back_propagation.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 92,
   "id": "b042291a",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Dataset FashionMNIST\n",
      "    Number of datapoints: 60000\n",
      "    Root location: data\n",
      "    Split: Train\n",
      "    StandardTransform\n",
      "Transform: ToTensor()\n",
      "Net(\n",
      "  (fc1): Linear(in_features=784, out_features=200, bias=True)\n",
      "  (fc2): Linear(in_features=200, out_features=10, bias=True)\n",
      ")\n",
      "Epoch [1/100], Step [100/300], Loss: 0.7097, Accuracy: 75.00%\n",
      "Epoch [1/100], Step [200/300], Loss: 0.7761, Accuracy: 72.50%\n",
      "Epoch [1/100], Step [300/300], Loss: 0.5534, Accuracy: 80.00%\n",
      "Epoch [2/100], Step [100/300], Loss: 0.4441, Accuracy: 84.50%\n",
      "Epoch [2/100], Step [200/300], Loss: 0.5506, Accuracy: 79.50%\n",
      "Epoch [2/100], Step [300/300], Loss: 0.4628, Accuracy: 85.50%\n",
      "Epoch [3/100], Step [100/300], Loss: 0.3788, Accuracy: 86.50%\n",
      "Epoch [3/100], Step [200/300], Loss: 0.4692, Accuracy: 84.50%\n",
      "Epoch [3/100], Step [300/300], Loss: 0.5224, Accuracy: 83.00%\n",
      "Epoch [4/100], Step [100/300], Loss: 0.5472, Accuracy: 81.00%\n",
      "Epoch [4/100], Step [200/300], Loss: 0.3511, Accuracy: 87.00%\n",
      "Epoch [4/100], Step [300/300], Loss: 0.3890, Accuracy: 90.00%\n",
      "Epoch [5/100], Step [100/300], Loss: 0.3791, Accuracy: 86.00%\n",
      "Epoch [5/100], Step [200/300], Loss: 0.3688, Accuracy: 87.00%\n",
      "Epoch [5/100], Step [300/300], Loss: 0.3169, Accuracy: 89.50%\n",
      "Epoch [6/100], Step [100/300], Loss: 0.3887, Accuracy: 86.00%\n",
      "Epoch [6/100], Step [200/300], Loss: 0.4222, Accuracy: 84.50%\n",
      "Epoch [6/100], Step [300/300], Loss: 0.4072, Accuracy: 88.50%\n",
      "Epoch [7/100], Step [100/300], Loss: 0.5027, Accuracy: 82.50%\n",
      "Epoch [7/100], Step [200/300], Loss: 0.3226, Accuracy: 89.50%\n",
      "Epoch [7/100], Step [300/300], Loss: 0.4238, Accuracy: 88.50%\n",
      "Epoch [8/100], Step [100/300], Loss: 0.4045, Accuracy: 85.50%\n",
      "Epoch [8/100], Step [200/300], Loss: 0.3414, Accuracy: 89.50%\n",
      "Epoch [8/100], Step [300/300], Loss: 0.4065, Accuracy: 87.50%\n",
      "Epoch [9/100], Step [100/300], Loss: 0.3668, Accuracy: 88.50%\n",
      "Epoch [9/100], Step [200/300], Loss: 0.3533, Accuracy: 86.00%\n",
      "Epoch [9/100], Step [300/300], Loss: 0.2998, Accuracy: 91.00%\n",
      "Epoch [10/100], Step [100/300], Loss: 0.3853, Accuracy: 86.00%\n",
      "Epoch [10/100], Step [200/300], Loss: 0.2968, Accuracy: 90.00%\n",
      "Epoch [10/100], Step [300/300], Loss: 0.2854, Accuracy: 90.00%\n",
      "Epoch [11/100], Step [100/300], Loss: 0.2566, Accuracy: 91.50%\n",
      "Epoch [11/100], Step [200/300], Loss: 0.4050, Accuracy: 87.00%\n",
      "Epoch [11/100], Step [300/300], Loss: 0.3330, Accuracy: 88.50%\n",
      "Epoch [12/100], Step [100/300], Loss: 0.3573, Accuracy: 90.00%\n",
      "Epoch [12/100], Step [200/300], Loss: 0.3228, Accuracy: 87.50%\n",
      "Epoch [12/100], Step [300/300], Loss: 0.2936, Accuracy: 90.00%\n",
      "Epoch [13/100], Step [100/300], Loss: 0.3296, Accuracy: 85.00%\n",
      "Epoch [13/100], Step [200/300], Loss: 0.3374, Accuracy: 85.50%\n",
      "Epoch [13/100], Step [300/300], Loss: 0.3607, Accuracy: 86.00%\n",
      "Epoch [14/100], Step [100/300], Loss: 0.2856, Accuracy: 90.00%\n",
      "Epoch [14/100], Step [200/300], Loss: 0.3316, Accuracy: 88.50%\n",
      "Epoch [14/100], Step [300/300], Loss: 0.2658, Accuracy: 90.50%\n",
      "Epoch [15/100], Step [100/300], Loss: 0.3028, Accuracy: 88.50%\n",
      "Epoch [15/100], Step [200/300], Loss: 0.3404, Accuracy: 89.00%\n",
      "Epoch [15/100], Step [300/300], Loss: 0.2786, Accuracy: 89.00%\n",
      "Epoch [16/100], Step [100/300], Loss: 0.2559, Accuracy: 92.00%\n",
      "Epoch [16/100], Step [200/300], Loss: 0.2838, Accuracy: 90.00%\n",
      "Epoch [16/100], Step [300/300], Loss: 0.3823, Accuracy: 86.00%\n",
      "Epoch [17/100], Step [100/300], Loss: 0.2589, Accuracy: 92.50%\n",
      "Epoch [17/100], Step [200/300], Loss: 0.3595, Accuracy: 89.00%\n",
      "Epoch [17/100], Step [300/300], Loss: 0.2930, Accuracy: 90.50%\n",
      "Epoch [18/100], Step [100/300], Loss: 0.3047, Accuracy: 88.50%\n",
      "Epoch [18/100], Step [200/300], Loss: 0.3182, Accuracy: 87.50%\n",
      "Epoch [18/100], Step [300/300], Loss: 0.3535, Accuracy: 84.50%\n",
      "Epoch [19/100], Step [100/300], Loss: 0.2942, Accuracy: 88.50%\n",
      "Epoch [19/100], Step [200/300], Loss: 0.3018, Accuracy: 88.00%\n",
      "Epoch [19/100], Step [300/300], Loss: 0.3620, Accuracy: 88.50%\n",
      "Epoch [20/100], Step [100/300], Loss: 0.2090, Accuracy: 92.50%\n",
      "Epoch [20/100], Step [200/300], Loss: 0.3317, Accuracy: 89.00%\n",
      "Epoch [20/100], Step [300/300], Loss: 0.2206, Accuracy: 92.50%\n",
      "Epoch [21/100], Step [100/300], Loss: 0.2712, Accuracy: 88.50%\n",
      "Epoch [21/100], Step [200/300], Loss: 0.2967, Accuracy: 88.00%\n",
      "Epoch [21/100], Step [300/300], Loss: 0.2738, Accuracy: 91.00%\n",
      "Epoch [22/100], Step [100/300], Loss: 0.3441, Accuracy: 90.50%\n",
      "Epoch [22/100], Step [200/300], Loss: 0.2824, Accuracy: 90.00%\n",
      "Epoch [22/100], Step [300/300], Loss: 0.3871, Accuracy: 87.00%\n",
      "Epoch [23/100], Step [100/300], Loss: 0.2751, Accuracy: 90.50%\n",
      "Epoch [23/100], Step [200/300], Loss: 0.2607, Accuracy: 90.50%\n",
      "Epoch [23/100], Step [300/300], Loss: 0.2857, Accuracy: 89.00%\n",
      "Epoch [24/100], Step [100/300], Loss: 0.2844, Accuracy: 90.00%\n",
      "Epoch [24/100], Step [200/300], Loss: 0.2508, Accuracy: 92.00%\n",
      "Epoch [24/100], Step [300/300], Loss: 0.2573, Accuracy: 93.00%\n",
      "Epoch [25/100], Step [100/300], Loss: 0.2733, Accuracy: 90.00%\n",
      "Epoch [25/100], Step [200/300], Loss: 0.2156, Accuracy: 91.50%\n",
      "Epoch [25/100], Step [300/300], Loss: 0.2873, Accuracy: 90.50%\n",
      "Epoch [26/100], Step [100/300], Loss: 0.1809, Accuracy: 94.50%\n",
      "Epoch [26/100], Step [200/300], Loss: 0.3415, Accuracy: 88.00%\n",
      "Epoch [26/100], Step [300/300], Loss: 0.2913, Accuracy: 90.00%\n",
      "Epoch [27/100], Step [100/300], Loss: 0.1877, Accuracy: 93.00%\n",
      "Epoch [27/100], Step [200/300], Loss: 0.2655, Accuracy: 88.50%\n",
      "Epoch [27/100], Step [300/300], Loss: 0.3410, Accuracy: 88.00%\n",
      "Epoch [28/100], Step [100/300], Loss: 0.2667, Accuracy: 90.00%\n",
      "Epoch [28/100], Step [200/300], Loss: 0.2231, Accuracy: 93.50%\n",
      "Epoch [28/100], Step [300/300], Loss: 0.2269, Accuracy: 92.50%\n",
      "Epoch [29/100], Step [100/300], Loss: 0.1987, Accuracy: 91.50%\n",
      "Epoch [29/100], Step [200/300], Loss: 0.2352, Accuracy: 92.50%\n",
      "Epoch [29/100], Step [300/300], Loss: 0.2374, Accuracy: 92.00%\n",
      "Epoch [30/100], Step [100/300], Loss: 0.2050, Accuracy: 95.00%\n",
      "Epoch [30/100], Step [200/300], Loss: 0.2440, Accuracy: 91.50%\n",
      "Epoch [30/100], Step [300/300], Loss: 0.2794, Accuracy: 88.00%\n",
      "Epoch [31/100], Step [100/300], Loss: 0.1890, Accuracy: 93.00%\n",
      "Epoch [31/100], Step [200/300], Loss: 0.3067, Accuracy: 89.00%\n",
      "Epoch [31/100], Step [300/300], Loss: 0.1616, Accuracy: 93.00%\n",
      "Epoch [32/100], Step [100/300], Loss: 0.3316, Accuracy: 84.50%\n",
      "Epoch [32/100], Step [200/300], Loss: 0.2670, Accuracy: 90.00%\n",
      "Epoch [32/100], Step [300/300], Loss: 0.2161, Accuracy: 92.00%\n",
      "Epoch [33/100], Step [100/300], Loss: 0.1956, Accuracy: 91.00%\n",
      "Epoch [33/100], Step [200/300], Loss: 0.2260, Accuracy: 92.00%\n",
      "Epoch [33/100], Step [300/300], Loss: 0.3247, Accuracy: 85.50%\n",
      "Epoch [34/100], Step [100/300], Loss: 0.3038, Accuracy: 89.00%\n",
      "Epoch [34/100], Step [200/300], Loss: 0.2790, Accuracy: 87.50%\n",
      "Epoch [34/100], Step [300/300], Loss: 0.2574, Accuracy: 94.00%\n",
      "Epoch [35/100], Step [100/300], Loss: 0.1929, Accuracy: 91.50%\n",
      "Epoch [35/100], Step [200/300], Loss: 0.2903, Accuracy: 91.50%\n",
      "Epoch [35/100], Step [300/300], Loss: 0.2173, Accuracy: 91.50%\n",
      "Epoch [36/100], Step [100/300], Loss: 0.2465, Accuracy: 90.50%\n",
      "Epoch [36/100], Step [200/300], Loss: 0.2899, Accuracy: 89.50%\n",
      "Epoch [36/100], Step [300/300], Loss: 0.2004, Accuracy: 94.00%\n",
      "Epoch [37/100], Step [100/300], Loss: 0.2542, Accuracy: 91.00%\n",
      "Epoch [37/100], Step [200/300], Loss: 0.2102, Accuracy: 92.50%\n",
      "Epoch [37/100], Step [300/300], Loss: 0.2972, Accuracy: 89.50%\n",
      "Epoch [38/100], Step [100/300], Loss: 0.2315, Accuracy: 89.50%\n",
      "Epoch [38/100], Step [200/300], Loss: 0.1691, Accuracy: 94.50%\n",
      "Epoch [38/100], Step [300/300], Loss: 0.2255, Accuracy: 94.00%\n",
      "Epoch [39/100], Step [100/300], Loss: 0.2054, Accuracy: 93.50%\n",
      "Epoch [39/100], Step [200/300], Loss: 0.1700, Accuracy: 94.50%\n",
      "Epoch [39/100], Step [300/300], Loss: 0.3197, Accuracy: 87.50%\n",
      "Epoch [40/100], Step [100/300], Loss: 0.2271, Accuracy: 92.50%\n",
      "Epoch [40/100], Step [200/300], Loss: 0.1730, Accuracy: 93.50%\n",
      "Epoch [40/100], Step [300/300], Loss: 0.2163, Accuracy: 93.00%\n",
      "Epoch [41/100], Step [100/300], Loss: 0.2417, Accuracy: 93.00%\n",
      "Epoch [41/100], Step [200/300], Loss: 0.2531, Accuracy: 92.00%\n",
      "Epoch [41/100], Step [300/300], Loss: 0.2656, Accuracy: 90.50%\n",
      "Epoch [42/100], Step [100/300], Loss: 0.2567, Accuracy: 90.00%\n",
      "Epoch [42/100], Step [200/300], Loss: 0.2123, Accuracy: 94.00%\n",
      "Epoch [42/100], Step [300/300], Loss: 0.3478, Accuracy: 87.00%\n",
      "Epoch [43/100], Step [100/300], Loss: 0.2287, Accuracy: 93.50%\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [43/100], Step [200/300], Loss: 0.2100, Accuracy: 92.00%\n",
      "Epoch [43/100], Step [300/300], Loss: 0.1877, Accuracy: 93.50%\n",
      "Epoch [44/100], Step [100/300], Loss: 0.1759, Accuracy: 92.50%\n",
      "Epoch [44/100], Step [200/300], Loss: 0.2791, Accuracy: 89.00%\n",
      "Epoch [44/100], Step [300/300], Loss: 0.2358, Accuracy: 92.00%\n",
      "Epoch [45/100], Step [100/300], Loss: 0.1917, Accuracy: 92.50%\n",
      "Epoch [45/100], Step [200/300], Loss: 0.2198, Accuracy: 93.00%\n",
      "Epoch [45/100], Step [300/300], Loss: 0.2186, Accuracy: 91.50%\n",
      "Epoch [46/100], Step [100/300], Loss: 0.2386, Accuracy: 92.00%\n",
      "Epoch [46/100], Step [200/300], Loss: 0.2286, Accuracy: 93.50%\n",
      "Epoch [46/100], Step [300/300], Loss: 0.2634, Accuracy: 91.00%\n",
      "Epoch [47/100], Step [100/300], Loss: 0.2409, Accuracy: 91.00%\n",
      "Epoch [47/100], Step [200/300], Loss: 0.1645, Accuracy: 93.50%\n",
      "Epoch [47/100], Step [300/300], Loss: 0.2235, Accuracy: 94.50%\n",
      "Epoch [48/100], Step [100/300], Loss: 0.1869, Accuracy: 93.00%\n",
      "Epoch [48/100], Step [200/300], Loss: 0.2243, Accuracy: 90.50%\n",
      "Epoch [48/100], Step [300/300], Loss: 0.2295, Accuracy: 94.00%\n",
      "Epoch [49/100], Step [100/300], Loss: 0.2214, Accuracy: 93.00%\n",
      "Epoch [49/100], Step [200/300], Loss: 0.2314, Accuracy: 92.50%\n",
      "Epoch [49/100], Step [300/300], Loss: 0.2633, Accuracy: 92.00%\n",
      "Epoch [50/100], Step [100/300], Loss: 0.1125, Accuracy: 97.00%\n",
      "Epoch [50/100], Step [200/300], Loss: 0.2239, Accuracy: 90.00%\n",
      "Epoch [50/100], Step [300/300], Loss: 0.2021, Accuracy: 93.00%\n",
      "Epoch [51/100], Step [100/300], Loss: 0.2628, Accuracy: 90.50%\n",
      "Epoch [51/100], Step [200/300], Loss: 0.2441, Accuracy: 90.00%\n",
      "Epoch [51/100], Step [300/300], Loss: 0.1673, Accuracy: 95.50%\n",
      "Epoch [52/100], Step [100/300], Loss: 0.2432, Accuracy: 91.00%\n",
      "Epoch [52/100], Step [200/300], Loss: 0.2148, Accuracy: 91.00%\n",
      "Epoch [52/100], Step [300/300], Loss: 0.1798, Accuracy: 94.00%\n",
      "Epoch [53/100], Step [100/300], Loss: 0.1362, Accuracy: 96.00%\n",
      "Epoch [53/100], Step [200/300], Loss: 0.1992, Accuracy: 91.50%\n",
      "Epoch [53/100], Step [300/300], Loss: 0.2213, Accuracy: 91.00%\n",
      "Epoch [54/100], Step [100/300], Loss: 0.2576, Accuracy: 91.00%\n",
      "Epoch [54/100], Step [200/300], Loss: 0.2218, Accuracy: 93.50%\n",
      "Epoch [54/100], Step [300/300], Loss: 0.2088, Accuracy: 92.00%\n",
      "Epoch [55/100], Step [100/300], Loss: 0.1699, Accuracy: 95.00%\n",
      "Epoch [55/100], Step [200/300], Loss: 0.1797, Accuracy: 90.50%\n",
      "Epoch [55/100], Step [300/300], Loss: 0.2450, Accuracy: 91.50%\n",
      "Epoch [56/100], Step [100/300], Loss: 0.1727, Accuracy: 93.50%\n",
      "Epoch [56/100], Step [200/300], Loss: 0.2100, Accuracy: 92.50%\n",
      "Epoch [56/100], Step [300/300], Loss: 0.1919, Accuracy: 92.50%\n",
      "Epoch [57/100], Step [100/300], Loss: 0.2505, Accuracy: 90.50%\n",
      "Epoch [57/100], Step [200/300], Loss: 0.2041, Accuracy: 92.50%\n",
      "Epoch [57/100], Step [300/300], Loss: 0.1344, Accuracy: 95.00%\n",
      "Epoch [58/100], Step [100/300], Loss: 0.1655, Accuracy: 94.00%\n",
      "Epoch [58/100], Step [200/300], Loss: 0.1252, Accuracy: 96.50%\n",
      "Epoch [58/100], Step [300/300], Loss: 0.1882, Accuracy: 93.00%\n",
      "Epoch [59/100], Step [100/300], Loss: 0.2025, Accuracy: 92.50%\n",
      "Epoch [59/100], Step [200/300], Loss: 0.2203, Accuracy: 91.00%\n",
      "Epoch [59/100], Step [300/300], Loss: 0.1898, Accuracy: 92.00%\n",
      "Epoch [60/100], Step [100/300], Loss: 0.2220, Accuracy: 89.00%\n",
      "Epoch [60/100], Step [200/300], Loss: 0.1423, Accuracy: 92.50%\n",
      "Epoch [60/100], Step [300/300], Loss: 0.2228, Accuracy: 91.50%\n",
      "Epoch [61/100], Step [100/300], Loss: 0.1886, Accuracy: 94.00%\n",
      "Epoch [61/100], Step [200/300], Loss: 0.1825, Accuracy: 95.50%\n",
      "Epoch [61/100], Step [300/300], Loss: 0.1525, Accuracy: 94.00%\n",
      "Epoch [62/100], Step [100/300], Loss: 0.2199, Accuracy: 93.50%\n",
      "Epoch [62/100], Step [200/300], Loss: 0.1500, Accuracy: 94.50%\n",
      "Epoch [62/100], Step [300/300], Loss: 0.1955, Accuracy: 93.00%\n",
      "Epoch [63/100], Step [100/300], Loss: 0.1707, Accuracy: 94.00%\n",
      "Epoch [63/100], Step [200/300], Loss: 0.2360, Accuracy: 90.00%\n",
      "Epoch [63/100], Step [300/300], Loss: 0.1725, Accuracy: 94.50%\n",
      "Epoch [64/100], Step [100/300], Loss: 0.1673, Accuracy: 95.00%\n",
      "Epoch [64/100], Step [200/300], Loss: 0.1629, Accuracy: 95.50%\n",
      "Epoch [64/100], Step [300/300], Loss: 0.1727, Accuracy: 94.50%\n",
      "Epoch [65/100], Step [100/300], Loss: 0.1608, Accuracy: 94.00%\n",
      "Epoch [65/100], Step [200/300], Loss: 0.1626, Accuracy: 94.00%\n",
      "Epoch [65/100], Step [300/300], Loss: 0.1374, Accuracy: 95.00%\n",
      "Epoch [66/100], Step [100/300], Loss: 0.1969, Accuracy: 94.50%\n",
      "Epoch [66/100], Step [200/300], Loss: 0.1051, Accuracy: 96.00%\n",
      "Epoch [66/100], Step [300/300], Loss: 0.1683, Accuracy: 93.00%\n",
      "Epoch [67/100], Step [100/300], Loss: 0.2153, Accuracy: 91.50%\n",
      "Epoch [67/100], Step [200/300], Loss: 0.2012, Accuracy: 93.50%\n",
      "Epoch [67/100], Step [300/300], Loss: 0.1036, Accuracy: 97.00%\n",
      "Epoch [68/100], Step [100/300], Loss: 0.1239, Accuracy: 95.50%\n",
      "Epoch [68/100], Step [200/300], Loss: 0.1779, Accuracy: 94.00%\n",
      "Epoch [68/100], Step [300/300], Loss: 0.1819, Accuracy: 92.50%\n",
      "Epoch [69/100], Step [100/300], Loss: 0.1207, Accuracy: 96.00%\n",
      "Epoch [69/100], Step [200/300], Loss: 0.1848, Accuracy: 93.50%\n",
      "Epoch [69/100], Step [300/300], Loss: 0.2002, Accuracy: 92.50%\n",
      "Epoch [70/100], Step [100/300], Loss: 0.1410, Accuracy: 95.00%\n",
      "Epoch [70/100], Step [200/300], Loss: 0.1495, Accuracy: 95.00%\n",
      "Epoch [70/100], Step [300/300], Loss: 0.1358, Accuracy: 93.50%\n",
      "Epoch [71/100], Step [100/300], Loss: 0.2078, Accuracy: 93.50%\n",
      "Epoch [71/100], Step [200/300], Loss: 0.1408, Accuracy: 94.50%\n",
      "Epoch [71/100], Step [300/300], Loss: 0.1784, Accuracy: 95.00%\n",
      "Epoch [72/100], Step [100/300], Loss: 0.1691, Accuracy: 93.00%\n",
      "Epoch [72/100], Step [200/300], Loss: 0.1310, Accuracy: 96.00%\n",
      "Epoch [72/100], Step [300/300], Loss: 0.1903, Accuracy: 93.50%\n",
      "Epoch [73/100], Step [100/300], Loss: 0.2204, Accuracy: 92.50%\n",
      "Epoch [73/100], Step [200/300], Loss: 0.1533, Accuracy: 95.00%\n",
      "Epoch [73/100], Step [300/300], Loss: 0.1961, Accuracy: 93.50%\n",
      "Epoch [74/100], Step [100/300], Loss: 0.1085, Accuracy: 97.00%\n",
      "Epoch [74/100], Step [200/300], Loss: 0.1289, Accuracy: 98.00%\n",
      "Epoch [74/100], Step [300/300], Loss: 0.1174, Accuracy: 97.00%\n",
      "Epoch [75/100], Step [100/300], Loss: 0.1574, Accuracy: 94.00%\n",
      "Epoch [75/100], Step [200/300], Loss: 0.1776, Accuracy: 93.00%\n",
      "Epoch [75/100], Step [300/300], Loss: 0.1034, Accuracy: 97.00%\n",
      "Epoch [76/100], Step [100/300], Loss: 0.1671, Accuracy: 93.00%\n",
      "Epoch [76/100], Step [200/300], Loss: 0.2452, Accuracy: 89.50%\n",
      "Epoch [76/100], Step [300/300], Loss: 0.1561, Accuracy: 94.00%\n",
      "Epoch [77/100], Step [100/300], Loss: 0.1402, Accuracy: 93.50%\n",
      "Epoch [77/100], Step [200/300], Loss: 0.1584, Accuracy: 94.00%\n",
      "Epoch [77/100], Step [300/300], Loss: 0.1350, Accuracy: 96.50%\n",
      "Epoch [78/100], Step [100/300], Loss: 0.1133, Accuracy: 96.50%\n",
      "Epoch [78/100], Step [200/300], Loss: 0.1092, Accuracy: 95.50%\n",
      "Epoch [78/100], Step [300/300], Loss: 0.1604, Accuracy: 95.00%\n",
      "Epoch [79/100], Step [100/300], Loss: 0.2074, Accuracy: 93.50%\n",
      "Epoch [79/100], Step [200/300], Loss: 0.1575, Accuracy: 95.00%\n",
      "Epoch [79/100], Step [300/300], Loss: 0.1623, Accuracy: 95.50%\n",
      "Epoch [80/100], Step [100/300], Loss: 0.1681, Accuracy: 95.50%\n",
      "Epoch [80/100], Step [200/300], Loss: 0.1891, Accuracy: 92.00%\n",
      "Epoch [80/100], Step [300/300], Loss: 0.1928, Accuracy: 92.50%\n",
      "Epoch [81/100], Step [100/300], Loss: 0.1658, Accuracy: 94.00%\n",
      "Epoch [81/100], Step [200/300], Loss: 0.1424, Accuracy: 94.00%\n",
      "Epoch [81/100], Step [300/300], Loss: 0.1397, Accuracy: 93.00%\n",
      "Epoch [82/100], Step [100/300], Loss: 0.1986, Accuracy: 95.00%\n",
      "Epoch [82/100], Step [200/300], Loss: 0.1350, Accuracy: 94.50%\n",
      "Epoch [82/100], Step [300/300], Loss: 0.1143, Accuracy: 97.00%\n",
      "Epoch [83/100], Step [100/300], Loss: 0.1400, Accuracy: 95.00%\n",
      "Epoch [83/100], Step [200/300], Loss: 0.1378, Accuracy: 96.00%\n",
      "Epoch [83/100], Step [300/300], Loss: 0.0941, Accuracy: 97.50%\n",
      "Epoch [84/100], Step [100/300], Loss: 0.1481, Accuracy: 94.50%\n",
      "Epoch [84/100], Step [200/300], Loss: 0.1621, Accuracy: 94.50%\n",
      "Epoch [84/100], Step [300/300], Loss: 0.1489, Accuracy: 95.50%\n",
      "Epoch [85/100], Step [100/300], Loss: 0.1684, Accuracy: 91.00%\n",
      "Epoch [85/100], Step [200/300], Loss: 0.1454, Accuracy: 96.50%\n",
      "Epoch [85/100], Step [300/300], Loss: 0.1532, Accuracy: 94.00%\n",
      "Epoch [86/100], Step [100/300], Loss: 0.1331, Accuracy: 95.00%\n",
      "Epoch [86/100], Step [200/300], Loss: 0.1855, Accuracy: 92.00%\n",
      "Epoch [86/100], Step [300/300], Loss: 0.1891, Accuracy: 92.00%\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [87/100], Step [100/300], Loss: 0.1147, Accuracy: 96.00%\n",
      "Epoch [87/100], Step [200/300], Loss: 0.1069, Accuracy: 97.50%\n",
      "Epoch [87/100], Step [300/300], Loss: 0.1070, Accuracy: 95.00%\n",
      "Epoch [88/100], Step [100/300], Loss: 0.1064, Accuracy: 95.50%\n",
      "Epoch [88/100], Step [200/300], Loss: 0.1589, Accuracy: 93.00%\n",
      "Epoch [88/100], Step [300/300], Loss: 0.1388, Accuracy: 94.50%\n",
      "Epoch [89/100], Step [100/300], Loss: 0.1647, Accuracy: 94.50%\n",
      "Epoch [89/100], Step [200/300], Loss: 0.1339, Accuracy: 95.50%\n",
      "Epoch [89/100], Step [300/300], Loss: 0.1272, Accuracy: 96.00%\n",
      "Epoch [90/100], Step [100/300], Loss: 0.0971, Accuracy: 97.50%\n",
      "Epoch [90/100], Step [200/300], Loss: 0.1629, Accuracy: 95.00%\n",
      "Epoch [90/100], Step [300/300], Loss: 0.1819, Accuracy: 94.00%\n",
      "Epoch [91/100], Step [100/300], Loss: 0.1723, Accuracy: 94.00%\n",
      "Epoch [91/100], Step [200/300], Loss: 0.1145, Accuracy: 96.00%\n",
      "Epoch [91/100], Step [300/300], Loss: 0.1140, Accuracy: 96.50%\n",
      "Epoch [92/100], Step [100/300], Loss: 0.0982, Accuracy: 96.00%\n",
      "Epoch [92/100], Step [200/300], Loss: 0.1725, Accuracy: 94.00%\n",
      "Epoch [92/100], Step [300/300], Loss: 0.1412, Accuracy: 94.50%\n",
      "Epoch [93/100], Step [100/300], Loss: 0.1033, Accuracy: 96.50%\n",
      "Epoch [93/100], Step [200/300], Loss: 0.1207, Accuracy: 95.00%\n",
      "Epoch [93/100], Step [300/300], Loss: 0.1083, Accuracy: 96.00%\n",
      "Epoch [94/100], Step [100/300], Loss: 0.0920, Accuracy: 97.00%\n",
      "Epoch [94/100], Step [200/300], Loss: 0.1038, Accuracy: 97.00%\n",
      "Epoch [94/100], Step [300/300], Loss: 0.1207, Accuracy: 95.00%\n",
      "Epoch [95/100], Step [100/300], Loss: 0.1172, Accuracy: 96.50%\n",
      "Epoch [95/100], Step [200/300], Loss: 0.1228, Accuracy: 95.00%\n",
      "Epoch [95/100], Step [300/300], Loss: 0.1122, Accuracy: 95.00%\n",
      "Epoch [96/100], Step [100/300], Loss: 0.0928, Accuracy: 97.00%\n",
      "Epoch [96/100], Step [200/300], Loss: 0.1084, Accuracy: 96.00%\n",
      "Epoch [96/100], Step [300/300], Loss: 0.1512, Accuracy: 94.50%\n",
      "Epoch [97/100], Step [100/300], Loss: 0.1851, Accuracy: 94.50%\n",
      "Epoch [97/100], Step [200/300], Loss: 0.1236, Accuracy: 96.50%\n",
      "Epoch [97/100], Step [300/300], Loss: 0.1431, Accuracy: 93.50%\n",
      "Epoch [98/100], Step [100/300], Loss: 0.1501, Accuracy: 96.00%\n",
      "Epoch [98/100], Step [200/300], Loss: 0.1307, Accuracy: 92.00%\n",
      "Epoch [98/100], Step [300/300], Loss: 0.1410, Accuracy: 96.00%\n",
      "Epoch [99/100], Step [100/300], Loss: 0.0956, Accuracy: 96.50%\n",
      "Epoch [99/100], Step [200/300], Loss: 0.1311, Accuracy: 95.50%\n",
      "Epoch [99/100], Step [300/300], Loss: 0.1419, Accuracy: 95.00%\n",
      "Epoch [100/100], Step [100/300], Loss: 0.1437, Accuracy: 94.50%\n",
      "Epoch [100/100], Step [200/300], Loss: 0.1291, Accuracy: 96.00%\n",
      "Epoch [100/100], Step [300/300], Loss: 0.1248, Accuracy: 95.00%\n",
      "Test Accuracy of the model on the 10000 test images: 96.19 %\n"
     ]
    }
   ],
   "source": [
    "import torch.nn.functional as F\n",
    "import torch\n",
    "from torch.utils.data import Dataset\n",
    "from torchvision import datasets\n",
    "from torchvision.transforms import ToTensor\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy\n",
    "import torch.optim\n",
    "from torch import nn\n",
    "\n",
    "training_data = datasets.FashionMNIST(\n",
    "    root=\"data\",\n",
    "    train=True,\n",
    "    download=True,\n",
    "    transform=ToTensor()\n",
    ")\n",
    "print(training_data)\n",
    "# (x_train, y_train)= tf.keras.datasets.fashion_mnist.load_data()\n",
    "# x_train = x_train.reshape(60000, 784)\n",
    "# print(x_train)\n",
    "from torch.utils.data import DataLoader\n",
    "\n",
    "train_loader = DataLoader(training_data, batch_size=200, shuffle=True)\n",
    "\n",
    "\n",
    "# class Net(nn.Module):\n",
    "#     def __init__(self):\n",
    "#         super(Net, self).__init__()\n",
    "#         self.fc1 = nn.Linear(28 * 28, 200)\n",
    "#         self.fc2 = nn.Linear(200, 9)\n",
    "\n",
    "#     def forward(self, x):\n",
    "#         x = F.relu(self.fc1(x))\n",
    "#         x = F.log_softmax(x)\n",
    "#         return x\n",
    "\n",
    "class Net(nn.Module):\n",
    "    def __init__(self):\n",
    "        super(Net, self).__init__()\n",
    "        self.fc1 = nn.Linear(28 * 28, 200)\n",
    "#         self.fc2 = nn.Linear(200, 20)\n",
    "        self.fc2 = nn.Linear(200, 10)\n",
    "\n",
    "    def forward(self, x):\n",
    "        x = F.relu(self.fc1(x))\n",
    "#         x = F.relu(self.fc2(x))\n",
    "        x = self.fc2(x)\n",
    "        return F.log_softmax(x, -1)\n",
    "\n",
    "\n",
    "net = Net()\n",
    "\n",
    "\n",
    "net = Net()\n",
    "print(net)\n",
    "\n",
    "learning_rate = 0.01\n",
    "epochs = 100\n",
    "batch_size = 200\n",
    "# optimizer = torch.optim.Adam(net.parameters(), lr=learning_rate)\n",
    "optimizer = torch.optim.SGD(net.parameters(), lr=0.01, momentum=0.9)\n",
    "\n",
    "criterion = nn.NLLLoss()\n",
    "\n",
    "# запускаем главный тренировочный цикл\n",
    "total_step = len(train_loader)\n",
    "loss_list = []\n",
    "acc_list = []\n",
    "\n",
    "for epoch in range(epochs):\n",
    "    for i, (images, labels) in enumerate(train_loader):\n",
    "        # Прямой запуск\n",
    "        images = images.reshape(-1, 28 * 28)\n",
    "        #         print(images.size())\n",
    "        outputs = net(images)\n",
    "        loss = criterion(outputs, labels)\n",
    "        loss_list.append(loss.item())\n",
    "        optimizer.zero_grad()\n",
    "        # Обратное распространение и оптимизатор\n",
    "\n",
    "        loss.backward()\n",
    "        optimizer.step()\n",
    "\n",
    "        # Отслеживание точности\n",
    "        total = labels.size(0)\n",
    "        _, predicted = torch.max(outputs.data, 1)\n",
    "        correct = (predicted == labels).sum().item()\n",
    "        acc_list.append(correct / total)\n",
    "\n",
    "        if (i + 1) % 100 == 0:\n",
    "            print('Epoch [{}/{}], Step [{}/{}], Loss: {:.4f}, Accuracy: {:.2f}%'\n",
    "                  .format(epoch + 1, epochs, i + 1, total_step, loss.item(),\n",
    "                          (correct / total) * 100))\n",
    "\n",
    "testing_data = datasets.FashionMNIST(\n",
    "    root=\"data\",\n",
    "    train=False,\n",
    "    download=True,\n",
    "    transform=ToTensor()\n",
    ")\n",
    "test_loader = DataLoader(training_data, batch_size=200, shuffle=True)\n",
    "# net.eval()\n",
    "with torch.no_grad():\n",
    "    correct = 0\n",
    "    total = 0\n",
    "    for images, labels in test_loader:\n",
    "        images = images.reshape(-1, 28 * 28)\n",
    "        outputs = net(images)\n",
    "        _, predicted = torch.max(outputs.data, 1)\n",
    "        total += labels.size(0)\n",
    "        correct += (predicted == labels).sum().item()\n",
    "\n",
    "    print('Test Accuracy of the model on the 10000 test images: {} %'.format((correct / total) * 100))\n",
    "    \n",
    "    \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "2493329f",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "id": "67fffc52",
   "metadata": {},
   "outputs": [],
   "source": [
    "from bokeh.plotting import figure\n",
    "from bokeh.io import show\n",
    "from bokeh.models import LinearAxis, Range1d\n",
    "\n",
    "p = figure(y_axis_label='Loss', width=1000, y_range=(0, 1), title='results')\n",
    "p.extra_y_ranges = {'Accuracy': Range1d(start=0, end=100)}\n",
    "p.add_layout(LinearAxis(y_range_name='Accuracy', axis_label='Accuracy (%)'), 'right')\n",
    "# p.line(numpy.arange(len(loss_list)), loss_list)\n",
    "p.line(numpy.arange(len(loss_list)), numpy.array(acc_list) * 100, y_range_name='Accuracy', color='red')\n",
    "show(p)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "3eff1cfe",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "id": "fb74ad4c",
   "metadata": {},
   "outputs": [],
   "source": [
    "import torchvision\n",
    "import torch.onnx\n",
    "channel = 1\n",
    "height = 28\n",
    "width = 28\n",
    "\n",
    "dummy_input = torch.randn(height * width)\n",
    "\n",
    "torch.onnx.export(net1, dummy_input, \"onnx_model.onnx\")\n",
    "\n",
    "\n",
    "\n"
   ]
  }
 ],
 "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": "3.8.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}