CIFAR10.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train mode : epoch =     1, loss = 1.909e-02\n",
      "trained model at current step was saved.\n",
      "train mode : epoch =     2, loss = 1.719e-02\n",
      "trained model at current step was saved.\n",
      "train mode : epoch =     3, loss = 1.637e-02\n",
      "trained model at current step was saved.\n",
      "train mode : epoch =     4, loss = 1.580e-02\n",
      "trained model at current step was saved.\n",
      "train mode : epoch =     5, loss = 1.540e-02\n",
      "trained model at current step was saved.\n",
      "train mode : epoch =     6, loss = 1.508e-02\n",
      "trained model at current step was saved.\n",
      "train mode : epoch =     7, loss = 1.478e-02\n",
      "trained model at current step was saved.\n",
      "train mode : epoch =     8, loss = 1.450e-02\n",
      "trained model at current step was saved.\n",
      "train mode : epoch =     9, loss = 1.424e-02\n",
      "trained model at current step was saved.\n",
      "train mode : epoch =    10, loss = 1.403e-02\n",
      "trained model at current step was saved.\n",
      "time : 43.416sec\n",
      "test mode : loss = 1.474e-02, eval = 4.800e-01\n"
     ]
    }
   ],
   "source": [
    "import torch\n",
    "import torch.nn as nn\n",
    "\n",
    "from framework import DataManager\n",
    "from framework import Network\n",
    "from framework import Timer\n",
    "\n",
    "from usr import CIFAR10\n",
    "from usr import Linear\n",
    "from usr import CorrectRate\n",
    "\n",
    "if __name__ == \"__main__\":\n",
    "  # define variables\n",
    "  device = \"cuda\"\n",
    "  device_ids = [0]\n",
    "  batch_size = 100\n",
    "  num_workers = 1\n",
    "  num_epochs = 10\n",
    "  learning_rate = 0.001\n",
    "\n",
    "  # create data set\n",
    "  cifar10_train = CIFAR10.CIFAR10(\"./data/cifar10/data_batch\")\n",
    "  cifar10_test = CIFAR10.CIFAR10(\"./data/cifar10/test_batch\")\n",
    "  \n",
    "  # create data manager\n",
    "  data_manager = DataManager.DataManager(train_data=cifar10_train, test_data=cifar10_test, batch_size=batch_size, num_workers=num_workers)\n",
    "\n",
    "  # define user-defined network\n",
    "  model_kwargs = {\"in_units\" : 3*32*32, \"mid_units\" : [100, 100, 100, 100], \"out_units\" : 10}\n",
    "  model = Linear.Linear(**model_kwargs)\n",
    "\n",
    "  # define user-defined criterion\n",
    "  criterion = nn.CrossEntropyLoss()\n",
    "\n",
    "  # define user-defined optimizer\n",
    "  optimizer = torch.optim.Adam(model.parameters(), lr=learning_rate)\n",
    "\n",
    "  # define user-defined evaluator\n",
    "  evaluator = CorrectRate.CorrectRate()\n",
    "\n",
    "  # define model\n",
    "  net = Network.Network(model=model, criterion=criterion, evaluator=evaluator, optimizer=optimizer,\n",
    "                        data_manager=data_manager, save_checkpoint=True,\n",
    "                        device=device, device_ids=device_ids, non_blocking=True)\n",
    "  \n",
    "  # training\n",
    "  with Timer.Timer() as timer:\n",
    "    for _ in range(num_epochs):\n",
    "      net.train()\n",
    "  \n",
    "  # test\n",
    "  net.test()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "trained model was loaded.\n",
      "test mode : loss = 1.474e-02, eval = 4.800e-01\n"
     ]
    }
   ],
   "source": [
    "import torch\n",
    "import torch.nn as nn\n",
    "\n",
    "from framework import DataManager\n",
    "from framework import Network\n",
    "from framework import Timer\n",
    "\n",
    "from usr import CIFAR10\n",
    "from usr import Linear\n",
    "from usr import CorrectRate\n",
    "\n",
    "if __name__ == \"__main__\":\n",
    "  # define variables\n",
    "  device = \"cuda\"\n",
    "  device_ids = [0]\n",
    "  batch_size = 100\n",
    "  num_workers = 1\n",
    "  num_epochs = 10\n",
    "  learning_rate = 0.001\n",
    "\n",
    "  # create data set\n",
    "  cifar10_train = CIFAR10.CIFAR10(\"./data/cifar10/data_batch\")\n",
    "  cifar10_test = CIFAR10.CIFAR10(\"./data/cifar10/test_batch\")\n",
    "  \n",
    "  # create data manager\n",
    "  data_manager = DataManager.DataManager(train_data=cifar10_train, test_data=cifar10_test, batch_size=batch_size, num_workers=num_workers)\n",
    "\n",
    "  # define user-defined network\n",
    "  model_kwargs = {\"in_units\" : 3*32*32, \"mid_units\" : [100, 100, 100, 100], \"out_units\" : 10}\n",
    "  model = Linear.Linear(**model_kwargs)\n",
    "\n",
    "  # define user-defined criterion\n",
    "  criterion = nn.CrossEntropyLoss()\n",
    "\n",
    "  # define user-defined optimizer\n",
    "  optimizer = torch.optim.Adam(model.parameters(), lr=learning_rate)\n",
    "\n",
    "  # define user-defined evaluator\n",
    "  evaluator = CorrectRate.CorrectRate()\n",
    "\n",
    "  # define model\n",
    "  net = Network.Network(model=model, criterion=criterion, evaluator=evaluator, optimizer=optimizer,\n",
    "                        data_manager=data_manager, save_checkpoint=True,\n",
    "                        device=device, device_ids=device_ids, non_blocking=True)\n",
    "\n",
    "  # load trained model\n",
    "  net.load()\n",
    "\n",
    "  # test\n",
    "  net.test()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3.10.6 64-bit",
   "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": "3.10.6"
  },
  "orig_nbformat": 4,
  "vscode": {
   "interpreter": {
    "hash": "916dbcbb3f70747c44a77c7bcd40155683ae19c65e1c03b4aa3499c5328201f1"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}