vgg13.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "# PyTorch for CIFAR10\n",
    "We test VGG classifier.\n",
    "First init some basic environment."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "\n",
    "%matplotlib inline\n",
    "plt.rcParams['figure.figsize'] = (10.0, 8.0) # set default size of plots\n",
    "plt.rcParams['image.interpolation'] = 'nearest'\n",
    "plt.rcParams['image.cmap'] = 'gray'\n",
    "\n",
    "%load_ext autoreload\n",
    "%autoreload 2"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# VGG13 classifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_VGG(\n",
      "  (layers): Sequential(\n",
      "    (0): Conv2d(3, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
      "    (1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    (2): ReLU(inplace)\n",
      "    (3): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
      "    (4): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    (5): ReLU(inplace)\n",
      "    (6): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
      "    (7): Conv2d(64, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
      "    (8): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    (9): ReLU(inplace)\n",
      "    (10): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
      "    (11): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    (12): ReLU(inplace)\n",
      "    (13): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
      "    (14): Conv2d(128, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
      "    (15): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    (16): ReLU(inplace)\n",
      "    (17): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
      "    (18): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    (19): ReLU(inplace)\n",
      "    (20): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
      "    (21): Conv2d(256, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
      "    (22): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    (23): ReLU(inplace)\n",
      "    (24): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
      "    (25): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    (26): ReLU(inplace)\n",
      "    (27): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
      "    (28): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
      "    (29): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    (30): ReLU(inplace)\n",
      "    (31): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
      "    (32): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    (33): ReLU(inplace)\n",
      "    (34): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
      "  )\n",
      "  (fc1): Linear(in_features=512, out_features=10, bias=True)\n",
      ")\n"
     ]
    }
   ],
   "source": [
    "import torch.optim as optim\n",
    "import torch.nn as nn\n",
    "from cifar10.tnt_solver import *\n",
    "from cifar10.classifiers.vgg import VGG13\n",
    "\n",
    "model = VGG13()\n",
    "print(model)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# train"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Files already downloaded and verified\n",
      "Files already downloaded and verified\n",
      "Files already downloaded and verified\n",
      "Epoch: 1/150, lr:1.00e-01\n",
      "100%|██████████| 625/625 [00:42<00:00, 14.68it/s, loss:2.0754, acc:19.7225%]\n",
      "Val loss: 1.8478, accuracy: 26.08%\n",
      "Epoch: 2/150, lr:1.00e-01\n",
      "100%|██████████| 625/625 [00:42<00:00, 14.66it/s, loss:1.6014, acc:38.3550%]\n",
      "Val loss: 1.2857, accuracy: 53.68%\n",
      "Epoch: 3/150, lr:1.00e-01\n",
      "100%|██████████| 625/625 [00:42<00:00, 14.60it/s, loss:1.2014, acc:56.8200%]\n",
      "Val loss: 1.1871, accuracy: 58.68%\n",
      "Epoch: 4/150, lr:1.00e-01\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.44it/s, loss:1.0056, acc:64.4350%]\n",
      "Val loss: 1.2032, accuracy: 60.23%\n",
      "Epoch: 5/150, lr:1.00e-01\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.40it/s, loss:0.8826, acc:69.3825%]\n",
      "Val loss: 1.5145, accuracy: 55.32%\n",
      "Epoch: 6/150, lr:1.00e-01\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.39it/s, loss:0.8006, acc:72.7175%]\n",
      "Val loss: 1.0807, accuracy: 64.48%\n",
      "Epoch: 7/150, lr:1.00e-01\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.35it/s, loss:0.7472, acc:74.6675%]\n",
      "Val loss: 0.8349, accuracy: 71.93%\n",
      "Epoch: 8/150, lr:1.00e-01\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.44it/s, loss:0.7112, acc:75.8850%]\n",
      "Val loss: 0.9853, accuracy: 68.07%\n",
      "Epoch: 9/150, lr:1.00e-01\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.41it/s, loss:0.6817, acc:76.9500%]\n",
      "Val loss: 0.9297, accuracy: 71.32%\n",
      "Epoch: 10/150, lr:1.00e-01\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.36it/s, loss:0.6628, acc:77.6000%]\n",
      "Val loss: 0.7067, accuracy: 76.64%\n",
      "Epoch: 11/150, lr:1.00e-01\n",
      "100%|██████████| 625/625 [01:07<00:00,  9.31it/s, loss:0.6517, acc:78.0550%]\n",
      "Val loss: 0.7087, accuracy: 76.52%\n",
      "Epoch: 12/150, lr:1.00e-01\n",
      "100%|██████████| 625/625 [01:05<00:00,  9.56it/s, loss:0.6306, acc:78.7475%]\n",
      "Val loss: 0.7150, accuracy: 76.04%\n",
      "Epoch: 13/150, lr:5.00e-02\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.37it/s, loss:0.4887, acc:83.2900%]\n",
      "Val loss: 0.4964, accuracy: 83.51%\n",
      "Epoch: 14/150, lr:5.00e-02\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.37it/s, loss:0.4755, acc:83.8600%]\n",
      "Val loss: 0.5728, accuracy: 80.63%\n",
      "Epoch: 15/150, lr:5.00e-02\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.38it/s, loss:0.4649, acc:84.3125%]\n",
      "Val loss: 0.5669, accuracy: 81.16%\n",
      "Epoch: 16/150, lr:5.00e-02\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.37it/s, loss:0.4691, acc:83.9750%]\n",
      "Val loss: 0.5350, accuracy: 82.04%\n",
      "Epoch: 17/150, lr:5.00e-02\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.37it/s, loss:0.4609, acc:84.2700%]\n",
      "Val loss: 0.6125, accuracy: 80.06%\n",
      "Epoch: 18/150, lr:5.00e-02\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.41it/s, loss:0.4497, acc:84.7350%]\n",
      "Val loss: 0.5658, accuracy: 80.51%\n",
      "Epoch: 19/150, lr:5.00e-02\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.42it/s, loss:0.4469, acc:84.6600%]\n",
      "Val loss: 0.6602, accuracy: 78.79%\n",
      "Epoch: 20/150, lr:5.00e-02\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.46it/s, loss:0.4408, acc:85.2050%]\n",
      "Val loss: 0.4746, accuracy: 83.98%\n",
      "Epoch: 21/150, lr:5.00e-02\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.38it/s, loss:0.4345, acc:85.2450%]\n",
      "Val loss: 0.5979, accuracy: 79.88%\n",
      "Epoch: 22/150, lr:5.00e-02\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.46it/s, loss:0.4281, acc:85.6675%]\n",
      "Val loss: 0.5372, accuracy: 82.02%\n",
      "Epoch: 23/150, lr:5.00e-02\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.43it/s, loss:0.4173, acc:85.8250%]\n",
      "Val loss: 0.5179, accuracy: 82.59%\n",
      "Epoch: 24/150, lr:2.50e-02\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.44it/s, loss:0.3067, acc:89.4725%]\n",
      "Val loss: 0.4303, accuracy: 85.63%\n",
      "Epoch: 25/150, lr:2.50e-02\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.42it/s, loss:0.2867, acc:90.1750%]\n",
      "Val loss: 0.3746, accuracy: 87.60%\n",
      "Epoch: 26/150, lr:2.50e-02\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.45it/s, loss:0.2789, acc:90.4425%]\n",
      "Val loss: 0.4317, accuracy: 85.75%\n",
      "Epoch: 27/150, lr:2.50e-02\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.41it/s, loss:0.2819, acc:90.3650%]\n",
      "Val loss: 0.3995, accuracy: 86.78%\n",
      "Epoch: 28/150, lr:2.50e-02\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.37it/s, loss:0.2819, acc:90.4650%]\n",
      "Val loss: 0.3982, accuracy: 86.75%\n",
      "Epoch: 29/150, lr:2.50e-02\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.41it/s, loss:0.2788, acc:90.5050%]\n",
      "Val loss: 0.3646, accuracy: 87.79%\n",
      "Epoch: 30/150, lr:2.50e-02\n",
      "100%|██████████| 625/625 [01:05<00:00,  9.52it/s, loss:0.2803, acc:90.2900%]\n",
      "Val loss: 0.3780, accuracy: 87.59%\n",
      "Epoch: 31/150, lr:2.50e-02\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.41it/s, loss:0.2776, acc:90.4000%]\n",
      "Val loss: 0.4447, accuracy: 85.29%\n",
      "Epoch: 32/150, lr:2.50e-02\n",
      "100%|██████████| 625/625 [01:05<00:00,  9.48it/s, loss:0.2765, acc:90.5500%]\n",
      "Val loss: 0.3679, accuracy: 87.88%\n",
      "Epoch: 33/150, lr:2.50e-02\n",
      "100%|██████████| 625/625 [01:07<00:00,  9.31it/s, loss:0.2690, acc:90.7800%]\n",
      "Val loss: 0.4006, accuracy: 86.88%\n",
      "Epoch: 34/150, lr:2.50e-02\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.42it/s, loss:0.2663, acc:90.9075%]\n",
      "Val loss: 0.4103, accuracy: 85.89%\n",
      "Epoch: 35/150, lr:1.25e-02\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.35it/s, loss:0.1718, acc:94.1100%]\n",
      "Val loss: 0.3488, accuracy: 89.11%\n",
      "Epoch: 36/150, lr:1.25e-02\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.46it/s, loss:0.1558, acc:94.6225%]\n",
      "Val loss: 0.3280, accuracy: 89.67%\n",
      "Epoch: 37/150, lr:1.25e-02\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.42it/s, loss:0.1501, acc:94.8850%]\n",
      "Val loss: 0.3161, accuracy: 89.90%\n",
      "Epoch: 38/150, lr:1.25e-02\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.45it/s, loss:0.1547, acc:94.6125%]\n",
      "Val loss: 0.3462, accuracy: 89.19%\n",
      "Epoch: 39/150, lr:1.25e-02\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.43it/s, loss:0.1563, acc:94.5500%]\n",
      "Val loss: 0.3225, accuracy: 89.90%\n",
      "Epoch: 40/150, lr:1.25e-02\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.43it/s, loss:0.1492, acc:94.7875%]\n",
      "Val loss: 0.4395, accuracy: 86.91%\n",
      "Epoch: 41/150, lr:1.25e-02\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.37it/s, loss:0.1486, acc:94.8875%]\n",
      "Val loss: 0.3438, accuracy: 89.44%\n",
      "Epoch: 42/150, lr:1.25e-02\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.42it/s, loss:0.1512, acc:94.7275%]\n",
      "Val loss: 0.3865, accuracy: 87.79%\n",
      "Epoch: 43/150, lr:1.25e-02\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.47it/s, loss:0.1517, acc:94.7575%]\n",
      "Val loss: 0.4413, accuracy: 86.56%\n",
      "Epoch: 44/150, lr:1.25e-02\n",
      "100%|██████████| 625/625 [01:05<00:00,  9.48it/s, loss:0.1516, acc:94.7125%]\n",
      "Val loss: 0.3539, accuracy: 88.78%\n",
      "Epoch: 45/150, lr:1.25e-02\n",
      "100%|██████████| 625/625 [01:07<00:00,  9.29it/s, loss:0.1506, acc:94.6975%]\n",
      "Val loss: 0.3810, accuracy: 88.52%\n",
      "Epoch: 46/150, lr:6.25e-03\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.45it/s, loss:0.0877, acc:96.9250%]\n",
      "Val loss: 0.2933, accuracy: 91.21%\n",
      "Epoch: 47/150, lr:6.25e-03\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.44it/s, loss:0.0714, acc:97.5225%]\n",
      "Val loss: 0.2980, accuracy: 91.55%\n",
      "Epoch: 48/150, lr:6.25e-03\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.42it/s, loss:0.0654, acc:97.7625%]\n",
      "Val loss: 0.3063, accuracy: 91.42%\n",
      "Epoch: 49/150, lr:6.25e-03\n",
      "100%|██████████| 625/625 [01:05<00:00,  9.53it/s, loss:0.0661, acc:97.7125%]\n",
      "Val loss: 0.3060, accuracy: 91.19%\n",
      "Epoch: 50/150, lr:6.25e-03\n",
      "100%|██████████| 625/625 [01:07<00:00,  9.31it/s, loss:0.0640, acc:97.7375%]\n",
      "Val loss: 0.3201, accuracy: 90.97%\n",
      "Epoch: 51/150, lr:6.25e-03\n",
      "100%|██████████| 625/625 [01:05<00:00,  9.55it/s, loss:0.0640, acc:97.8525%]\n",
      "Val loss: 0.2974, accuracy: 91.42%\n",
      "Epoch: 52/150, lr:6.25e-03\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.35it/s, loss:0.0627, acc:97.8100%]\n",
      "Val loss: 0.3218, accuracy: 90.88%\n",
      "Epoch: 53/150, lr:6.25e-03\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.47it/s, loss:0.0623, acc:97.8600%]\n",
      "Val loss: 0.3110, accuracy: 91.03%\n",
      "Epoch: 54/150, lr:6.25e-03\n",
      "100%|██████████| 625/625 [01:05<00:00,  9.49it/s, loss:0.0688, acc:97.6325%]\n",
      "Val loss: 0.3348, accuracy: 90.83%\n",
      "Epoch: 55/150, lr:6.25e-03\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.47it/s, loss:0.0639, acc:97.8125%]\n",
      "Val loss: 0.3470, accuracy: 90.49%\n",
      "Epoch: 56/150, lr:6.25e-03\n",
      "100%|██████████| 625/625 [01:05<00:00,  9.48it/s, loss:0.0653, acc:97.7950%]\n",
      "Val loss: 0.3705, accuracy: 89.76%\n",
      "Epoch: 57/150, lr:3.13e-03\n",
      "100%|██████████| 625/625 [01:05<00:00,  9.49it/s, loss:0.0359, acc:98.8425%]\n",
      "Val loss: 0.2877, accuracy: 91.94%\n",
      "Epoch: 58/150, lr:3.13e-03\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.47it/s, loss:0.0267, acc:99.1650%]\n",
      "Val loss: 0.3108, accuracy: 91.81%\n",
      "Epoch: 59/150, lr:3.13e-03\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "100%|██████████| 625/625 [01:06<00:00,  9.40it/s, loss:0.0260, acc:99.1150%]\n",
      "Val loss: 0.3018, accuracy: 91.97%\n",
      "Epoch: 60/150, lr:3.13e-03\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.39it/s, loss:0.0201, acc:99.4250%]\n",
      "Val loss: 0.2987, accuracy: 92.18%\n",
      "Epoch: 61/150, lr:3.13e-03\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.37it/s, loss:0.0183, acc:99.4175%]\n",
      "Val loss: 0.2893, accuracy: 92.35%\n",
      "Epoch: 62/150, lr:3.13e-03\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.34it/s, loss:0.0201, acc:99.3325%]\n",
      "Val loss: 0.3024, accuracy: 92.01%\n",
      "Epoch: 63/150, lr:3.13e-03\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.43it/s, loss:0.0171, acc:99.4800%]\n",
      "Val loss: 0.3020, accuracy: 92.06%\n",
      "Epoch: 64/150, lr:3.13e-03\n",
      "100%|██████████| 625/625 [01:05<00:00,  9.49it/s, loss:0.0184, acc:99.4250%]\n",
      "Val loss: 0.3110, accuracy: 92.03%\n",
      "Epoch: 65/150, lr:3.13e-03\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.47it/s, loss:0.0189, acc:99.3800%]\n",
      "Val loss: 0.3114, accuracy: 92.06%\n",
      "Epoch: 66/150, lr:3.13e-03\n",
      "100%|██████████| 625/625 [01:05<00:00,  9.47it/s, loss:0.0192, acc:99.3775%]\n",
      "Val loss: 0.3022, accuracy: 92.29%\n",
      "Epoch: 67/150, lr:3.13e-03\n",
      "100%|██████████| 625/625 [01:05<00:00,  9.53it/s, loss:0.0172, acc:99.4500%]\n",
      "Val loss: 0.3117, accuracy: 92.23%\n",
      "Epoch: 68/150, lr:1.56e-03\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.43it/s, loss:0.0107, acc:99.6975%]\n",
      "Val loss: 0.2981, accuracy: 92.44%\n",
      "Epoch: 69/150, lr:1.56e-03\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.37it/s, loss:0.0105, acc:99.6975%]\n",
      "Val loss: 0.3104, accuracy: 92.06%\n",
      "Epoch: 70/150, lr:1.56e-03\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.45it/s, loss:0.0090, acc:99.7550%]\n",
      "Val loss: 0.3039, accuracy: 92.47%\n",
      "Epoch: 71/150, lr:1.56e-03\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.41it/s, loss:0.0084, acc:99.7775%]\n",
      "Val loss: 0.2999, accuracy: 92.51%\n",
      "Epoch: 72/150, lr:1.56e-03\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.36it/s, loss:0.0075, acc:99.8025%]\n",
      "Val loss: 0.2979, accuracy: 92.68%\n",
      "Epoch: 73/150, lr:1.56e-03\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.47it/s, loss:0.0073, acc:99.8100%]\n",
      "Val loss: 0.3008, accuracy: 92.71%\n",
      "Epoch: 74/150, lr:1.56e-03\n",
      "100%|██████████| 625/625 [01:07<00:00,  9.29it/s, loss:0.0070, acc:99.7875%]\n",
      "Val loss: 0.2943, accuracy: 92.73%\n",
      "Epoch: 75/150, lr:1.56e-03\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.35it/s, loss:0.0069, acc:99.8025%]\n",
      "Val loss: 0.2936, accuracy: 92.73%\n",
      "Epoch: 76/150, lr:1.56e-03\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.42it/s, loss:0.0060, acc:99.8475%]\n",
      "Val loss: 0.3003, accuracy: 92.55%\n",
      "Epoch: 77/150, lr:1.56e-03\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.37it/s, loss:0.0060, acc:99.8750%]\n",
      "Val loss: 0.3035, accuracy: 92.84%\n",
      "Epoch: 78/150, lr:1.56e-03\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.39it/s, loss:0.0059, acc:99.8500%]\n",
      "Val loss: 0.2976, accuracy: 92.70%\n",
      "Epoch: 79/150, lr:7.81e-04\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.41it/s, loss:0.0045, acc:99.9075%]\n",
      "Val loss: 0.2999, accuracy: 92.89%\n",
      "Epoch: 80/150, lr:7.81e-04\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.35it/s, loss:0.0041, acc:99.9350%]\n",
      "Val loss: 0.2945, accuracy: 92.68%\n",
      "Epoch: 81/150, lr:7.81e-04\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.39it/s, loss:0.0045, acc:99.9075%]\n",
      "Val loss: 0.2964, accuracy: 92.84%\n",
      "Epoch: 82/150, lr:7.81e-04\n",
      "100%|██████████| 625/625 [01:05<00:00,  9.48it/s, loss:0.0038, acc:99.9175%]\n",
      "Val loss: 0.2978, accuracy: 92.83%\n",
      "Epoch: 83/150, lr:7.81e-04\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.36it/s, loss:0.0041, acc:99.9150%]\n",
      "Val loss: 0.2973, accuracy: 92.87%\n",
      "Epoch: 84/150, lr:7.81e-04\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.45it/s, loss:0.0035, acc:99.9400%]\n",
      "Val loss: 0.2940, accuracy: 92.86%\n",
      "Epoch: 85/150, lr:7.81e-04\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.41it/s, loss:0.0035, acc:99.9375%]\n",
      "Val loss: 0.2955, accuracy: 92.77%\n",
      "Epoch: 86/150, lr:7.81e-04\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.36it/s, loss:0.0033, acc:99.9475%]\n",
      "Val loss: 0.2965, accuracy: 92.86%\n",
      "Epoch: 87/150, lr:7.81e-04\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.46it/s, loss:0.0032, acc:99.9375%]\n",
      "Val loss: 0.2960, accuracy: 92.81%\n",
      "Epoch: 88/150, lr:7.81e-04\n",
      "100%|██████████| 625/625 [01:07<00:00,  9.31it/s, loss:0.0031, acc:99.9575%]\n",
      "Val loss: 0.2953, accuracy: 92.93%\n",
      "Epoch: 89/150, lr:7.81e-04\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.37it/s, loss:0.0030, acc:99.9525%]\n",
      "Val loss: 0.2970, accuracy: 92.83%\n",
      "Epoch: 90/150, lr:3.91e-04\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.34it/s, loss:0.0033, acc:99.9325%]\n",
      "Val loss: 0.2938, accuracy: 92.85%\n",
      "Epoch: 91/150, lr:3.91e-04\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.42it/s, loss:0.0030, acc:99.9400%]\n",
      "Val loss: 0.2942, accuracy: 92.98%\n",
      "Epoch: 92/150, lr:3.91e-04\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.40it/s, loss:0.0032, acc:99.9375%]\n",
      "Val loss: 0.2943, accuracy: 92.99%\n",
      "Epoch: 93/150, lr:3.91e-04\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.40it/s, loss:0.0028, acc:99.9550%]\n",
      "Val loss: 0.2950, accuracy: 92.95%\n",
      "Epoch: 94/150, lr:3.91e-04\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.45it/s, loss:0.0031, acc:99.9575%]\n",
      "Val loss: 0.2915, accuracy: 92.91%\n",
      "Epoch: 95/150, lr:3.91e-04\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.38it/s, loss:0.0029, acc:99.9425%]\n",
      "Val loss: 0.2930, accuracy: 92.87%\n",
      "Epoch: 96/150, lr:3.91e-04\n",
      "100%|██████████| 625/625 [01:07<00:00,  9.21it/s, loss:0.0026, acc:99.9625%]\n",
      "Val loss: 0.2947, accuracy: 92.96%\n",
      "Epoch: 97/150, lr:3.91e-04\n",
      "100%|██████████| 625/625 [01:07<00:00,  9.30it/s, loss:0.0026, acc:99.9500%]\n",
      "Val loss: 0.2903, accuracy: 92.92%\n",
      "Epoch: 98/150, lr:3.91e-04\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.41it/s, loss:0.0023, acc:99.9725%]\n",
      "Val loss: 0.2938, accuracy: 92.98%\n",
      "Epoch: 99/150, lr:3.91e-04\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.37it/s, loss:0.0027, acc:99.9625%]\n",
      "Val loss: 0.2910, accuracy: 92.89%\n",
      "Epoch: 100/150, lr:3.91e-04\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.36it/s, loss:0.0027, acc:99.9550%]\n",
      "Val loss: 0.2915, accuracy: 92.97%\n",
      "Epoch: 101/150, lr:1.95e-04\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.38it/s, loss:0.0025, acc:99.9700%]\n",
      "Val loss: 0.2885, accuracy: 92.92%\n",
      "Epoch: 102/150, lr:1.95e-04\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.40it/s, loss:0.0025, acc:99.9600%]\n",
      "Val loss: 0.2873, accuracy: 93.02%\n",
      "Epoch: 103/150, lr:1.95e-04\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.41it/s, loss:0.0025, acc:99.9625%] \n",
      "Val loss: 0.2882, accuracy: 92.97%\n",
      "Epoch: 104/150, lr:1.95e-04\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.43it/s, loss:0.0025, acc:99.9600%]\n",
      "Val loss: 0.2888, accuracy: 93.09%\n",
      "Epoch: 105/150, lr:1.95e-04\n",
      "100%|██████████| 625/625 [01:07<00:00,  9.33it/s, loss:0.0025, acc:99.9575%]\n",
      "Val loss: 0.2895, accuracy: 93.03%\n",
      "Epoch: 106/150, lr:1.95e-04\n",
      "100%|██████████| 625/625 [01:07<00:00,  9.25it/s, loss:0.0021, acc:99.9725%]\n",
      "Val loss: 0.2904, accuracy: 92.92%\n",
      "Epoch: 107/150, lr:1.95e-04\n",
      "100%|██████████| 625/625 [01:07<00:00,  9.19it/s, loss:0.0023, acc:99.9675%]\n",
      "Val loss: 0.2884, accuracy: 93.01%\n",
      "Epoch: 108/150, lr:1.95e-04\n",
      "100%|██████████| 625/625 [01:07<00:00,  9.26it/s, loss:0.0023, acc:99.9700%]\n",
      "Val loss: 0.2931, accuracy: 92.85%\n",
      "Epoch: 109/150, lr:1.95e-04\n",
      "100%|██████████| 625/625 [01:07<00:00,  9.20it/s, loss:0.0021, acc:99.9800%]\n",
      "Val loss: 0.2873, accuracy: 93.00%\n",
      "Epoch: 110/150, lr:1.95e-04\n",
      "100%|██████████| 625/625 [01:07<00:00,  9.27it/s, loss:0.0025, acc:99.9550%]\n",
      "Val loss: 0.2872, accuracy: 93.01%\n",
      "Epoch: 111/150, lr:1.95e-04\n",
      "100%|██████████| 625/625 [01:08<00:00,  9.16it/s, loss:0.0023, acc:99.9725%] \n",
      "Val loss: 0.2899, accuracy: 93.05%\n",
      "Epoch: 112/150, lr:9.77e-05\n",
      "100%|██████████| 625/625 [01:07<00:00,  9.24it/s, loss:0.0024, acc:99.9675%]\n",
      "Val loss: 0.2903, accuracy: 93.00%\n",
      "Epoch: 113/150, lr:9.77e-05\n",
      "100%|██████████| 625/625 [01:07<00:00,  9.24it/s, loss:0.0025, acc:99.9525%]\n",
      "Val loss: 0.2892, accuracy: 93.00%\n",
      "Epoch: 114/150, lr:9.77e-05\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.36it/s, loss:0.0022, acc:99.9750%] \n",
      "Val loss: 0.2890, accuracy: 93.13%\n",
      "Epoch: 115/150, lr:9.77e-05\n",
      "100%|██████████| 625/625 [01:07<00:00,  9.31it/s, loss:0.0022, acc:99.9725%]\n",
      "Val loss: 0.2875, accuracy: 92.99%\n",
      "Epoch: 116/150, lr:9.77e-05\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.37it/s, loss:0.0024, acc:99.9625%]\n",
      "Val loss: 0.2924, accuracy: 93.02%\n",
      "Epoch: 117/150, lr:9.77e-05\n",
      "100%|██████████| 625/625 [01:07<00:00,  9.32it/s, loss:0.0022, acc:99.9725%] \n",
      "Val loss: 0.2902, accuracy: 93.05%\n",
      "Epoch: 118/150, lr:9.77e-05\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "100%|██████████| 625/625 [01:06<00:00,  9.41it/s, loss:0.0023, acc:99.9650%]\n",
      "Val loss: 0.2894, accuracy: 93.08%\n",
      "Epoch: 119/150, lr:9.77e-05\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.45it/s, loss:0.0022, acc:99.9800%]\n",
      "Val loss: 0.2889, accuracy: 93.00%\n",
      "Epoch: 120/150, lr:9.77e-05\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.37it/s, loss:0.0023, acc:99.9725%]\n",
      "Val loss: 0.2895, accuracy: 93.04%\n",
      "Epoch: 121/150, lr:9.77e-05\n",
      "100%|██████████| 625/625 [01:07<00:00,  9.22it/s, loss:0.0024, acc:99.9750%]\n",
      "Val loss: 0.2888, accuracy: 92.95%\n",
      "Epoch: 122/150, lr:9.77e-05\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.36it/s, loss:0.0021, acc:99.9775%]\n",
      "Val loss: 0.2907, accuracy: 93.02%\n",
      "Epoch: 123/150, lr:4.88e-05\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.37it/s, loss:0.0020, acc:99.9800%]\n",
      "Val loss: 0.2869, accuracy: 93.06%\n",
      "Epoch: 124/150, lr:4.88e-05\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.38it/s, loss:0.0022, acc:99.9675%] \n",
      "Val loss: 0.2898, accuracy: 93.00%\n",
      "Epoch: 125/150, lr:4.88e-05\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.39it/s, loss:0.0021, acc:99.9725%]\n",
      "Val loss: 0.2870, accuracy: 92.99%\n",
      "Epoch: 126/150, lr:4.88e-05\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.38it/s, loss:0.0021, acc:99.9750%] \n",
      "Val loss: 0.2867, accuracy: 93.01%\n",
      "Epoch: 127/150, lr:4.88e-05\n",
      "100%|██████████| 625/625 [01:07<00:00,  9.32it/s, loss:0.0020, acc:99.9750%]\n",
      "Val loss: 0.2896, accuracy: 93.00%\n",
      "Epoch: 128/150, lr:4.88e-05\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.43it/s, loss:0.0026, acc:99.9575%]\n",
      "Val loss: 0.2876, accuracy: 92.95%\n",
      "Epoch: 129/150, lr:4.88e-05\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.43it/s, loss:0.0022, acc:99.9675%] \n",
      "Val loss: 0.2885, accuracy: 92.98%\n",
      "Epoch: 130/150, lr:4.88e-05\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.46it/s, loss:0.0020, acc:99.9775%]\n",
      "Val loss: 0.2902, accuracy: 93.01%\n",
      "Epoch: 131/150, lr:4.88e-05\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.44it/s, loss:0.0021, acc:99.9800%]\n",
      "Val loss: 0.2895, accuracy: 93.04%\n",
      "Epoch: 132/150, lr:4.88e-05\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.35it/s, loss:0.0022, acc:99.9800%]\n",
      "Val loss: 0.2921, accuracy: 92.93%\n",
      "Epoch: 133/150, lr:4.88e-05\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.35it/s, loss:0.0024, acc:99.9750%]\n",
      "Val loss: 0.2870, accuracy: 93.03%\n",
      "Epoch: 134/150, lr:2.44e-05\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.34it/s, loss:0.0020, acc:99.9825%] \n",
      "Val loss: 0.2882, accuracy: 92.97%\n",
      "Epoch: 135/150, lr:2.44e-05\n",
      "100%|██████████| 625/625 [01:07<00:00,  9.28it/s, loss:0.0021, acc:99.9700%]\n",
      "Val loss: 0.2879, accuracy: 92.99%\n",
      "Epoch: 136/150, lr:2.44e-05\n",
      "100%|██████████| 625/625 [01:07<00:00,  9.33it/s, loss:0.0021, acc:99.9750%]\n",
      "Val loss: 0.2903, accuracy: 93.02%\n",
      "Epoch: 137/150, lr:2.44e-05\n",
      "100%|██████████| 625/625 [01:07<00:00,  9.21it/s, loss:0.0021, acc:99.9750%]\n",
      "Val loss: 0.2902, accuracy: 93.05%\n",
      "Epoch: 138/150, lr:2.44e-05\n",
      "100%|██████████| 625/625 [01:06<00:00,  9.41it/s, loss:0.0020, acc:99.9775%]\n",
      "Val loss: 0.2873, accuracy: 93.02%\n",
      "Epoch: 139/150, lr:2.44e-05\n",
      "100%|██████████| 625/625 [00:57<00:00, 10.79it/s, loss:0.0020, acc:99.9775%]\n",
      "Val loss: 0.2901, accuracy: 92.93%\n",
      "Epoch: 140/150, lr:2.44e-05\n",
      "100%|██████████| 625/625 [00:53<00:00, 11.58it/s, loss:0.0024, acc:99.9600%] \n",
      "Val loss: 0.2891, accuracy: 93.03%\n",
      "Epoch: 141/150, lr:2.44e-05\n",
      "100%|██████████| 625/625 [00:54<00:00, 11.54it/s, loss:0.0021, acc:99.9850%]\n",
      "Val loss: 0.2880, accuracy: 92.95%\n",
      "Epoch: 142/150, lr:2.44e-05\n",
      "100%|██████████| 625/625 [00:53<00:00, 11.57it/s, loss:0.0020, acc:99.9675%]\n",
      "Val loss: 0.2889, accuracy: 92.99%\n",
      "Epoch: 143/150, lr:2.44e-05\n",
      "100%|██████████| 625/625 [00:53<00:00, 11.59it/s, loss:0.0021, acc:99.9700%]\n",
      "Val loss: 0.2871, accuracy: 93.03%\n",
      "Epoch: 144/150, lr:2.44e-05\n",
      "100%|██████████| 625/625 [00:54<00:00, 11.55it/s, loss:0.0020, acc:99.9800%]\n",
      "Val loss: 0.2893, accuracy: 92.91%\n",
      "Epoch: 145/150, lr:1.22e-05\n",
      "100%|██████████| 625/625 [00:54<00:00, 11.46it/s, loss:0.0024, acc:99.9500%]\n",
      "Val loss: 0.2872, accuracy: 92.98%\n",
      "Epoch: 146/150, lr:1.22e-05\n",
      "100%|██████████| 625/625 [00:54<00:00, 11.40it/s, loss:0.0021, acc:99.9725%]\n",
      "Val loss: 0.2879, accuracy: 92.97%\n",
      "Epoch: 147/150, lr:1.22e-05\n",
      "100%|██████████| 625/625 [00:54<00:00, 11.38it/s, loss:0.0021, acc:99.9650%]\n",
      "Val loss: 0.2893, accuracy: 93.02%\n",
      "Epoch: 148/150, lr:1.22e-05\n",
      "100%|██████████| 625/625 [00:55<00:00, 11.34it/s, loss:0.0019, acc:99.9850%]\n",
      "Val loss: 0.2880, accuracy: 92.97%\n",
      "Epoch: 149/150, lr:1.22e-05\n",
      "100%|██████████| 625/625 [00:55<00:00, 11.33it/s, loss:0.0021, acc:99.9750%]\n",
      "Val loss: 0.2913, accuracy: 92.97%\n",
      "Epoch: 150/150, lr:1.22e-05\n",
      "100%|██████████| 625/625 [00:55<00:00, 11.30it/s, loss:0.0021, acc:99.9725%]\n",
      "Val loss: 0.2901, accuracy: 92.95%\n",
      "Test loss: 0.2954, accuracy: 92.84%\n",
      "CPU times: user 2h 3min 29s, sys: 48min 38s, total: 2h 52min 7s\n",
      "Wall time: 2h 50min 50s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "opt = optim.SGD(model.parameters(), lr=1e-1, momentum=0.9, weight_decay=5e-4, nesterov=True)\n",
    "loss_fn = F.cross_entropy\n",
    "lr_scheduler = optim.lr_scheduler.ReduceLROnPlateau(opt, 'min', factor=0.5)\n",
    "\n",
    "# lr_scheduler = None\n",
    "history = main(model, opt, epoch=150, loss_fn=loss_fn, lr_scheduler=lr_scheduler)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# plot loss history\n",
    "We check the classifier by plotting the loss and acc."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_history(history)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "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": "3.5.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}