{ "nbformat": 4, "nbformat_minor": 0, "metadata": { "colab": { "provenance": [], "gpuType": "T4" }, "kernelspec": { "name": "python3", "display_name": "Python 3" }, "language_info": { "name": "python" }, "accelerator": "GPU" }, "cells": [ { "cell_type": "code", "execution_count": 1, "metadata": { "id": "yDgsM3-9rlBJ" }, "outputs": [], "source": [ "import tensorflow as tf\n", "from tensorflow import keras\n", "import numpy as np\n", "import matplotlib.pyplot as plt\n", "from tqdm import tqdm\n", "from IPython import display" ] }, { "cell_type": "code", "source": [ "(x_train, y_train), (x_test, y_test) = tf.keras.datasets.fashion_mnist.load_data()\n", "x_train = x_train.astype(np.float32) / 255.0\n", "x_test = x_test.astype(np.float32) / 255.0\n", "x_train.shape, x_test.shape" ], "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "-B_kH_xormEI", "outputId": "262633e6-22d3-422c-bc3b-ac58867d222f" }, "execution_count": 2, "outputs": [ { "output_type": "stream", "name": "stdout", "text": [ "Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/train-labels-idx1-ubyte.gz\n", "\u001b[1m29515/29515\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 0us/step\n", "Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/train-images-idx3-ubyte.gz\n", "\u001b[1m26421880/26421880\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 0us/step\n", "Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/t10k-labels-idx1-ubyte.gz\n", "\u001b[1m5148/5148\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 0us/step\n", "Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/t10k-images-idx3-ubyte.gz\n", "\u001b[1m4422102/4422102\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 0us/step\n" ] }, { "output_type": "execute_result", "data": { "text/plain": [ "((60000, 28, 28), (10000, 28, 28))" ] }, "metadata": {}, "execution_count": 2 } ] }, { "cell_type": "code", "source": [ "plt.figure(figsize =(10, 10))\n", "for i in range(25):\n", " plt.subplot(5, 5, i + 1)\n", " plt.xticks([])\n", " plt.yticks([])\n", " plt.grid(False)\n", " plt.imshow(x_train[i], cmap = plt.cm.binary)\n", "plt.show()" ], "metadata": { "id": "t1WehxpFrmGv" }, "execution_count": null, "outputs": [] }, { "cell_type": "code", "source": [ "batch_size = 32\n", "def create_batch(x_train):\n", " dataset = tf.data.Dataset.from_tensor_slices(x_train).shuffle(1000)\n", " dataset = dataset.batch(batch_size, drop_remainder=True).prefetch(1)\n", " return dataset" ], "metadata": { "id": "CABqNhMurmJP" }, "execution_count": 4, "outputs": [] }, { "cell_type": "code", "source": [ "num_features = 100\n", "\n", "generator = keras.models.Sequential([\n", " keras.layers.Dense(7 * 7 * 128, input_shape=[num_features]),\n", " keras.layers.Reshape([7, 7, 128]),\n", " keras.layers.BatchNormalization(),\n", "\n", " keras.layers.Conv2DTranspose(\n", " 64, (5, 5), (2, 2),\n", " padding=\"same\",\n", " activation=\"selu\"\n", " ),\n", "\n", " keras.layers.BatchNormalization(),\n", "\n", " keras.layers.Conv2DTranspose(\n", " 1, (5, 5), (2, 2),\n", " padding=\"same\",\n", " activation=\"tanh\"\n", " ),\n", "])\n", "\n", "generator.summary()" ], "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 455 }, "id": "O9fXIWqcrmLn", "outputId": "d622c695-9a49-4c94-df05-6637334301c9" }, "execution_count": 5, "outputs": [ { "output_type": "stream", "name": "stderr", "text": [ "/usr/local/lib/python3.12/dist-packages/keras/src/layers/core/dense.py:106: UserWarning: Do not pass an `input_shape`/`input_dim` argument to a layer. When using Sequential models, prefer using an `Input(shape)` object as the first layer in the model instead.\n", " super().__init__(activity_regularizer=activity_regularizer, **kwargs)\n" ] }, { "output_type": "display_data", "data": { "text/plain": [ "\u001b[1mModel: \"sequential\"\u001b[0m\n" ], "text/html": [ "
Model: \"sequential\"\n",
              "
\n" ] }, "metadata": {} }, { "output_type": "display_data", "data": { "text/plain": [ "┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓\n", "┃\u001b[1m \u001b[0m\u001b[1mLayer (type) \u001b[0m\u001b[1m \u001b[0m┃\u001b[1m \u001b[0m\u001b[1mOutput Shape \u001b[0m\u001b[1m \u001b[0m┃\u001b[1m \u001b[0m\u001b[1m Param #\u001b[0m\u001b[1m \u001b[0m┃\n", "┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩\n", "│ dense (\u001b[38;5;33mDense\u001b[0m) │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m6272\u001b[0m) │ \u001b[38;5;34m633,472\u001b[0m │\n", "├─────────────────────────────────┼────────────────────────┼───────────────┤\n", "│ reshape (\u001b[38;5;33mReshape\u001b[0m) │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m7\u001b[0m, \u001b[38;5;34m7\u001b[0m, \u001b[38;5;34m128\u001b[0m) │ \u001b[38;5;34m0\u001b[0m │\n", "├─────────────────────────────────┼────────────────────────┼───────────────┤\n", "│ batch_normalization │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m7\u001b[0m, \u001b[38;5;34m7\u001b[0m, \u001b[38;5;34m128\u001b[0m) │ \u001b[38;5;34m512\u001b[0m │\n", "│ (\u001b[38;5;33mBatchNormalization\u001b[0m) │ │ │\n", "├─────────────────────────────────┼────────────────────────┼───────────────┤\n", "│ conv2d_transpose │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m14\u001b[0m, \u001b[38;5;34m14\u001b[0m, \u001b[38;5;34m64\u001b[0m) │ \u001b[38;5;34m204,864\u001b[0m │\n", "│ (\u001b[38;5;33mConv2DTranspose\u001b[0m) │ │ │\n", "├─────────────────────────────────┼────────────────────────┼───────────────┤\n", "│ batch_normalization_1 │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m14\u001b[0m, \u001b[38;5;34m14\u001b[0m, \u001b[38;5;34m64\u001b[0m) │ \u001b[38;5;34m256\u001b[0m │\n", "│ (\u001b[38;5;33mBatchNormalization\u001b[0m) │ │ │\n", "├─────────────────────────────────┼────────────────────────┼───────────────┤\n", "│ conv2d_transpose_1 │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m28\u001b[0m, \u001b[38;5;34m28\u001b[0m, \u001b[38;5;34m1\u001b[0m) │ \u001b[38;5;34m1,601\u001b[0m │\n", "│ (\u001b[38;5;33mConv2DTranspose\u001b[0m) │ │ │\n", "└─────────────────────────────────┴────────────────────────┴───────────────┘\n" ], "text/html": [ "
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓\n",
              "┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃\n",
              "┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩\n",
              "│ dense (Dense)                   │ (None, 6272)           │       633,472 │\n",
              "├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
              "│ reshape (Reshape)               │ (None, 7, 7, 128)      │             0 │\n",
              "├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
              "│ batch_normalization             │ (None, 7, 7, 128)      │           512 │\n",
              "│ (BatchNormalization)            │                        │               │\n",
              "├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
              "│ conv2d_transpose                │ (None, 14, 14, 64)     │       204,864 │\n",
              "│ (Conv2DTranspose)               │                        │               │\n",
              "├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
              "│ batch_normalization_1           │ (None, 14, 14, 64)     │           256 │\n",
              "│ (BatchNormalization)            │                        │               │\n",
              "├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
              "│ conv2d_transpose_1              │ (None, 28, 28, 1)      │         1,601 │\n",
              "│ (Conv2DTranspose)               │                        │               │\n",
              "└─────────────────────────────────┴────────────────────────┴───────────────┘\n",
              "
\n" ] }, "metadata": {} }, { "output_type": "display_data", "data": { "text/plain": [ "\u001b[1m Total params: \u001b[0m\u001b[38;5;34m840,705\u001b[0m (3.21 MB)\n" ], "text/html": [ "
 Total params: 840,705 (3.21 MB)\n",
              "
\n" ] }, "metadata": {} }, { "output_type": "display_data", "data": { "text/plain": [ "\u001b[1m Trainable params: \u001b[0m\u001b[38;5;34m840,321\u001b[0m (3.21 MB)\n" ], "text/html": [ "
 Trainable params: 840,321 (3.21 MB)\n",
              "
\n" ] }, "metadata": {} }, { "output_type": "display_data", "data": { "text/plain": [ "\u001b[1m Non-trainable params: \u001b[0m\u001b[38;5;34m384\u001b[0m (1.50 KB)\n" ], "text/html": [ "
 Non-trainable params: 384 (1.50 KB)\n",
              "
\n" ] }, "metadata": {} } ] }, { "cell_type": "code", "source": [ "discriminator = keras.models.Sequential([\n", " keras.layers.Conv2D(\n", " 64, (5, 5), (2, 2),\n", " padding=\"same\",\n", " input_shape=[28, 28, 1]\n", " ),\n", "\n", " keras.layers.LeakyReLU(0.2),\n", " keras.layers.Dropout(0.3),\n", "\n", " keras.layers.Conv2D(\n", " 128, (5, 5), (2, 2),\n", " padding=\"same\"\n", " ),\n", "\n", " keras.layers.LeakyReLU(0.2),\n", " keras.layers.Dropout(0.3),\n", "\n", " keras.layers.Flatten(),\n", "\n", " keras.layers.Dense(1, activation='sigmoid')\n", "])\n", "\n", "discriminator.summary()" ], "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 455 }, "id": "4OTK3mDvrmOH", "outputId": "30934ae8-0bc7-4d5f-be72-8c8935d1eb5e" }, "execution_count": 6, "outputs": [ { "output_type": "stream", "name": "stderr", "text": [ "/usr/local/lib/python3.12/dist-packages/keras/src/layers/convolutional/base_conv.py:113: UserWarning: Do not pass an `input_shape`/`input_dim` argument to a layer. When using Sequential models, prefer using an `Input(shape)` object as the first layer in the model instead.\n", " super().__init__(activity_regularizer=activity_regularizer, **kwargs)\n" ] }, { "output_type": "display_data", "data": { "text/plain": [ "\u001b[1mModel: \"sequential_1\"\u001b[0m\n" ], "text/html": [ "
Model: \"sequential_1\"\n",
              "
\n" ] }, "metadata": {} }, { "output_type": "display_data", "data": { "text/plain": [ "┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓\n", "┃\u001b[1m \u001b[0m\u001b[1mLayer (type) \u001b[0m\u001b[1m \u001b[0m┃\u001b[1m \u001b[0m\u001b[1mOutput Shape \u001b[0m\u001b[1m \u001b[0m┃\u001b[1m \u001b[0m\u001b[1m Param #\u001b[0m\u001b[1m \u001b[0m┃\n", "┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩\n", "│ conv2d (\u001b[38;5;33mConv2D\u001b[0m) │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m14\u001b[0m, \u001b[38;5;34m14\u001b[0m, \u001b[38;5;34m64\u001b[0m) │ \u001b[38;5;34m1,664\u001b[0m │\n", "├─────────────────────────────────┼────────────────────────┼───────────────┤\n", "│ leaky_re_lu (\u001b[38;5;33mLeakyReLU\u001b[0m) │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m14\u001b[0m, \u001b[38;5;34m14\u001b[0m, \u001b[38;5;34m64\u001b[0m) │ \u001b[38;5;34m0\u001b[0m │\n", "├─────────────────────────────────┼────────────────────────┼───────────────┤\n", "│ dropout (\u001b[38;5;33mDropout\u001b[0m) │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m14\u001b[0m, \u001b[38;5;34m14\u001b[0m, \u001b[38;5;34m64\u001b[0m) │ \u001b[38;5;34m0\u001b[0m │\n", "├─────────────────────────────────┼────────────────────────┼───────────────┤\n", "│ conv2d_1 (\u001b[38;5;33mConv2D\u001b[0m) │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m7\u001b[0m, \u001b[38;5;34m7\u001b[0m, \u001b[38;5;34m128\u001b[0m) │ \u001b[38;5;34m204,928\u001b[0m │\n", "├─────────────────────────────────┼────────────────────────┼───────────────┤\n", "│ leaky_re_lu_1 (\u001b[38;5;33mLeakyReLU\u001b[0m) │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m7\u001b[0m, \u001b[38;5;34m7\u001b[0m, \u001b[38;5;34m128\u001b[0m) │ \u001b[38;5;34m0\u001b[0m │\n", "├─────────────────────────────────┼────────────────────────┼───────────────┤\n", "│ dropout_1 (\u001b[38;5;33mDropout\u001b[0m) │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m7\u001b[0m, \u001b[38;5;34m7\u001b[0m, \u001b[38;5;34m128\u001b[0m) │ \u001b[38;5;34m0\u001b[0m │\n", "├─────────────────────────────────┼────────────────────────┼───────────────┤\n", "│ flatten (\u001b[38;5;33mFlatten\u001b[0m) │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m6272\u001b[0m) │ \u001b[38;5;34m0\u001b[0m │\n", "├─────────────────────────────────┼────────────────────────┼───────────────┤\n", "│ dense_1 (\u001b[38;5;33mDense\u001b[0m) │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m1\u001b[0m) │ \u001b[38;5;34m6,273\u001b[0m │\n", "└─────────────────────────────────┴────────────────────────┴───────────────┘\n" ], "text/html": [ "
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓\n",
              "┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃\n",
              "┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩\n",
              "│ conv2d (Conv2D)                 │ (None, 14, 14, 64)     │         1,664 │\n",
              "├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
              "│ leaky_re_lu (LeakyReLU)         │ (None, 14, 14, 64)     │             0 │\n",
              "├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
              "│ dropout (Dropout)               │ (None, 14, 14, 64)     │             0 │\n",
              "├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
              "│ conv2d_1 (Conv2D)               │ (None, 7, 7, 128)      │       204,928 │\n",
              "├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
              "│ leaky_re_lu_1 (LeakyReLU)       │ (None, 7, 7, 128)      │             0 │\n",
              "├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
              "│ dropout_1 (Dropout)             │ (None, 7, 7, 128)      │             0 │\n",
              "├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
              "│ flatten (Flatten)               │ (None, 6272)           │             0 │\n",
              "├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
              "│ dense_1 (Dense)                 │ (None, 1)              │         6,273 │\n",
              "└─────────────────────────────────┴────────────────────────┴───────────────┘\n",
              "
\n" ] }, "metadata": {} }, { "output_type": "display_data", "data": { "text/plain": [ "\u001b[1m Total params: \u001b[0m\u001b[38;5;34m212,865\u001b[0m (831.50 KB)\n" ], "text/html": [ "
 Total params: 212,865 (831.50 KB)\n",
              "
\n" ] }, "metadata": {} }, { "output_type": "display_data", "data": { "text/plain": [ "\u001b[1m Trainable params: \u001b[0m\u001b[38;5;34m212,865\u001b[0m (831.50 KB)\n" ], "text/html": [ "
 Trainable params: 212,865 (831.50 KB)\n",
              "
\n" ] }, "metadata": {} }, { "output_type": "display_data", "data": { "text/plain": [ "\u001b[1m Non-trainable params: \u001b[0m\u001b[38;5;34m0\u001b[0m (0.00 B)\n" ], "text/html": [ "
 Non-trainable params: 0 (0.00 B)\n",
              "
\n" ] }, "metadata": {} } ] }, { "cell_type": "code", "source": [ "discriminator.compile(\n", " loss=\"binary_crossentropy\",\n", " optimizer=\"adam\"\n", ")\n", "\n", "discriminator.trainable = False\n", "\n", "gan = keras.models.Sequential([\n", " generator,\n", " discriminator\n", "])\n", "\n", "gan.compile(\n", " loss=\"binary_crossentropy\",\n", " optimizer=\"adam\"\n", ")" ], "metadata": { "id": "O0rsmBt2rmQW" }, "execution_count": 7, "outputs": [] }, { "cell_type": "code", "source": [ "seed = tf.random.normal(shape=[batch_size, 100])\n", "\n", "def train_dcgan(gan, dataset, batch_size, num_features, epochs=5):\n", "\n", " generator, discriminator = gan.layers\n", "\n", " for epoch in tqdm(range(epochs)):\n", "\n", " print()\n", " print(\"Epoch {}/{}\".format(epoch + 1, epochs))\n", "\n", " for X_batch in dataset:\n", "\n", " noise = tf.random.normal(\n", " shape=[batch_size, num_features]\n", " )\n", "\n", " generated_images = generator(noise)\n", "\n", " X_fake_and_real = tf.concat(\n", " [generated_images, X_batch],\n", " axis=0\n", " )\n", "\n", " y1 = tf.constant(\n", " [[0.]] * batch_size +\n", " [[1.]] * batch_size\n", " )\n", "\n", " discriminator.trainable = True\n", " discriminator.train_on_batch(\n", " X_fake_and_real,\n", " y1\n", " )\n", "\n", " noise = tf.random.normal(\n", " shape=[batch_size, num_features]\n", " )\n", "\n", " y2 = tf.constant([[1.]] * batch_size)\n", "\n", " discriminator.trainable = False\n", "\n", " gan.train_on_batch(\n", " noise,\n", " y2\n", " )\n", "\n", " generate_and_save_images(\n", " generator,\n", " epoch + 1,\n", " seed\n", " )\n", "\n", " generate_and_save_images(\n", " generator,\n", " epochs,\n", " seed\n", " )" ], "metadata": { "id": "e1LzY0P3xhON" }, "execution_count": 8, "outputs": [] }, { "cell_type": "code", "source": [ "def generate_and_save_images(model, epoch, test_input):\n", "\n", " predictions = model(\n", " test_input,\n", " training=False\n", " )\n", "\n", " fig = plt.figure(figsize=(10, 10))\n", "\n", " for i in range(25):\n", "\n", " plt.subplot(5, 5, i + 1)\n", "\n", " plt.imshow(\n", " predictions[i, :, :, 0] * 127.5 + 127.5,\n", " cmap='binary'\n", " )\n", "\n", " plt.axis('off')\n", "\n", " plt.savefig(\n", " 'image_epoch_{:04d}.png'.format(epoch)\n", " )" ], "metadata": { "id": "j80we_QrxhQP" }, "execution_count": 9, "outputs": [] }, { "cell_type": "code", "source": [ "x_train_dcgan = x_train.reshape(\n", " -1, 28, 28, 1\n", ") * 2. - 1.\n", "\n", "dataset = create_batch(x_train_dcgan)\n", "\n", "train_dcgan(\n", " gan,\n", " dataset,\n", " batch_size,\n", " num_features,\n", " epochs=10\n", ")" ], "metadata": { "id": "SJzVTEiPxhSn" }, "execution_count": null, "outputs": [] }, { "cell_type": "code", "source": [ "import imageio\n", "import glob\n", "\n", "anim_file = 'dcgan_results.gif'\n", "\n", "with imageio.get_writer(anim_file, mode='I') as writer:\n", "\n", " filenames = glob.glob('image*.png')\n", " filenames = sorted(filenames)\n", "\n", " last = -1\n", "\n", " for i, filename in enumerate(filenames):\n", "\n", " frame = 2 * (i)\n", "\n", " if round(frame) > round(last):\n", " last = frame\n", " else:\n", " continue\n", "\n", " image = imageio.imread(filename)\n", " writer.append_data(image)\n", "\n", " image = imageio.imread(filename)\n", " writer.append_data(image)\n", "\n", "display.Image(filename=anim_file)" ], "metadata": { "id": "ZCxk2A-JxhUo" }, "execution_count": null, "outputs": [] }, { "cell_type": "code", "source": [], "metadata": { "id": "ErknNhUpxhWk" }, "execution_count": null, "outputs": [] } ] }