model-mlp-tf.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "d3a1a0e1",
   "metadata": {},
   "outputs": [],
   "source": [
    "from matplotlib import pyplot as plt\n",
    "import tensorflow_datasets\n",
    "import tensorflow as tf\n",
    "import numpy as np\n",
    "import os"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "6a5cda32",
   "metadata": {},
   "outputs": [],
   "source": [
    "directory = 'models'\n",
    "if not os.path.exists(directory):\n",
    "    os.mkdir(directory)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "a436fb70",
   "metadata": {},
   "outputs": [],
   "source": [
    "def normilize(image, label):\n",
    "    return tf.cast(image, tf.float32) / 256, label"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "886005e8",
   "metadata": {},
   "outputs": [],
   "source": [
    "(train, test), info = tensorflow_datasets.load('mnist', split=['train', 'test'], as_supervised=True, with_info=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "a5ac7036",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'train': <SplitInfo num_examples=60000, num_shards=1>,\n",
       " 'test': <SplitInfo num_examples=10000, num_shards=1>}"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "info.splits"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "ef429428",
   "metadata": {},
   "outputs": [],
   "source": [
    "train = train.map(normilize)\n",
    "train = train.cache()\n",
    "train = train.shuffle(info.splits['train'].num_examples)\n",
    "train = train.batch(256)\n",
    "train = train.prefetch(tf.data.experimental.AUTOTUNE)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "7b2581f0",
   "metadata": {},
   "outputs": [],
   "source": [
    "test = test.map(normilize)\n",
    "test = test.cache()\n",
    "test = test.batch(256)\n",
    "test = test.prefetch(tf.data.experimental.AUTOTUNE)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "2994ddc9",
   "metadata": {},
   "outputs": [],
   "source": [
    "model = tf.keras.models.Sequential([\n",
    "    tf.keras.layers.Flatten(input_shape=(28, 28)),\n",
    "    tf.keras.layers.Dropout(.15),\n",
    "    tf.keras.layers.Dense(128, activation='linear'),\n",
    "    tf.keras.layers.LeakyReLU(),\n",
    "    tf.keras.layers.Dropout(.1),\n",
    "    tf.keras.layers.Dense(64, activation='relu'),\n",
    "    tf.keras.layers.Dense(32, activation='relu'),\n",
    "    tf.keras.layers.Dense(16, activation='relu'),\n",
    "    tf.keras.layers.Dense(10, activation='softmax')\n",
    "])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "d10c295d",
   "metadata": {},
   "outputs": [],
   "source": [
    "model.compile(\n",
    "    optimizer = tf.keras.optimizers.Adam(.001),\n",
    "    loss = tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True),\n",
    "    metrics = [tf.keras.metrics.SparseCategoricalAccuracy()]\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "ca2887f6",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/15\n",
      "235/235 [==============================] - 7s 14ms/step - loss: 0.6131 - sparse_categorical_accuracy: 0.8143 - val_loss: 0.2307 - val_sparse_categorical_accuracy: 0.9353\n",
      "Epoch 2/15\n",
      "235/235 [==============================] - 2s 9ms/step - loss: 0.2456 - sparse_categorical_accuracy: 0.9279 - val_loss: 0.1648 - val_sparse_categorical_accuracy: 0.9528\n",
      "Epoch 3/15\n",
      "235/235 [==============================] - 2s 9ms/step - loss: 0.1913 - sparse_categorical_accuracy: 0.9433 - val_loss: 0.1442 - val_sparse_categorical_accuracy: 0.9578\n",
      "Epoch 4/15\n",
      "235/235 [==============================] - 2s 8ms/step - loss: 0.1613 - sparse_categorical_accuracy: 0.9517 - val_loss: 0.1196 - val_sparse_categorical_accuracy: 0.9650\n",
      "Epoch 5/15\n",
      "235/235 [==============================] - 2s 8ms/step - loss: 0.1402 - sparse_categorical_accuracy: 0.9572 - val_loss: 0.1142 - val_sparse_categorical_accuracy: 0.9662\n",
      "Epoch 6/15\n",
      "235/235 [==============================] - 3s 11ms/step - loss: 0.1269 - sparse_categorical_accuracy: 0.9620 - val_loss: 0.1061 - val_sparse_categorical_accuracy: 0.9682\n",
      "Epoch 7/15\n",
      "235/235 [==============================] - 3s 11ms/step - loss: 0.1124 - sparse_categorical_accuracy: 0.9656 - val_loss: 0.0996 - val_sparse_categorical_accuracy: 0.9705\n",
      "Epoch 8/15\n",
      "235/235 [==============================] - 2s 9ms/step - loss: 0.1056 - sparse_categorical_accuracy: 0.9674 - val_loss: 0.0976 - val_sparse_categorical_accuracy: 0.9734\n",
      "Epoch 9/15\n",
      "235/235 [==============================] - 2s 8ms/step - loss: 0.0983 - sparse_categorical_accuracy: 0.9688 - val_loss: 0.0924 - val_sparse_categorical_accuracy: 0.9738\n",
      "Epoch 10/15\n",
      "235/235 [==============================] - 2s 9ms/step - loss: 0.0904 - sparse_categorical_accuracy: 0.9716 - val_loss: 0.0825 - val_sparse_categorical_accuracy: 0.9755\n",
      "Epoch 11/15\n",
      "235/235 [==============================] - 2s 8ms/step - loss: 0.0854 - sparse_categorical_accuracy: 0.9727 - val_loss: 0.0835 - val_sparse_categorical_accuracy: 0.9736\n",
      "Epoch 12/15\n",
      "235/235 [==============================] - 2s 8ms/step - loss: 0.0827 - sparse_categorical_accuracy: 0.9736 - val_loss: 0.0831 - val_sparse_categorical_accuracy: 0.9758\n",
      "Epoch 13/15\n",
      "235/235 [==============================] - 2s 9ms/step - loss: 0.0788 - sparse_categorical_accuracy: 0.9748 - val_loss: 0.0845 - val_sparse_categorical_accuracy: 0.9753\n",
      "Epoch 14/15\n",
      "235/235 [==============================] - 2s 8ms/step - loss: 0.0744 - sparse_categorical_accuracy: 0.9762 - val_loss: 0.0815 - val_sparse_categorical_accuracy: 0.9764\n",
      "Epoch 15/15\n",
      "235/235 [==============================] - 2s 8ms/step - loss: 0.0704 - sparse_categorical_accuracy: 0.9767 - val_loss: 0.0820 - val_sparse_categorical_accuracy: 0.9763\n"
     ]
    }
   ],
   "source": [
    "history = model.fit(train, epochs=15, validation_data=test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "16ed973f",
   "metadata": {},
   "outputs": [],
   "source": [
    "model.save(os.path.join(directory, \"MLP.keras\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "2d65425e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(history.history['sparse_categorical_accuracy'], color='green', label='Train')\n",
    "plt.plot(history.history['val_sparse_categorical_accuracy'], color='red', label='Validation')\n",
    "plt.title('Accuracy')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}