training_keras_basics.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "training keras basics.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "authorship_tag": "ABX9TyNHPQHiE02N36vRqbocFXLM",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/faheem-khaskheli/tensorflow_basics/blob/main/training_keras_basics.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "KMtCNPsS09dd"
      },
      "source": [
        "import tensorflow as tf\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "13T7MJkv2uph"
      },
      "source": [
        "class Model(object):\n",
        "  def __init__(self):\n",
        "    self.w = tf.Variable(2.0)\n",
        "    self.b = tf.Variable(1.0)\n",
        "  \n",
        "  def __call__(self, x):\n",
        "    return self.w * x + self.b\n",
        "  \n",
        "model = Model()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ddMsbTLL3Y54"
      },
      "source": [
        "def loss(predicted_y, target_y):\n",
        "  return tf.reduce_mean(tf.square(predicted_y - target_y))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "AhRby25w3lM5"
      },
      "source": [
        "TRUE_w = 3.0\n",
        "TRUE_b = 2.0\n",
        "NUM_EXAMPLES = 1000\n",
        "\n",
        "xs = tf.random.normal(shape=[NUM_EXAMPLES])\n",
        "\n",
        "ys = (TRUE_w * xs) + TRUE_b"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 282
        },
        "id": "1Fpwoi0O31cB",
        "outputId": "8db2d2d9-f897-4ac0-a29f-3ea0b7e1477c"
      },
      "source": [
        "def plot_data(inputs, outputs, predicted_outputs):\n",
        "  real = plt.scatter(inputs, outputs, c='b')\n",
        "  predicted = plt.scatter(inputs, predicted_outputs, c='r')\n",
        "  plt.legend((real, predicted), ('Real Data', 'Predicted Data'))\n",
        "  plt.show()\n",
        "\n",
        "plot_data(xs, ys, model(xs))\n",
        "print('Current loss: %1.6f' % loss(model(xs), ys).numpy())"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "text": [
            "Current loss: 1.938138\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2ymaq1Fc4WsJ"
      },
      "source": [
        "def train(model, inputs, outputs, learning_rate):\n",
        "  with tf.GradientTape() as t:\n",
        "    current_loss = loss(model(inputs), outputs)\n",
        "  dw, db = t.gradient(current_loss, [model.w, model.b])\n",
        "  model.w.assign_sub(learning_rate * dw)\n",
        "  model.b.assign_sub(learning_rate * db)\n",
        "\n",
        "  return current_loss"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "rkp7APSw5P84",
        "outputId": "5348e73f-0bb2-4e6c-b468-14089a40720c"
      },
      "source": [
        "model = Model()\n",
        "\n",
        "list_w, list_b = [], []\n",
        "epochs = range(40)\n",
        "losses = []\n",
        "for epoch in epochs:\n",
        "  list_w.append(model.w.numpy())\n",
        "  list_b.append(model.b.numpy())\n",
        "  current_loss = train(model, xs, ys, learning_rate=0.1)\n",
        "  losses.append(current_loss)\n",
        "  print('Epoch %2d: w=%1.2f b=%1.2f, loss=%2.5f' %\n",
        "        (epoch, \n",
        "         list_w[-1], \n",
        "         list_b[-1], \n",
        "         current_loss))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch  0: w=2.00 b=1.00, loss=1.93814\n",
            "Epoch  1: w=2.19 b=1.20, loss=1.25875\n",
            "Epoch  2: w=2.34 b=1.36, loss=0.81783\n",
            "Epoch  3: w=2.46 b=1.49, loss=0.53157\n",
            "Epoch  4: w=2.56 b=1.59, loss=0.34564\n",
            "Epoch  5: w=2.64 b=1.68, loss=0.22484\n",
            "Epoch  6: w=2.71 b=1.74, loss=0.14631\n",
            "Epoch  7: w=2.76 b=1.79, loss=0.09525\n",
            "Epoch  8: w=2.81 b=1.84, loss=0.06204\n",
            "Epoch  9: w=2.84 b=1.87, loss=0.04042\n",
            "Epoch 10: w=2.87 b=1.90, loss=0.02635\n",
            "Epoch 11: w=2.90 b=1.92, loss=0.01718\n",
            "Epoch 12: w=2.91 b=1.93, loss=0.01121\n",
            "Epoch 13: w=2.93 b=1.95, loss=0.00731\n",
            "Epoch 14: w=2.94 b=1.96, loss=0.00478\n",
            "Epoch 15: w=2.95 b=1.97, loss=0.00312\n",
            "Epoch 16: w=2.96 b=1.97, loss=0.00204\n",
            "Epoch 17: w=2.97 b=1.98, loss=0.00133\n",
            "Epoch 18: w=2.98 b=1.98, loss=0.00087\n",
            "Epoch 19: w=2.98 b=1.99, loss=0.00057\n",
            "Epoch 20: w=2.98 b=1.99, loss=0.00037\n",
            "Epoch 21: w=2.99 b=1.99, loss=0.00024\n",
            "Epoch 22: w=2.99 b=1.99, loss=0.00016\n",
            "Epoch 23: w=2.99 b=1.99, loss=0.00010\n",
            "Epoch 24: w=2.99 b=2.00, loss=0.00007\n",
            "Epoch 25: w=2.99 b=2.00, loss=0.00005\n",
            "Epoch 26: w=3.00 b=2.00, loss=0.00003\n",
            "Epoch 27: w=3.00 b=2.00, loss=0.00002\n",
            "Epoch 28: w=3.00 b=2.00, loss=0.00001\n",
            "Epoch 29: w=3.00 b=2.00, loss=0.00001\n",
            "Epoch 30: w=3.00 b=2.00, loss=0.00001\n",
            "Epoch 31: w=3.00 b=2.00, loss=0.00000\n",
            "Epoch 32: w=3.00 b=2.00, loss=0.00000\n",
            "Epoch 33: w=3.00 b=2.00, loss=0.00000\n",
            "Epoch 34: w=3.00 b=2.00, loss=0.00000\n",
            "Epoch 35: w=3.00 b=2.00, loss=0.00000\n",
            "Epoch 36: w=3.00 b=2.00, loss=0.00000\n",
            "Epoch 37: w=3.00 b=2.00, loss=0.00000\n",
            "Epoch 38: w=3.00 b=2.00, loss=0.00000\n",
            "Epoch 39: w=3.00 b=2.00, loss=0.00000\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 265
        },
        "id": "R8zt-xov6FND",
        "outputId": "ad0f656a-e28a-44db-9ebd-9b1a3eb40783"
      },
      "source": [
        "plt.plot(epochs, list_w, 'r',\n",
        "         epochs, list_b, 'b')\n",
        "plt.plot([TRUE_w] * len(epochs), 'r--',\n",
        "         [TRUE_b] * len(epochs), 'b--')\n",
        "plt.legend(['w', 'b', 'True w', 'True b'])\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 265
        },
        "id": "k2Ug003e6wf_",
        "outputId": "2130c869-cdd5-43c8-daee-fb94bc42b1e8"
      },
      "source": [
        "test_inputs = tf.random.normal(shape=[NUM_EXAMPLES])\n",
        "test_outputs = test_inputs * TRUE_w + TRUE_b\n",
        "\n",
        "predicted_test_outputs = model(test_inputs)\n",
        "plot_data(test_inputs, test_outputs, predicted_test_outputs)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YTp7SpIr7bvZ"
      },
      "source": [
        "def plot_loss_for_weights(weights_list, losses):\n",
        "  for idx, weights in enumerate(weights_list):\n",
        "    plt.subplot(120 + idx + 1)\n",
        "    plt.plot(weights['values'], losses, 'r')\n",
        "    plt.plot(weights['values'], losses, 'bo')\n",
        "    plt.xlabel(weights['name'])\n",
        "    plt.ylabel(['Loss'])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 280
        },
        "id": "7FIM-WEl8G_k",
        "outputId": "dd9e75c6-4ed9-418b-cfb7-61ec11494da2"
      },
      "source": [
        "weights_list = [{'name': 'w',\n",
        "                 'values': list_w},\n",
        "                {'name': 'b',\n",
        "                 'values': list_b}]\n",
        "            \n",
        "plot_loss_for_weights(weights_list, losses)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "GAAaeHtKDwZ1"
      },
      "source": [
        "import tensorflow_datasets as tfds"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "QmBtghMeC464"
      },
      "source": [
        "train_data = tfds.load(\"fashion_mnist\", split='train')\n",
        "test_data = tfds.load(\"fashion_mnist\", split='test')\n",
        "\n",
        "def format_image(data):\n",
        "  image = data[\"image\"]\n",
        "  image = tf.reshape(image, [-1])\n",
        "  image = tf.cast(image, 'float32')\n",
        "  image = image / 255.0\n",
        "  return image, data[\"label\"]\n",
        "\n",
        "train_data = train_data.map(format_image)\n",
        "test_data = test_data.map(format_image)\n",
        "\n",
        "batch_size = 64\n",
        "train = train_data.shuffle(buffer_size=1024).batch(batch_size)\n",
        "test = test_data.batch(batch_size)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ASRQ-KeJE5np"
      },
      "source": [
        "loss_object = tf.keras.losses.SparseCategoricalCrossentropy()\n",
        "\n",
        "optimizer = tf.keras.optimizers.Adam()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3zhnGO9zCQNh"
      },
      "source": [
        "def base_model():\n",
        "  inputs = tf.keras.Input(shape=(784,), \n",
        "                          name='clothing')\n",
        "  x = tf.keras.layers.Dense(64, \n",
        "                            activation='relu', \n",
        "                            name='dense_1')(inputs)\n",
        "  x = tf.keras.layers.Dense(64, \n",
        "                            activation='relu', \n",
        "                            name='dense_2')(x)\n",
        "  outputs = tf.keras.layers.Dense(10, \n",
        "                                  activation='softmax', \n",
        "                                  name='predictions')(x)\n",
        "  model = tf.keras.Model(inputs=inputs, \n",
        "                         outputs=outputs)\n",
        "  return model"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TUSwLh1QGPlh"
      },
      "source": [
        "def apply_gradient(optimizer, model, x, y):\n",
        "  with tf.GradientTape() as tape:\n",
        "    y_pred = model(x)\n",
        "    loss_value = loss_object(y_true=y, y_pred=y_pred)\n",
        "    gradients = tape.gradient(loss_value, model.trainable_weights)\n",
        "  optimizer.apply_gradients(zip(gradients, model.trainable_weights))\n",
        "  return y_pred, loss_value"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "w57oI2L0Brdv"
      },
      "source": [
        " def train_data_for_one_epoch():\n",
        "  losses = []\n",
        "  for step, (x_batch, y_batch) in enumerate(train_dataset):\n",
        "    logits, loss_value = apply_gradient(optimizer, \n",
        "                                        model, \n",
        "                                        x_batch, \n",
        "                                        y_batch)\n",
        "    losses.append(loss_value)\n",
        "  return losses"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "AXDO42vZ8dPi"
      },
      "source": [
        "model = base_model()\n",
        "epochs = 20\n",
        "for epoch in range(epochs):\n",
        "\n",
        "  losses_train = train_data_for_one_epoch()\n",
        "\n",
        "  losses_val = perform_validation()\n",
        "\n",
        "  losses_train_mean = np.mean(losses_train)\n",
        "  losses_val_mean = np.mean(losses_val)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "UDC-NZmtLdGb"
      },
      "source": [
        "def apply_gradient(optimizer, model, X_batch, y_batch):\n",
        "  with tf.GradientTape() as tape:\n",
        "    y_pred  = model(x_batch)\n",
        "    loss = loss_object(y_batch, y_pred)\n",
        "  grad = tape.gradient(loss, y_batch)\n",
        "  optimizer.apply_gradients(zip(grad), model.trainable_variables)\n",
        "  return y_pred, loss"
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}