 - 9ms/epoch - 3ms/step\n",

            "Epoch 158/300\n",

            "3/3 - 0s - loss: 1.3945e-05 - 10ms/epoch - 3ms/step\n",

            "Epoch 159/300\n",

            "3/3 - 0s - loss: 1.3617e-05 - 10ms/epoch - 3ms/step\n",

            "Epoch 160/300\n",

            "3/3 - 0s - loss: 1.3157e-05 - 11ms/epoch - 4ms/step\n",

            "Epoch 161/300\n",

            "3/3 - 0s - loss: 1.2828e-05 - 10ms/epoch - 3ms/step\n",

            "Epoch 162/300\n",

            "3/3 - 0s - loss: 1.2695e-05 - 14ms/epoch - 5ms/step\n",

            "Epoch 163/300\n",

            "3/3 - 0s - loss: 1.2114e-05 - 11ms/epoch - 4ms/step\n",

            "Epoch 164/300\n",

            "3/3 - 0s - loss: 1.2344e-05 - 11ms/epoch - 4ms/step\n",

            "Epoch 165/300\n",

            "3/3 - 0s - loss: 1.2030e-05 - 10ms/epoch - 3ms/step\n",

            "Epoch 166/300\n",

            "3/3 - 0s - loss: 1.1489e-05 - 10ms/epoch - 3ms/step\n",

            "Epoch 167/300\n",

            "3/3 - 0s - loss: 1.1404e-05 - 10ms/epoch - 3ms/step\n",

            "Epoch 168/300\n",

            "3/3 - 0s - loss: 1.0903e-05 - 11ms/epoch - 4ms/step\n",

            "Epoch 169/300\n",

            "3/3 - 0s - loss: 1.1463e-05 - 10ms/epoch - 3ms/step\n",

            "Epoch 170/300\n",

            "3/3 - 0s - loss: 1.0795e-05 - 11ms/epoch - 4ms/step\n",

            "Epoch 171/300\n",

            "3/3 - 0s - loss: 1.0607e-05 - 10ms/epoch - 3ms/step\n",

            "Epoch 172/300\n",

            "3/3 - 0s - loss: 1.1208e-05 - 10ms/epoch - 3ms/step\n",

            "Epoch 173/300\n",

            "3/3 - 0s - loss: 9.3615e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 174/300\n",

            "3/3 - 0s - loss: 9.6801e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 175/300\n",

            "3/3 - 0s - loss: 9.5731e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 176/300\n",

            "3/3 - 0s - loss: 8.6858e-06 - 12ms/epoch - 4ms/step\n",

            "Epoch 177/300\n",

            "3/3 - 0s - loss: 9.1015e-06 - 12ms/epoch - 4ms/step\n",

            "Epoch 178/300\n",

            "3/3 - 0s - loss: 8.4440e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 179/300\n",

            "3/3 - 0s - loss: 8.5789e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 180/300\n",

            "3/3 - 0s - loss: 8.1958e-06 - 9ms/epoch - 3ms/step\n",

            "Epoch 181/300\n",

            "3/3 - 0s - loss: 8.6107e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 182/300\n",

            "3/3 - 0s - loss: 8.1158e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 183/300\n",

            "3/3 - 0s - loss: 8.0382e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 184/300\n",

            "3/3 - 0s - loss: 8.3136e-06 - 12ms/epoch - 4ms/step\n",

            "Epoch 185/300\n",

            "3/3 - 0s - loss: 7.4128e-06 - 9ms/epoch - 3ms/step\n",

            "Epoch 186/300\n",

            "3/3 - 0s - loss: 7.8207e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 187/300\n",

            "3/3 - 0s - loss: 7.5321e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 188/300\n",

            "3/3 - 0s - loss: 7.2220e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 189/300\n",

            "3/3 - 0s - loss: 7.7224e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 190/300\n",

            "3/3 - 0s - loss: 7.0167e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 191/300\n",

            "3/3 - 0s - loss: 6.7280e-06 - 9ms/epoch - 3ms/step\n",

            "Epoch 192/300\n",

            "3/3 - 0s - loss: 7.3436e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 193/300\n",

            "3/3 - 0s - loss: 6.4096e-06 - 12ms/epoch - 4ms/step\n",

            "Epoch 194/300\n",

            "3/3 - 0s - loss: 7.1740e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 195/300\n",

            "3/3 - 0s - loss: 6.2640e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 196/300\n",

            "3/3 - 0s - loss: 5.9245e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 197/300\n",

            "3/3 - 0s - loss: 5.7851e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 198/300\n",

            "3/3 - 0s - loss: 6.1468e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 199/300\n",

            "3/3 - 0s - loss: 6.6124e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 200/300\n",

            "3/3 - 0s - loss: 6.1927e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 201/300\n",

            "3/3 - 0s - loss: 6.0981e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 202/300\n",

            "3/3 - 0s - loss: 5.6132e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 203/300\n",

            "3/3 - 0s - loss: 5.4261e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 204/300\n",

            "3/3 - 0s - loss: 5.8134e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 205/300\n",

            "3/3 - 0s - loss: 5.3261e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 206/300\n",

            "3/3 - 0s - loss: 5.6922e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 207/300\n",

            "3/3 - 0s - loss: 5.3886e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 208/300\n",

            "3/3 - 0s - loss: 5.2678e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 209/300\n",

            "3/3 - 0s - loss: 5.1627e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 210/300\n",

            "3/3 - 0s - loss: 5.3975e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 211/300\n",

            "3/3 - 0s - loss: 4.5620e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 212/300\n",

            "3/3 - 0s - loss: 5.5581e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 213/300\n",

            "3/3 - 0s - loss: 4.7086e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 214/300\n",

            "3/3 - 0s - loss: 5.0151e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 215/300\n",

            "3/3 - 0s - loss: 4.8620e-06 - 12ms/epoch - 4ms/step\n",

            "Epoch 216/300\n",

            "3/3 - 0s - loss: 4.2306e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 217/300\n",

            "3/3 - 0s - loss: 4.8022e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 218/300\n",

            "3/3 - 0s - loss: 4.1304e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 219/300\n",

            "3/3 - 0s - loss: 4.4333e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 220/300\n",

            "3/3 - 0s - loss: 4.3856e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 221/300\n",

            "3/3 - 0s - loss: 3.7280e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 222/300\n",

            "3/3 - 0s - loss: 4.2963e-06 - 12ms/epoch - 4ms/step\n",

            "Epoch 223/300\n",

            "3/3 - 0s - loss: 3.8061e-06 - 12ms/epoch - 4ms/step\n",

            "Epoch 224/300\n",

            "3/3 - 0s - loss: 4.0099e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 225/300\n",

            "3/3 - 0s - loss: 4.1208e-06 - 12ms/epoch - 4ms/step\n",

            "Epoch 226/300\n",

            "3/3 - 0s - loss: 3.7706e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 227/300\n",

            "3/3 - 0s - loss: 3.6554e-06 - 12ms/epoch - 4ms/step\n",

            "Epoch 228/300\n",

            "3/3 - 0s - loss: 3.8271e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 229/300\n",

            "3/3 - 0s - loss: 3.6408e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 230/300\n",

            "3/3 - 0s - loss: 3.8301e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 231/300\n",

            "3/3 - 0s - loss: 3.9109e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 232/300\n",

            "3/3 - 0s - loss: 3.7608e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 233/300\n",

            "3/3 - 0s - loss: 4.0293e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 234/300\n",

            "3/3 - 0s - loss: 3.4236e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 235/300\n",

            "3/3 - 0s - loss: 3.7379e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 236/300\n",

            "3/3 - 0s - loss: 3.6925e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 237/300\n",

            "3/3 - 0s - loss: 3.3738e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 238/300\n",

            "3/3 - 0s - loss: 3.6437e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 239/300\n",

            "3/3 - 0s - loss: 3.1176e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 240/300\n",

            "3/3 - 0s - loss: 3.1168e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 241/300\n",

            "3/3 - 0s - loss: 3.1751e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 242/300\n",

            "3/3 - 0s - loss: 3.0683e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 243/300\n",

            "3/3 - 0s - loss: 2.9676e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 244/300\n",

            "3/3 - 0s - loss: 3.3126e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 245/300\n",

            "3/3 - 0s - loss: 2.9347e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 246/300\n",

            "3/3 - 0s - loss: 3.2757e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 247/300\n",

            "3/3 - 0s - loss: 3.2142e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 248/300\n",

            "3/3 - 0s - loss: 3.2176e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 249/300\n",

            "3/3 - 0s - loss: 2.8382e-06 - 9ms/epoch - 3ms/step\n",

            "Epoch 250/300\n",

            "3/3 - 0s - loss: 3.0712e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 251/300\n",

            "3/3 - 0s - loss: 2.9336e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 252/300\n",

            "3/3 - 0s - loss: 2.7713e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 253/300\n",

            "3/3 - 0s - loss: 3.1205e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 254/300\n",

            "3/3 - 0s - loss: 2.7059e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 255/300\n",

            "3/3 - 0s - loss: 2.9267e-06 - 12ms/epoch - 4ms/step\n",

            "Epoch 256/300\n",

            "3/3 - 0s - loss: 2.6573e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 257/300\n",

            "3/3 - 0s - loss: 2.7912e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 258/300\n",

            "3/3 - 0s - loss: 2.9237e-06 - 12ms/epoch - 4ms/step\n",

            "Epoch 259/300\n",

            "3/3 - 0s - loss: 3.3254e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 260/300\n",

            "3/3 - 0s - loss: 3.2088e-06 - 9ms/epoch - 3ms/step\n",

            "Epoch 261/300\n",

            "3/3 - 0s - loss: 2.8664e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 262/300\n",

            "3/3 - 0s - loss: 2.8006e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 263/300\n",

            "3/3 - 0s - loss: 4.1102e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 264/300\n",

            "3/3 - 0s - loss: 3.0658e-06 - 9ms/epoch - 3ms/step\n",

            "Epoch 265/300\n",

            "3/3 - 0s - loss: 2.1988e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 266/300\n",

            "3/3 - 0s - loss: 2.3797e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 267/300\n",

            "3/3 - 0s - loss: 2.1137e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 268/300\n",

            "3/3 - 0s - loss: 2.2700e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 269/300\n",

            "3/3 - 0s - loss: 2.3573e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 270/300\n",

            "3/3 - 0s - loss: 2.0647e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 271/300\n",

            "3/3 - 0s - loss: 2.2215e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 272/300\n",

            "3/3 - 0s - loss: 2.2609e-06 - 12ms/epoch - 4ms/step\n",

            "Epoch 273/300\n",

            "3/3 - 0s - loss: 2.1239e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 274/300\n",

            "3/3 - 0s - loss: 2.0631e-06 - 13ms/epoch - 4ms/step\n",

            "Epoch 275/300\n",

            "3/3 - 0s - loss: 2.1012e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 276/300\n",

            "3/3 - 0s - loss: 2.0158e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 277/300\n",

            "3/3 - 0s - loss: 2.0523e-06 - 13ms/epoch - 4ms/step\n",

            "Epoch 278/300\n",

            "3/3 - 0s - loss: 2.0416e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 279/300\n",

            "3/3 - 0s - loss: 2.1009e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 280/300\n",

            "3/3 - 0s - loss: 1.9421e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 281/300\n",

            "3/3 - 0s - loss: 1.9659e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 282/300\n",

            "3/3 - 0s - loss: 1.9462e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 283/300\n",

            "3/3 - 0s - loss: 1.7357e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 284/300\n",

            "3/3 - 0s - loss: 1.7213e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 285/300\n",

            "3/3 - 0s - loss: 1.7748e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 286/300\n",

            "3/3 - 0s - loss: 1.9336e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 287/300\n",

            "3/3 - 0s - loss: 1.7405e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 288/300\n",

            "3/3 - 0s - loss: 2.1245e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 289/300\n",

            "3/3 - 0s - loss: 2.0561e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 290/300\n",

            "3/3 - 0s - loss: 2.0798e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 291/300\n",

            "3/3 - 0s - loss: 1.7068e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 292/300\n",

            "3/3 - 0s - loss: 1.9205e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 293/300\n",

            "3/3 - 0s - loss: 1.7442e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 294/300\n",

            "3/3 - 0s - loss: 1.7597e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 295/300\n",

            "3/3 - 0s - loss: 1.5517e-06 - 11ms/epoch - 4ms/step\n",

            "Epoch 296/300\n",

            "3/3 - 0s - loss: 1.7890e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 297/300\n",

            "3/3 - 0s - loss: 1.8468e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 298/300\n",

            "3/3 - 0s - loss: 1.6581e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 299/300\n",

            "3/3 - 0s - loss: 1.6944e-06 - 10ms/epoch - 3ms/step\n",

            "Epoch 300/300\n",

            "3/3 - 0s - loss: 1.7180e-06 - 11ms/epoch - 4ms/step\n"

          ]

        },

        {

          "output_type": "execute_result",

          "data": {

            "text/plain": [

              "<keras.src.callbacks.History at 0x786a65815d20>"

            ]

          },

          "metadata": {},

          "execution_count": 56

        }

      ]

    },

    {

      "cell_type": "code",

      "metadata": {

        "id": "n-aNP4n3sqG_",

        "outputId": "e3333292-23f1-4a0a-c06f-e01a74493282",

        "colab": {

          "base_uri": "https://localhost:8080/",

          "height": 443

        }

      },

      "source": [

        "# Plotting code, feel free to ignore.\n",

        "h = 1.0\n",

        "x_min, x_max = X[:, 0].min() - 5, X[:, 0].max() + 5\n",

        "y_min, y_max = X[:, 1].min() - 5, X[:, 1].max() + 5\n",

        "xx, yy = np.meshgrid(np.arange(x_min, x_max, h),\n",

        "                     np.arange(y_min, y_max, h))\n",

        "\n",

        "# here \"model\" is your model's prediction (classification) function\n",

        "Z = tn_model.predict(np.c_[xx.ravel(), yy.ravel()])\n",

        "\n",

        "# Put the result into a color plot\n",

        "Z = Z.reshape(xx.shape)\n",

        "plt.contourf(xx, yy, Z)\n",

        "plt.axis('off')\n",

        "\n",

        "# Plot also the training points\n",

        "plt.scatter(X[:, 0], X[:, 1], c=Y, cmap=plt.cm.Paired)"

      ],

      "execution_count": 57,

      "outputs": [

        {

          "output_type": "stream",

          "name": "stdout",

          "text": [

            "14/14 [==============================] - 0s 2ms/step\n"

          ]

        },

        {

          "output_type": "execute_result",

          "data": {

            "text/plain": [

              "<matplotlib.collections.PathCollection at 0x786a65602620>"

            ]

          },

          "metadata": {},

          "execution_count": 57

        },

        {

          "output_type": "display_data",

          "data": {

            "text/plain": [

              "<Figure size 640x480 with 1 Axes>"

            ],

            "image/png": "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\n"

          },

          "metadata": {}

        }

      ]

    },

    {

      "cell_type": "code",

      "source": [

        "seconds = time.time()\n",

        "print(\"Time in seconds since end of run:\", seconds)\n",

        "local_time = time.ctime(seconds)\n",

        "print(local_time)"

      ],

      "metadata": {

        "colab": {

          "base_uri": "https://localhost:8080/",

          "height": 0

        },

        "id": "wfZCzuq9KY9b",

        "outputId": "55094d23-e05a-4e9e-9bc8-e8b5d91e44ef"

      },

      "execution_count": 58,

      "outputs": [

        {

          "output_type": "stream",

          "name": "stdout",

          "text": [

            "Time in seconds since end of run: 1709531172.417285\n",

            "Mon Mar  4 05:46:12 2024\n"

          ]

        }

      ]

    },

    {

      "cell_type": "code",

      "source": [

        "seconds = time.time()\n",

        "print(\"Time in seconds since beginning of run:\", seconds)\n",

        "local_time = time.ctime(seconds)\n",

        "print(local_time)"

      ],

      "metadata": {

        "colab": {

          "base_uri": "https://localhost:8080/",

          "height": 0

        },

        "id": "Ft6S13x6KuEQ",

        "outputId": "425bcc82-b55e-49f2-86b6-1932e5ed8025"

      },

      "execution_count": 59,

      "outputs": [

        {

          "output_type": "stream",

          "name": "stdout",

          "text": [

            "Time in seconds since beginning of run: 1709531172.4279752\n",

            "Mon Mar  4 05:46:12 2024\n"

          ]

        }

      ]

    },

    {

      "cell_type": "markdown",

      "metadata": {

        "id": "BMxSJo5gtOmQ"

      },

      "source": [

        "# VS Fully Connected"

      ]

    },

    {

      "cell_type": "code",

      "metadata": {

        "id": "NKQx7stYswzU",

        "outputId": "fd3f1273-2a7c-4e72-eb34-1da87a65216d",

        "colab": {

          "base_uri": "https://localhost:8080/",

          "height": 11458

        }

      },

      "source": [

        "fc_model.compile(optimizer=\"adam\", loss=\"mean_squared_error\")\n",

        "fc_model.fit(X, Y, epochs=300, verbose=2)\n",

        "# Plotting code, feel free to ignore.\n",

        "h = 1.0\n",

        "x_min, x_max = X[:, 0].min() - 5, X[:, 0].max() + 5\n",

        "y_min, y_max = X[:, 1].min() - 5, X[:, 1].max() + 5\n",

        "xx, yy = np.meshgrid(np.arange(x_min, x_max, h),\n",

        "                     np.arange(y_min, y_max, h))\n",

        "\n",

        "# here \"model\" is your model's prediction (classification) function\n",

        "Z = fc_model.predict(np.c_[xx.ravel(), yy.ravel()])\n",

        "\n",

        "# Put the result into a color plot\n",

        "Z = Z.reshape(xx.shape)\n",

        "plt.contourf(xx, yy, Z)\n",

        "plt.axis('off')\n",

        "\n",

        "# Plot also the training points\n",

        "plt.scatter(X[:, 0], X[:, 1], c=Y, cmap=plt.cm.Paired)"

      ],

      "execution_count": 60,

      "outputs": [

        {

          "output_type": "stream",

          "name": "stdout",

          "text": [

            "Epoch 1/300\n",

            "3/3 - 1s - loss: 0.5656 - 654ms/epoch - 218ms/step\n",

            "Epoch 2/300\n",

            "3/3 - 0s - loss: 0.1959 - 24ms/epoch - 8ms/step\n",

            "Epoch 3/300\n",

            "3/3 - 0s - loss: 0.1423 - 22ms/epoch - 7ms/step\n",

            "Epoch 4/300\n",

            "3/3 - 0s - loss: 0.0917 - 23ms/epoch - 8ms/step\n",

            "Epoch 5/300\n",

            "3/3 - 0s - loss: 0.0828 - 25ms/epoch - 8ms/step\n",

            "Epoch 6/300\n",

            "3/3 - 0s - loss: 0.0827 - 23ms/epoch - 8ms/step\n",

            "Epoch 7/300\n",

            "3/3 - 0s - loss: 0.0680 - 25ms/epoch - 8ms/step\n",

            "Epoch 8/300\n",

            "3/3 - 0s - loss: 0.0680 - 24ms/epoch - 8ms/step\n",

            "Epoch 9/300\n",

            "3/3 - 0s - loss: 0.0605 - 28ms/epoch - 9ms/step\n",

            "Epoch 10/300\n",

            "3/3 - 0s - loss: 0.0632 - 24ms/epoch - 8ms/step\n",

            "Epoch 11/300\n",

            "3/3 - 0s - loss: 0.0537 - 22ms/epoch - 7ms/step\n",

            "Epoch 12/300\n",

            "3/3 - 0s - loss: 0.0523 - 27ms/epoch - 9ms/step\n",

            "Epoch 13/300\n",

            "3/3 - 0s - loss: 0.0522 - 24ms/epoch - 8ms/step\n",

            "Epoch 14/300\n",

            "3/3 - 0s - loss: 0.0483 - 21ms/epoch - 7ms/step\n",

            "Epoch 15/300\n",

            "3/3 - 0s - loss: 0.0498 - 23ms/epoch - 8ms/step\n",

            "Epoch 16/300\n",

            "3/3 - 0s - loss: 0.0444 - 24ms/epoch - 8ms/step\n",

            "Epoch 17/300\n",

            "3/3 - 0s - loss: 0.0487 - 27ms/epoch - 9ms/step\n",

            "Epoch 18/300\n",

            "3/3 - 0s - loss: 0.0467 - 25ms/epoch - 8ms/step\n",

            "Epoch 19/300\n",

            "3/3 - 0s - loss: 0.0419 - 26ms/epoch - 9ms/step\n",

            "Epoch 20/300\n",

            "3/3 - 0s - loss: 0.0439 - 27ms/epoch - 9ms/step\n",

            "Epoch 21/300\n",

            "3/3 - 0s - loss: 0.0406 - 28ms/epoch - 9ms/step\n",

            "Epoch 22/300\n",

            "3/3 - 0s - loss: 0.0414 - 23ms/epoch - 8ms/step\n",

            "Epoch 23/300\n",