Classification/LeNet_5(1998).ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "LeNet-5.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "authorship_tag": "ABX9TyMwodz3tBF4icPXNjMjiCiu",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU",
    "widgets": {
      "application/vnd.jupyter.widget-state+json": {
        "fddb8bb17dde43f4b6db310571d4d7fe": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "state": {
            "_view_name": "HBoxView",
            "_dom_classes": [],
            "_model_name": "HBoxModel",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "box_style": "",
            "layout": "IPY_MODEL_e1b5c2ffeb4d4cb8958f5268c9c3e314",
            "_model_module": "@jupyter-widgets/controls",
            "children": [
              "IPY_MODEL_22b0a61e45b64a70998d05f7c8f6dc02",
              "IPY_MODEL_fa7bad13be60496784e49a1252786126"
            ]
          }
        },
        "e1b5c2ffeb4d4cb8958f5268c9c3e314": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "22b0a61e45b64a70998d05f7c8f6dc02": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "state": {
            "_view_name": "ProgressView",
            "style": "IPY_MODEL_15594db186994d5ba8c965be2d2b1c6d",
            "_dom_classes": [],
            "description": "",
            "_model_name": "FloatProgressModel",
            "bar_style": "success",
            "max": 1,
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": 1,
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "orientation": "horizontal",
            "min": 0,
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_0b1516af113c42f092432b96e1fdbf75"
          }
        },
        "fa7bad13be60496784e49a1252786126": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_7f996bad066f4570a15c91b4634f5562",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": " 9920512/? [00:01&lt;00:00, 8483832.39it/s]",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_848cc9dddd4b421eabe19acbf345509b"
          }
        },
        "15594db186994d5ba8c965be2d2b1c6d": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "ProgressStyleModel",
            "description_width": "initial",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "bar_color": null,
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "0b1516af113c42f092432b96e1fdbf75": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "7f996bad066f4570a15c91b4634f5562": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "DescriptionStyleModel",
            "description_width": "",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "848cc9dddd4b421eabe19acbf345509b": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "29a320d657af4f6d9b112ef7142b1efe": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "state": {
            "_view_name": "HBoxView",
            "_dom_classes": [],
            "_model_name": "HBoxModel",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "box_style": "",
            "layout": "IPY_MODEL_709cf9f4002c47219e2cec7ab69caa20",
            "_model_module": "@jupyter-widgets/controls",
            "children": [
              "IPY_MODEL_607836d4f6b74f2089596f30780c5b69",
              "IPY_MODEL_9041d718fbe94027894316849d891d9c"
            ]
          }
        },
        "709cf9f4002c47219e2cec7ab69caa20": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "607836d4f6b74f2089596f30780c5b69": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "state": {
            "_view_name": "ProgressView",
            "style": "IPY_MODEL_f58a8a22090f4a0c83f205a8a5943094",
            "_dom_classes": [],
            "description": "",
            "_model_name": "FloatProgressModel",
            "bar_style": "success",
            "max": 1,
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": 1,
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "orientation": "horizontal",
            "min": 0,
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_9e7dc43820f4411bba9bf6ab316a326f"
          }
        },
        "9041d718fbe94027894316849d891d9c": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_0366caeae06f4f27873336066d92bf61",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": " 32768/? [00:00&lt;00:00, 108648.60it/s]",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_8d76b3d020a84684951f483f4f5d4fac"
          }
        },
        "f58a8a22090f4a0c83f205a8a5943094": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "ProgressStyleModel",
            "description_width": "initial",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "bar_color": null,
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "9e7dc43820f4411bba9bf6ab316a326f": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "0366caeae06f4f27873336066d92bf61": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "DescriptionStyleModel",
            "description_width": "",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "8d76b3d020a84684951f483f4f5d4fac": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "fa72678cd3b84b068c306d1feb216aef": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "state": {
            "_view_name": "HBoxView",
            "_dom_classes": [],
            "_model_name": "HBoxModel",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "box_style": "",
            "layout": "IPY_MODEL_ec9839ea64964e2a9011b6b03eff4c56",
            "_model_module": "@jupyter-widgets/controls",
            "children": [
              "IPY_MODEL_a72f6cf8bd4a4fd28dd40cc5bf674628",
              "IPY_MODEL_9eae4fb6f2f64f66b48c6c9ac0cb6fe3"
            ]
          }
        },
        "ec9839ea64964e2a9011b6b03eff4c56": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "a72f6cf8bd4a4fd28dd40cc5bf674628": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "state": {
            "_view_name": "ProgressView",
            "style": "IPY_MODEL_a1b5ca0ee4084db786424e334c491f18",
            "_dom_classes": [],
            "description": "",
            "_model_name": "FloatProgressModel",
            "bar_style": "success",
            "max": 1,
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": 1,
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "orientation": "horizontal",
            "min": 0,
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_159177f6fd51406da6e6202a156f996f"
          }
        },
        "9eae4fb6f2f64f66b48c6c9ac0cb6fe3": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_024ab54ce9b84f38a9fecde6a3981298",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": " 1654784/? [00:00&lt;00:00, 7368316.89it/s]",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_37b7958c691b4e43918baa62460ed6d9"
          }
        },
        "a1b5ca0ee4084db786424e334c491f18": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "ProgressStyleModel",
            "description_width": "initial",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "bar_color": null,
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "159177f6fd51406da6e6202a156f996f": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "024ab54ce9b84f38a9fecde6a3981298": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "DescriptionStyleModel",
            "description_width": "",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "37b7958c691b4e43918baa62460ed6d9": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "d653fd88f39a4562a58249a3465268d7": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "state": {
            "_view_name": "HBoxView",
            "_dom_classes": [],
            "_model_name": "HBoxModel",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "box_style": "",
            "layout": "IPY_MODEL_5b60ed3df0e14ecc88f0f7f6be90c7aa",
            "_model_module": "@jupyter-widgets/controls",
            "children": [
              "IPY_MODEL_b7fc408a6e724caa853aa6966f3256a9",
              "IPY_MODEL_561b693c951047bbaca3ab15871d2f88"
            ]
          }
        },
        "5b60ed3df0e14ecc88f0f7f6be90c7aa": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "b7fc408a6e724caa853aa6966f3256a9": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "state": {
            "_view_name": "ProgressView",
            "style": "IPY_MODEL_26f0968adf634dbf8a644b4c06fee77d",
            "_dom_classes": [],
            "description": "",
            "_model_name": "FloatProgressModel",
            "bar_style": "success",
            "max": 1,
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": 1,
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "orientation": "horizontal",
            "min": 0,
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_ea621cca67704ea489015440d2609669"
          }
        },
        "561b693c951047bbaca3ab15871d2f88": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_22c1af08f45949debfdcea08fb3647c4",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": " 8192/? [00:00&lt;00:00, 28046.91it/s]",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_55d4ca968c36409a91815027c609ee64"
          }
        },
        "26f0968adf634dbf8a644b4c06fee77d": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "ProgressStyleModel",
            "description_width": "initial",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "bar_color": null,
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "ea621cca67704ea489015440d2609669": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "22c1af08f45949debfdcea08fb3647c4": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "DescriptionStyleModel",
            "description_width": "",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "55d4ca968c36409a91815027c609ee64": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        }
      }
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/Seonghoon-Yu/paper-implement-in-pytorch/blob/master/Classification/LeNet_5.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "AhJvaSkTioch",
        "outputId": "e11ea912-dced-4c67-b071-4178c7b30b1c"
      },
      "source": [
        "from google.colab import drive\r\n",
        "drive.mount('cookbook')"
      ],
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Mounted at cookbook\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "2WOqdicXjfF_"
      },
      "source": [
        "# CUDA check\r\n",
        "우선, 사용가능한 GPU를 확인합니다."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ZaY_Gb_jizuY",
        "outputId": "67ece961-a7fa-4a63-e1a8-28677ade2b1d"
      },
      "source": [
        "import torch\r\n",
        "if torch.cuda.is_available():\r\n",
        "    device = torch.device('cuda:0')\r\n",
        "else:\r\n",
        "    device = torch.device('cpu')\r\n",
        "print(device)"
      ],
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "cuda:0\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Ss5EHfkxk14m"
      },
      "source": [
        "# MINST dataset 불러오기\r\n",
        "LeNet-5를 학습하기 위한 MINST dataset을 불러옵니다. MNIST dataset은 torchvision 패키지에서 제공합니다."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PSquLTpZldYG"
      },
      "source": [
        "# 우선, MNIST dataset에 적용할 transformation 객체를 생성합니다.\r\n",
        "from torchvision import transforms\r\n",
        "\r\n",
        "# transformation 정의하기\r\n",
        "data_transform = transforms.Compose([\r\n",
        "            transforms.Resize((32, 32)),\r\n",
        "            transforms.ToTensor(),\r\n",
        "])"
      ],
      "execution_count": 3,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vStLrJZe05UG"
      },
      "source": [
        "from six.moves import urllib    \r\n",
        "opener = urllib.request.build_opener()\r\n",
        "opener.addheaders = [('User-agent', 'Mozilla/5.0')]\r\n",
        "urllib.request.install_opener(opener)"
      ],
      "execution_count": 4,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "JZS-g-VwkL0H",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 417,
          "referenced_widgets": [
            "fddb8bb17dde43f4b6db310571d4d7fe",
            "e1b5c2ffeb4d4cb8958f5268c9c3e314",
            "22b0a61e45b64a70998d05f7c8f6dc02",
            "fa7bad13be60496784e49a1252786126",
            "15594db186994d5ba8c965be2d2b1c6d",
            "0b1516af113c42f092432b96e1fdbf75",
            "7f996bad066f4570a15c91b4634f5562",
            "848cc9dddd4b421eabe19acbf345509b",
            "29a320d657af4f6d9b112ef7142b1efe",
            "709cf9f4002c47219e2cec7ab69caa20",
            "607836d4f6b74f2089596f30780c5b69",
            "9041d718fbe94027894316849d891d9c",
            "f58a8a22090f4a0c83f205a8a5943094",
            "9e7dc43820f4411bba9bf6ab316a326f",
            "0366caeae06f4f27873336066d92bf61",
            "8d76b3d020a84684951f483f4f5d4fac",
            "fa72678cd3b84b068c306d1feb216aef",
            "ec9839ea64964e2a9011b6b03eff4c56",
            "a72f6cf8bd4a4fd28dd40cc5bf674628",
            "9eae4fb6f2f64f66b48c6c9ac0cb6fe3",
            "a1b5ca0ee4084db786424e334c491f18",
            "159177f6fd51406da6e6202a156f996f",
            "024ab54ce9b84f38a9fecde6a3981298",
            "37b7958c691b4e43918baa62460ed6d9",
            "d653fd88f39a4562a58249a3465268d7",
            "5b60ed3df0e14ecc88f0f7f6be90c7aa",
            "b7fc408a6e724caa853aa6966f3256a9",
            "561b693c951047bbaca3ab15871d2f88",
            "26f0968adf634dbf8a644b4c06fee77d",
            "ea621cca67704ea489015440d2609669",
            "22c1af08f45949debfdcea08fb3647c4",
            "55d4ca968c36409a91815027c609ee64"
          ]
        },
        "outputId": "72645630-415e-4b8e-f5a2-1453b61931d8"
      },
      "source": [
        "# MNIST training dataset 불러오기\r\n",
        "from torchvision import datasets\r\n",
        "\r\n",
        "# 데이터를 저장할 경로 설정\r\n",
        "path2data = '/content/data'\r\n",
        "\r\n",
        "# training data 불러오기\r\n",
        "train_data = datasets.MNIST(path2data, train=True, download=True, transform=data_transform)\r\n",
        "\r\n",
        "# MNIST test dataset 불러오기\r\n",
        "val_data = datasets.MNIST(path2data, train=False, download=True, transform=data_transform)"
      ],
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Downloading http://yann.lecun.com/exdb/mnist/train-images-idx3-ubyte.gz to /content/data/MNIST/raw/train-images-idx3-ubyte.gz\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "fddb8bb17dde43f4b6db310571d4d7fe",
              "version_minor": 0,
              "version_major": 2
            },
            "text/plain": [
              "HBox(children=(FloatProgress(value=1.0, bar_style='info', max=1.0), HTML(value='')))"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "Extracting /content/data/MNIST/raw/train-images-idx3-ubyte.gz to /content/data/MNIST/raw\n",
            "Downloading http://yann.lecun.com/exdb/mnist/train-labels-idx1-ubyte.gz to /content/data/MNIST/raw/train-labels-idx1-ubyte.gz\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "29a320d657af4f6d9b112ef7142b1efe",
              "version_minor": 0,
              "version_major": 2
            },
            "text/plain": [
              "HBox(children=(FloatProgress(value=1.0, bar_style='info', max=1.0), HTML(value='')))"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "Extracting /content/data/MNIST/raw/train-labels-idx1-ubyte.gz to /content/data/MNIST/raw\n",
            "Downloading http://yann.lecun.com/exdb/mnist/t10k-images-idx3-ubyte.gz to /content/data/MNIST/raw/t10k-images-idx3-ubyte.gz\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "fa72678cd3b84b068c306d1feb216aef",
              "version_minor": 0,
              "version_major": 2
            },
            "text/plain": [
              "HBox(children=(FloatProgress(value=1.0, bar_style='info', max=1.0), HTML(value='')))"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "Extracting /content/data/MNIST/raw/t10k-images-idx3-ubyte.gz to /content/data/MNIST/raw\n",
            "Downloading http://yann.lecun.com/exdb/mnist/t10k-labels-idx1-ubyte.gz to /content/data/MNIST/raw/t10k-labels-idx1-ubyte.gz\n",
            "\n",
            "\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "d653fd88f39a4562a58249a3465268d7",
              "version_minor": 0,
              "version_major": 2
            },
            "text/plain": [
              "HBox(children=(FloatProgress(value=1.0, bar_style='info', max=1.0), HTML(value='')))"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "Extracting /content/data/MNIST/raw/t10k-labels-idx1-ubyte.gz to /content/data/MNIST/raw\n",
            "Processing...\n",
            "Done!\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/torchvision/datasets/mnist.py:480: UserWarning: The given NumPy array is not writeable, and PyTorch does not support non-writeable tensors. This means you can write to the underlying (supposedly non-writeable) NumPy array using the tensor. You may want to copy the array to protect its data or make it writeable before converting it to a tensor. This type of warning will be suppressed for the rest of this program. (Triggered internally at  /pytorch/torch/csrc/utils/tensor_numpy.cpp:141.)\n",
            "  return torch.from_numpy(parsed.astype(m[2], copy=False)).view(*s)\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "p-fASIT0mhbC",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 297
        },
        "outputId": "d3d17139-74ab-4136-d37f-90e9f05fd6c8"
      },
      "source": [
        "# sample images를 확인합니다.\r\n",
        "from torchvision import utils\r\n",
        "import matplotlib.pyplot as plt\r\n",
        "import numpy as np\r\n",
        "%matplotlib inline\r\n",
        "\r\n",
        "# training data를 추출합니다.\r\n",
        "x_train, y_train = train_data.data, train_data.targets\r\n",
        "\r\n",
        "# val data를 추출합니다.\r\n",
        "x_val, y_val = val_data.data, val_data.targets\r\n",
        "\r\n",
        "# 차원을 추가하여 B*C*H*W 가 되도록 합니다.\r\n",
        "if len(x_train.shape) == 3:\r\n",
        "    x_train = x_train.unsqueeze(1)\r\n",
        "\r\n",
        "if len(x_val.shape) == 3:\r\n",
        "    x_val = x_val.unsqueeze(1)\r\n",
        "\r\n",
        "# tensor를 image로 변경하는 함수를 정의합니다.\r\n",
        "def show(img):\r\n",
        "    # tensor를 numpy array로 변경합니다.\r\n",
        "    npimg = img.numpy()\r\n",
        "    # C*H*W를 H*W*C로 변경합니다.\r\n",
        "    npimg_tr = npimg.transpose((1,2,0))\r\n",
        "    plt.imshow(npimg_tr, interpolation='nearest')\r\n",
        "\r\n",
        "# images grid를 생성하고 출력합니다.\r\n",
        "# 총 40개 이미지, 행당 8개 이미지를 출력합니다.\r\n",
        "x_grid = utils.make_grid(x_train[:40], nrow=8, padding=2)\r\n",
        "\r\n",
        "show(x_grid)"
      ],
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\n",
            "\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7CBlZ4vd3a3f"
      },
      "source": [
        "# data loader 를 생성합니다.\r\n",
        "from torch.utils.data import DataLoader\r\n",
        "\r\n",
        "train_dl = DataLoader(train_data, batch_size=32, shuffle=True)\r\n",
        "val_dl = DataLoader(val_data, batch_size=32)"
      ],
      "execution_count": 7,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "uVjYZCRP35VR"
      },
      "source": [
        "# LeNet-5 모델 구축하기\r\n",
        "data 전처리는 끝났습니다.\r\n",
        "이제 LeNet-5 모델을 구현하겠습니다."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "lFdG2DWm334y",
        "outputId": "4dc95019-5255-4111-ca8e-69617aab1c2b"
      },
      "source": [
        "from torch import nn\r\n",
        "import torch.nn.functional as F\r\n",
        "\r\n",
        "class LeNet_5(nn.Module):\r\n",
        "    def __init__(self):\r\n",
        "        super(LeNet_5,self).__init__()\r\n",
        "        self.conv1 = nn.Conv2d(1, 6, kernel_size=5, stride=1)\r\n",
        "        self.conv2 = nn.Conv2d(6, 16, kernel_size=5, stride=1)\r\n",
        "        self.conv3 = nn.Conv2d(16, 120, kernel_size=5, stride=1)\r\n",
        "        self.fc1 = nn.Linear(120, 84)\r\n",
        "        self.fc2 = nn.Linear(84, 10)\r\n",
        "\r\n",
        "    def forward(self, x):\r\n",
        "        x = F.tanh(self.conv1(x))\r\n",
        "        x = F.avg_pool2d(x, 2, 2)\r\n",
        "        x = F.tanh(self.conv2(x))\r\n",
        "        x = F.avg_pool2d(x, 2, 2)\r\n",
        "        x = F.tanh(self.conv3(x))\r\n",
        "        x = x.view(-1, 120)\r\n",
        "        x = F.tanh(self.fc1(x))\r\n",
        "        x = self.fc2(x)\r\n",
        "        return F.softmax(x, dim=1)\r\n",
        "\r\n",
        "model = LeNet_5()\r\n",
        "print(model)"
      ],
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "LeNet_5(\n",
            "  (conv1): Conv2d(1, 6, kernel_size=(5, 5), stride=(1, 1))\n",
            "  (conv2): Conv2d(6, 16, kernel_size=(5, 5), stride=(1, 1))\n",
            "  (conv3): Conv2d(16, 120, kernel_size=(5, 5), stride=(1, 1))\n",
            "  (fc1): Linear(in_features=120, out_features=84, bias=True)\n",
            "  (fc2): Linear(in_features=84, out_features=10, bias=True)\n",
            ")\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "XVKmZqhe7VdN",
        "outputId": "78b0acd4-6922-42ec-f435-e754a260f483"
      },
      "source": [
        "# 모델을 CUDA로 전달합니다.\r\n",
        "model.to(device)\r\n",
        "print(next(model.parameters()).device)"
      ],
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "cuda:0\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "S_ijm8P0803G",
        "outputId": "18632175-11bc-44e0-eb72-5483013d0dca"
      },
      "source": [
        "# 모델 summary를 확인합니다.\r\n",
        "from torchsummary import summary\r\n",
        "summary(model, input_size=(1, 32, 32))"
      ],
      "execution_count": 10,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "----------------------------------------------------------------\n",
            "        Layer (type)               Output Shape         Param #\n",
            "================================================================\n",
            "            Conv2d-1            [-1, 6, 28, 28]             156\n",
            "            Conv2d-2           [-1, 16, 10, 10]           2,416\n",
            "            Conv2d-3            [-1, 120, 1, 1]          48,120\n",
            "            Linear-4                   [-1, 84]          10,164\n",
            "            Linear-5                   [-1, 10]             850\n",
            "================================================================\n",
            "Total params: 61,706\n",
            "Trainable params: 61,706\n",
            "Non-trainable params: 0\n",
            "----------------------------------------------------------------\n",
            "Input size (MB): 0.00\n",
            "Forward/backward pass size (MB): 0.05\n",
            "Params size (MB): 0.24\n",
            "Estimated Total Size (MB): 0.29\n",
            "----------------------------------------------------------------\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/torch/nn/functional.py:1628: UserWarning: nn.functional.tanh is deprecated. Use torch.tanh instead.\n",
            "  warnings.warn(\"nn.functional.tanh is deprecated. Use torch.tanh instead.\")\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "6TOxPM01BIYE"
      },
      "source": [
        "# loss function, optimizer 정의하기"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lz3Kw-cR89yv"
      },
      "source": [
        "# loss function 정의합니다.\r\n",
        "loss_func = nn.CrossEntropyLoss(reduction='sum')"
      ],
      "execution_count": 11,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "w8kg4U60BtW0"
      },
      "source": [
        "# optimizer 정의합니다.\r\n",
        "from torch import optim\r\n",
        "opt = optim.Adam(model.parameters(), lr=0.001)\r\n",
        "\r\n",
        "# 현재 lr을 계산하는 함수를 정의합니다.\r\n",
        "def get_lr(opt):\r\n",
        "    for param_group in opt.param_groups:\r\n",
        "        return param_group['lr']\r\n",
        "    \r\n",
        "# 러닝레이트 스케쥴러를 정의합니다.\r\n",
        "from torch.optim.lr_scheduler import CosineAnnealingLR\r\n",
        "lr_scheduler = CosineAnnealingLR(opt, T_max=2, eta_min=1e-05)"
      ],
      "execution_count": 12,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "gpWm4vz4D4eL"
      },
      "source": [
        "# 학습을 위한 함수 정의"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1NKkLHtWCoNF"
      },
      "source": [
        "# 배치당 performance metric 을 계산하는 함수 정의\r\n",
        "def metrics_batch(output, target):\r\n",
        "    pred = output.argmax(dim=1, keepdim=True)\r\n",
        "    corrects = pred.eq(target.view_as(pred)).sum().item()\r\n",
        "    return corrects"
      ],
      "execution_count": 13,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6IpaGjkcGJqF"
      },
      "source": [
        "# 배치당 loss를 계산하는 함수를 정의\r\n",
        "def loss_batch(loss_func, output, target, opt=None):\r\n",
        "    loss = loss_func(output, target)\r\n",
        "    metric_b = metrics_batch(output, target)\r\n",
        "    if opt is not None:\r\n",
        "        opt.zero_grad()\r\n",
        "        loss.backward()\r\n",
        "        opt.step()\r\n",
        "    return loss.item(), metric_b"
      ],
      "execution_count": 14,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "QhIZlp3tGMzP"
      },
      "source": [
        "# epoch당 loss와 performance metric을 계산하는 함수 정의\r\n",
        "def loss_epoch(model, loss_func, dataset_dl, sanity_check=False, opt=None):\r\n",
        "    running_loss = 0.0\r\n",
        "    running_metric = 0.0\r\n",
        "    len_data = len(dataset_dl.dataset)\r\n",
        "\r\n",
        "    for xb, yb in dataset_dl:\r\n",
        "        xb = xb.type(torch.float).to(device)\r\n",
        "        yb = yb.to(device)\r\n",
        "        output = model(xb)\r\n",
        "        loss_b, metric_b = loss_batch(loss_func, output, yb, opt)\r\n",
        "        running_loss += loss_b\r\n",
        "\r\n",
        "        if metric_b is not None:\r\n",
        "            running_metric += metric_b\r\n",
        "        \r\n",
        "        if sanity_check is True: # sanity_check가 True이면 1epoch만 학습합니다.\r\n",
        "            break\r\n",
        "\r\n",
        "    loss = running_loss / float(len_data)\r\n",
        "    metric = running_metric / float(len_data)\r\n",
        "    return loss, metric"
      ],
      "execution_count": 15,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7SPClyXWGxKP"
      },
      "source": [
        "# train_val 함수 정의\r\n",
        "def train_val(model, params):\r\n",
        "    num_epochs = params['num_epochs']\r\n",
        "    loss_func = params['loss_func']\r\n",
        "    opt = params['optimizer']\r\n",
        "    train_dl = params['train_dl']\r\n",
        "    val_dl = params['val_dl']\r\n",
        "    sanity_check = params['sanity_check']\r\n",
        "    lr_scheduler = params['lr_scheduler']\r\n",
        "    path2weights = params['path2weights']\r\n",
        "\r\n",
        "    loss_history = {\r\n",
        "        'train': [],\r\n",
        "        'val': [],\r\n",
        "    }\r\n",
        "\r\n",
        "    metric_history = {\r\n",
        "        'train': [],\r\n",
        "        'val': [],\r\n",
        "    }\r\n",
        "\r\n",
        "    # best model parameter를 저장합니다.\r\n",
        "    best_model_wts = copy.deepcopy(model.state_dict())\r\n",
        "    best_loss = float('inf')\r\n",
        "\r\n",
        "    for epoch in range(num_epochs):\r\n",
        "        current_lr = get_lr(opt)\r\n",
        "        print('Epoch {}/{}, current lr={}'.format(epoch, num_epochs-1, current_lr))\r\n",
        "        model.train()\r\n",
        "        train_loss, train_metric = loss_epoch(model, loss_func, train_dl, sanity_check, opt)\r\n",
        "\r\n",
        "        loss_history['train'].append(train_loss)\r\n",
        "        metric_history['train'].append(train_metric)\r\n",
        "\r\n",
        "        model.eval()\r\n",
        "        with torch.no_grad():\r\n",
        "            val_loss, val_metric = loss_epoch(model, loss_func, val_dl, sanity_check)\r\n",
        "            loss_history['val'].append(val_loss)\r\n",
        "            metric_history['val'].append(val_metric)\r\n",
        "\r\n",
        "        if val_loss < best_loss:\r\n",
        "            best_loss = val_loss\r\n",
        "            best_model_wts = copy.deepcopy(model.state_dict())\r\n",
        "            torch.save(model.state_dict(), path2weights)\r\n",
        "            print('Copied best model weights')\r\n",
        "\r\n",
        "        lr_scheduler.step()\r\n",
        "\r\n",
        "        print('train loss: %.6f, dev loss: %.6f, accuracy: %.2f' %(train_loss, val_loss, 100*val_metric))\r\n",
        "        print('-'*10)\r\n",
        "\r\n",
        "    # best model을 반환합니다.\r\n",
        "    model.load_state_dict(best_model_wts)\r\n",
        "    return model, loss_history, metric_history"
      ],
      "execution_count": 16,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "-KO-MY7XHG1u"
      },
      "source": [
        "# 모델 학습하기"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7vjW8hY3HFMq"
      },
      "source": [
        "import copy\r\n",
        "import os\r\n",
        "\r\n",
        "# 학습된 모델의 가중치를 저장할 폴더를 만듭니다.\r\n",
        "os.makedirs('/models', exist_ok=True)\r\n",
        "\r\n",
        "# 하이퍼 파라미터를 설정합니다.\r\n",
        "params_train={\r\n",
        " \"num_epochs\": 3,\r\n",
        " \"optimizer\": opt,\r\n",
        " \"loss_func\": loss_func,\r\n",
        " \"train_dl\": train_dl,\r\n",
        " \"val_dl\": val_dl,\r\n",
        " \"sanity_check\": False,\r\n",
        " \"lr_scheduler\": lr_scheduler,\r\n",
        " \"path2weights\": \"/models/LeNet-5.pt\",\r\n",
        "}"
      ],
      "execution_count": 17,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "RrmGDOPjHcGz",
        "outputId": "31495918-ab99-4eb5-d4ac-ed5bb212238a"
      },
      "source": [
        "# 모델을 학습합니다.\r\n",
        "model,loss_hist,metric_hist=train_val(model,params_train)"
      ],
      "execution_count": 18,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/torch/nn/functional.py:1628: UserWarning: nn.functional.tanh is deprecated. Use torch.tanh instead.\n",
            "  warnings.warn(\"nn.functional.tanh is deprecated. Use torch.tanh instead.\")\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch 0/2, current lr=0.001\n",
            "Copied best model weights\n",
            "train loss: 1.562686, dev loss: 1.503760, accuracy: 96.11\n",
            "----------\n",
            "Epoch 1/2, current lr=0.000505\n",
            "Copied best model weights\n",
            "train loss: 1.496906, dev loss: 1.489321, accuracy: 97.44\n",
            "----------\n",
            "Epoch 2/2, current lr=1e-05\n",
            "Copied best model weights\n",
            "train loss: 1.486404, dev loss: 1.487152, accuracy: 97.64\n",
            "----------\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "iiB4hBMyHkJO",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "outputId": "4a84b75b-95d4-4fbc-916a-cf12d497922f"
      },
      "source": [
        "num_epochs=params_train[\"num_epochs\"]\r\n",
        "\r\n",
        "plt.title(\"Train-Val Loss\")\r\n",
        "plt.plot(range(1,num_epochs+1),loss_hist[\"train\"],label=\"train\")\r\n",
        "plt.plot(range(1,num_epochs+1),loss_hist[\"val\"],label=\"val\")\r\n",
        "plt.ylabel(\"Loss\")\r\n",
        "plt.xlabel(\"Training Epochs\")\r\n",
        "plt.legend()\r\n",
        "plt.show()"
      ],
      "execution_count": 20,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "id": "rDIQb-WjSaJg",
        "outputId": "d949ef60-2f4c-44af-a0ec-7fe448c2e9fc"
      },
      "source": [
        "# plot accuracy progress\r\n",
        "plt.title(\"Train-Val Accuracy\")\r\n",
        "plt.plot(range(1,num_epochs+1),metric_hist[\"train\"],label=\"train\")\r\n",
        "plt.plot(range(1,num_epochs+1),metric_hist[\"val\"],label=\"val\")\r\n",
        "plt.ylabel(\"Accuracy\")\r\n",
        "plt.xlabel(\"Training Epochs\")\r\n",
        "plt.legend()\r\n",
        "plt.show()"
      ],
      "execution_count": 21,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    }
  ]
}