Copy_of_Eva5_Assignment_6_Regularization_2.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Copy of Eva5 Assignment 6-Regularization_2.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU",
    "widgets": {
      "application/vnd.jupyter.widget-state+json": {
        "fbae6be0e95b43c1a2210c99faf69125": {
          "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_ef5e9b1eaba7494dbf9db7534ec0391a",
            "_model_module": "@jupyter-widgets/controls",
            "children": [
              "IPY_MODEL_8c3a4b48901a47de8ac806776fcb1400",
              "IPY_MODEL_b6b451639a0d4561aa59ae86fd4810e6"
            ]
          }
        },
        "ef5e9b1eaba7494dbf9db7534ec0391a": {
          "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
          }
        },
        "8c3a4b48901a47de8ac806776fcb1400": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "state": {
            "_view_name": "ProgressView",
            "style": "IPY_MODEL_1e25f9ff9fc04a20b497f1c45511bf0f",
            "_dom_classes": [],
            "description": "",
            "_model_name": "FloatProgressModel",
            "bar_style": "info",
            "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_573f1779d0e44dbda6d831cd8327cdc1"
          }
        },
        "b6b451639a0d4561aa59ae86fd4810e6": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_57fff04375c84761beda805944942e0e",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": " 9920512/? [00:30&lt;00:00, 1002467.71it/s]",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_accf783a7f364215b8d9e72341757b43"
          }
        },
        "1e25f9ff9fc04a20b497f1c45511bf0f": {
          "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"
          }
        },
        "573f1779d0e44dbda6d831cd8327cdc1": {
          "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
          }
        },
        "57fff04375c84761beda805944942e0e": {
          "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"
          }
        },
        "accf783a7f364215b8d9e72341757b43": {
          "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
          }
        },
        "996264dab392409fa5a33e8d9ab1c359": {
          "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_7addc3be8e2340018c2d293e86769fa9",
            "_model_module": "@jupyter-widgets/controls",
            "children": [
              "IPY_MODEL_e52e4ebf669343639aa4f793f68e5380",
              "IPY_MODEL_cec63b38264546758a938ec44a1bc6ae"
            ]
          }
        },
        "7addc3be8e2340018c2d293e86769fa9": {
          "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
          }
        },
        "e52e4ebf669343639aa4f793f68e5380": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "state": {
            "_view_name": "ProgressView",
            "style": "IPY_MODEL_0b15cec38fde4170a520b5e05185040c",
            "_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_1f48078881b74193b99e6df082363f02"
          }
        },
        "cec63b38264546758a938ec44a1bc6ae": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_526ab50a939f412280b015d63955d68e",
            "_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, 41468.95it/s]",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_1122367bbcff4a46b36e069722e648b7"
          }
        },
        "0b15cec38fde4170a520b5e05185040c": {
          "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"
          }
        },
        "1f48078881b74193b99e6df082363f02": {
          "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
          }
        },
        "526ab50a939f412280b015d63955d68e": {
          "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"
          }
        },
        "1122367bbcff4a46b36e069722e648b7": {
          "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
          }
        },
        "3088c2e7a7b14aadbdda553ecd46dd2a": {
          "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_d35a2e1efaa140a29362b13b8b31fe91",
            "_model_module": "@jupyter-widgets/controls",
            "children": [
              "IPY_MODEL_07386b57684d4d628d17ede7ec482df6",
              "IPY_MODEL_618dff3a3ec44b6392c53ddd81e0519e"
            ]
          }
        },
        "d35a2e1efaa140a29362b13b8b31fe91": {
          "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
          }
        },
        "07386b57684d4d628d17ede7ec482df6": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "state": {
            "_view_name": "ProgressView",
            "style": "IPY_MODEL_cf3a20896f694d66ae7dbbf5ed7c3686",
            "_dom_classes": [],
            "description": "",
            "_model_name": "FloatProgressModel",
            "bar_style": "info",
            "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_38ef28032ed84a7fb7b9c0659d548d00"
          }
        },
        "618dff3a3ec44b6392c53ddd81e0519e": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_0697a4aa29a545baaf76cb0d44b320da",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": " 1654784/? [00:15&lt;00:00, 248096.10it/s]",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_9bed8e04aa7845d3b88b974e70b261fa"
          }
        },
        "cf3a20896f694d66ae7dbbf5ed7c3686": {
          "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"
          }
        },
        "38ef28032ed84a7fb7b9c0659d548d00": {
          "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
          }
        },
        "0697a4aa29a545baaf76cb0d44b320da": {
          "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"
          }
        },
        "9bed8e04aa7845d3b88b974e70b261fa": {
          "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
          }
        },
        "58362950493c4854bde5ff95e3b78813": {
          "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_89c1b037ff594023bd0e78128541f883",
            "_model_module": "@jupyter-widgets/controls",
            "children": [
              "IPY_MODEL_af064033125b49ff9ecbf0a1920e6116",
              "IPY_MODEL_0d1a0f5c1e3942548ce9b152fceb373b"
            ]
          }
        },
        "89c1b037ff594023bd0e78128541f883": {
          "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
          }
        },
        "af064033125b49ff9ecbf0a1920e6116": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "state": {
            "_view_name": "ProgressView",
            "style": "IPY_MODEL_ae6047a23689453486ae396d558d3ee2",
            "_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_bd771a42c9a44fceb58fe85c106bed1e"
          }
        },
        "0d1a0f5c1e3942548ce9b152fceb373b": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_ac001c761e1e41d89cffe94d89a4e96d",
            "_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, 10044.38it/s]",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_8bbc81e905474d88b98dd6c049140c13"
          }
        },
        "ae6047a23689453486ae396d558d3ee2": {
          "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"
          }
        },
        "bd771a42c9a44fceb58fe85c106bed1e": {
          "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
          }
        },
        "ac001c761e1e41d89cffe94d89a4e96d": {
          "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"
          }
        },
        "8bbc81e905474d88b98dd6c049140c13": {
          "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/AyusheeMittal/legendary-guide/blob/master/Copy_of_Eva5_Assignment_6_Regularization_2.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "zZaxUFh-upV8",
        "colab_type": "text"
      },
      "source": [
        "**Team Members**\n",
        "\n",
        "Avneesh Nolkha\n",
        "\n",
        "Ayushee Mittal\n",
        "\n",
        "Smita Sasindran\n",
        "\n",
        "Vishesh Sethi\n",
        "\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9Pv-t5JpdDS0",
        "colab_type": "text"
      },
      "source": [
        "## **Importing Libraries**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "wGRJvQsUc3DC",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from __future__ import print_function\n",
        "import torch\n",
        "import torch.nn as nn\n",
        "import torch.nn.functional as F\n",
        "import torch.optim as optim\n",
        "from torchvision import datasets, transforms\n",
        "from torch.optim.lr_scheduler import StepLR"
      ],
      "execution_count": 1,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "U2GI9sAFdIQJ",
        "colab_type": "text"
      },
      "source": [
        "## **Transforms**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "EX9nSD-C_e2d",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Train Phase transformations\n",
        "train_transforms = transforms.Compose([\n",
        "                                      #  transforms.Resize((28, 28)),\n",
        "                                      #  transforms.ColorJitter(brightness=0.10, contrast=0.1, saturation=0.10, hue=0.1),\n",
        "                                      transforms.RandomRotation((-7.0, 7.0), fill=(1,)),\n",
        "                                       transforms.ToTensor(),\n",
        "                                       transforms.Normalize((0.1307,), (0.3081,)) # The mean and std have to be sequences (e.g., tuples), therefore you should add a comma after the values. \n",
        "                                       # Note the difference between (0.1307) and (0.1307,)\n",
        "                                       ])\n",
        "\n",
        "# Test Phase transformations\n",
        "test_transforms = transforms.Compose([\n",
        "                                      #  transforms.Resize((28, 28)),\n",
        "                                      #  transforms.ColorJitter(brightness=0.10, contrast=0.1, saturation=0.10, hue=0.1),\n",
        "                                       transforms.ToTensor(),\n",
        "                                       transforms.Normalize((0.1307,), (0.3081,))\n",
        "                                       ])"
      ],
      "execution_count": 2,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Uz8iGrH0dUxY",
        "colab_type": "text"
      },
      "source": [
        "## **Dataset**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "j9UB_t0s_pKe",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 386,
          "referenced_widgets": [
            "fbae6be0e95b43c1a2210c99faf69125",
            "ef5e9b1eaba7494dbf9db7534ec0391a",
            "8c3a4b48901a47de8ac806776fcb1400",
            "b6b451639a0d4561aa59ae86fd4810e6",
            "1e25f9ff9fc04a20b497f1c45511bf0f",
            "573f1779d0e44dbda6d831cd8327cdc1",
            "57fff04375c84761beda805944942e0e",
            "accf783a7f364215b8d9e72341757b43",
            "996264dab392409fa5a33e8d9ab1c359",
            "7addc3be8e2340018c2d293e86769fa9",
            "e52e4ebf669343639aa4f793f68e5380",
            "cec63b38264546758a938ec44a1bc6ae",
            "0b15cec38fde4170a520b5e05185040c",
            "1f48078881b74193b99e6df082363f02",
            "526ab50a939f412280b015d63955d68e",
            "1122367bbcff4a46b36e069722e648b7",
            "3088c2e7a7b14aadbdda553ecd46dd2a",
            "d35a2e1efaa140a29362b13b8b31fe91",
            "07386b57684d4d628d17ede7ec482df6",
            "618dff3a3ec44b6392c53ddd81e0519e",
            "cf3a20896f694d66ae7dbbf5ed7c3686",
            "38ef28032ed84a7fb7b9c0659d548d00",
            "0697a4aa29a545baaf76cb0d44b320da",
            "9bed8e04aa7845d3b88b974e70b261fa",
            "58362950493c4854bde5ff95e3b78813",
            "89c1b037ff594023bd0e78128541f883",
            "af064033125b49ff9ecbf0a1920e6116",
            "0d1a0f5c1e3942548ce9b152fceb373b",
            "ae6047a23689453486ae396d558d3ee2",
            "bd771a42c9a44fceb58fe85c106bed1e",
            "ac001c761e1e41d89cffe94d89a4e96d",
            "8bbc81e905474d88b98dd6c049140c13"
          ]
        },
        "outputId": "09316aa0-135c-4d32-d386-e4a98e00b92a"
      },
      "source": [
        "train = datasets.MNIST('./data', train=True, download=True, transform=train_transforms)\n",
        "test = datasets.MNIST('./data', train=False, download=True, transform=test_transforms)"
      ],
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Downloading http://yann.lecun.com/exdb/mnist/train-images-idx3-ubyte.gz to ./data/MNIST/raw/train-images-idx3-ubyte.gz\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "fbae6be0e95b43c1a2210c99faf69125",
              "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 ./data/MNIST/raw/train-images-idx3-ubyte.gz to ./data/MNIST/raw\n",
            "Downloading http://yann.lecun.com/exdb/mnist/train-labels-idx1-ubyte.gz to ./data/MNIST/raw/train-labels-idx1-ubyte.gz\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "996264dab392409fa5a33e8d9ab1c359",
              "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 ./data/MNIST/raw/train-labels-idx1-ubyte.gz to ./data/MNIST/raw\n",
            "Downloading http://yann.lecun.com/exdb/mnist/t10k-images-idx3-ubyte.gz to ./data/MNIST/raw/t10k-images-idx3-ubyte.gz\n",
            "\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "3088c2e7a7b14aadbdda553ecd46dd2a",
              "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 ./data/MNIST/raw/t10k-images-idx3-ubyte.gz to ./data/MNIST/raw\n",
            "Downloading http://yann.lecun.com/exdb/mnist/t10k-labels-idx1-ubyte.gz to ./data/MNIST/raw/t10k-labels-idx1-ubyte.gz\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "58362950493c4854bde5ff95e3b78813",
              "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 ./data/MNIST/raw/t10k-labels-idx1-ubyte.gz to ./data/MNIST/raw\n",
            "Processing...\n",
            "Done!\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/torchvision/datasets/mnist.py:469: 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": "markdown",
      "metadata": {
        "id": "Bblr6MqYdXm_",
        "colab_type": "text"
      },
      "source": [
        "## **Dataloader**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Ja6Yrcq9_w22",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "a348984f-6c66-400f-9423-376665056fb9"
      },
      "source": [
        "SEED = 1\n",
        "\n",
        "# CUDA?\n",
        "cuda = torch.cuda.is_available()\n",
        "print(\"CUDA Available?\", cuda)\n",
        "\n",
        "# For reproducibility\n",
        "torch.manual_seed(SEED)\n",
        "\n",
        "if cuda:\n",
        "    torch.cuda.manual_seed(SEED)\n",
        "\n",
        "# dataloader arguments - something you'll fetch these from cmdprmt\n",
        "dataloader_args = dict(shuffle=True, batch_size=128, num_workers=4, pin_memory=True) if cuda else dict(shuffle=True, batch_size=64)\n",
        "\n",
        "# train dataloader\n",
        "train_loader = torch.utils.data.DataLoader(train, **dataloader_args)\n",
        "\n",
        "# test dataloader\n",
        "test_loader = torch.utils.data.DataLoader(test, **dataloader_args)"
      ],
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "CUDA Available? True\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ZrsymfKpCnXV",
        "colab_type": "text"
      },
      "source": [
        "## **Ghost Batch Normalization classes**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BbF_tA_eCnr2",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "c1609dd2-7264-4fe9-cc93-b67f2d0a9959"
      },
      "source": [
        "class BatchNorm(nn.BatchNorm2d):\n",
        "    def __init__(self, num_features, eps=1e-05, momentum=0.1, weight=True, bias=True):\n",
        "        super().__init__(num_features, eps=eps, momentum=momentum)\n",
        "        self.weight.data.fill_(1.0)\n",
        "        self.bias.data.fill_(0.0)\n",
        "        self.weight.requires_grad = weight\n",
        "        self.bias.requires_grad = bias\n",
        "\n",
        "\n",
        "class GhostBatchNorm(BatchNorm):\n",
        "    def __init__(self, num_features, num_splits, **kw):\n",
        "        super().__init__(num_features, **kw)\n",
        "        self.num_splits = num_splits\n",
        "        self.register_buffer('running_mean', torch.zeros(num_features * self.num_splits))\n",
        "        self.register_buffer('running_var', torch.ones(num_features * self.num_splits))\n",
        "\n",
        "    def train(self, mode=True):\n",
        "        if (self.training is True) and (mode is False):  # lazily collate stats when we are going to use them\n",
        "            self.running_mean = torch.mean(self.running_mean.view(self.num_splits, self.num_features), dim=0).repeat(\n",
        "                self.num_splits)\n",
        "            self.running_var = torch.mean(self.running_var.view(self.num_splits, self.num_features), dim=0).repeat(\n",
        "                self.num_splits)\n",
        "        return super().train(mode)\n",
        "\n",
        "    def forward(self, input):\n",
        "        N, C, H, W = input.shape\n",
        "        if self.training or not self.track_running_stats:\n",
        "            return F.batch_norm(\n",
        "                input.view(-1, C * self.num_splits, H, W), self.running_mean, self.running_var,\n",
        "                self.weight.repeat(self.num_splits), self.bias.repeat(self.num_splits),\n",
        "                True, self.momentum, self.eps).view(N, C, H, W)\n",
        "        else:\n",
        "            return F.batch_norm(\n",
        "                input, self.running_mean[:self.num_features], self.running_var[:self.num_features],\n",
        "                self.weight, self.bias, False, self.momentum, self.eps)\n",
        "\n"
      ],
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "io7IDVsvdhVB",
        "colab_type": "text"
      },
      "source": [
        "## **Model**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "K8r-6s9mHAPw",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "dropout_value = 0.09\n",
        "class Net(nn.Module):\n",
        "    def __init__(self, gbn, gbn_split):\n",
        "        super(Net, self).__init__()\n",
        "\n",
        "        self.gbn = gbn\n",
        "        self.gbn_split = gbn_split\n",
        "\n",
        "        print(\"Is GBN run:\", gbn)\n",
        "        # Add BatchNorm or GhostBatchNorm\n",
        "        def BNLayer(n_features):\n",
        "            bn = nn.BatchNorm2d(n_features) if not self.gbn else GhostBatchNorm(n_features, self.gbn_split) \n",
        "            return bn\n",
        "\n",
        "\n",
        "        # Input Block\n",
        "        self.convblock1 = nn.Sequential(\n",
        "            nn.Conv2d(in_channels=1, out_channels=10, kernel_size=(3, 3), padding=0, bias=False),\n",
        "            nn.ReLU(),\n",
        "            #nn.BatchNorm2d(10),\n",
        "            BNLayer(10),\n",
        "            nn.Dropout(dropout_value)\n",
        "        ) # output_size = 26\n",
        "\n",
        "        # CONVOLUTION BLOCK 1\n",
        "        self.convblock2 = nn.Sequential(\n",
        "            nn.Conv2d(in_channels=10, out_channels=16, kernel_size=(3, 3), padding=0, bias=False),\n",
        "            nn.ReLU(),\n",
        "            #nn.BatchNorm2d(16),\n",
        "            BNLayer(16),\n",
        "            nn.Dropout(dropout_value)\n",
        "        ) # output_size = 24\n",
        "\n",
        "        # TRANSITION BLOCK 1\n",
        "        self.convblock3 = nn.Sequential(\n",
        "            nn.Conv2d(in_channels=16, out_channels=10, kernel_size=(1, 1), padding=0, bias=False),\n",
        "            nn.ReLU(),\n",
        "            #nn.BatchNorm2d(10)\n",
        "            nn.BatchNorm2d(10)\n",
        "        ) # output_size = 24\n",
        "        self.pool1 = nn.MaxPool2d(2, 2) # output_size = 12\n",
        "\n",
        "        # CONVOLUTION BLOCK 2\n",
        "        self.convblock4 = nn.Sequential(\n",
        "            nn.Conv2d(in_channels=10, out_channels=10, kernel_size=(3, 3), padding=0, bias=False),\n",
        "            nn.ReLU(),            \n",
        "            #nn.BatchNorm2d(10),\n",
        "            nn.BatchNorm2d(10),\n",
        "            nn.Dropout(dropout_value)\n",
        "        ) # output_size = 10\n",
        "        self.convblock5 = nn.Sequential(\n",
        "            nn.Conv2d(in_channels=10, out_channels=12, kernel_size=(3, 3), padding=0, bias=False),\n",
        "            nn.ReLU(),            \n",
        "            #nn.BatchNorm2d(12),\n",
        "            nn.BatchNorm2d(12),\n",
        "            nn.Dropout(dropout_value)\n",
        "        ) # output_size = 8\n",
        "        self.convblock6 = nn.Sequential(\n",
        "            nn.Conv2d(in_channels=12, out_channels=16, kernel_size=(3, 3), padding=0, bias=False),\n",
        "            nn.ReLU(),            \n",
        "            #nn.BatchNorm2d(16),\n",
        "            nn.BatchNorm2d(16),\n",
        "            nn.Dropout(dropout_value)\n",
        "        ) # output_size = 6\n",
        "        self.convblock7 = nn.Sequential(\n",
        "            nn.Conv2d(in_channels=16, out_channels=20, kernel_size=(3, 3), padding=1, bias=False),\n",
        "            nn.ReLU(),            \n",
        "            #nn.BatchNorm2d(20),\n",
        "            nn.BatchNorm2d(20),\n",
        "            nn.Dropout(dropout_value)\n",
        "        ) # output_size = 6\n",
        "        \n",
        "        # OUTPUT BLOCK\n",
        "        self.gap = nn.Sequential(\n",
        "            nn.AvgPool2d(kernel_size=6)\n",
        "        ) # output_size = 1\n",
        "\n",
        "        self.convblock8 = nn.Sequential(\n",
        "            nn.Conv2d(in_channels=20, out_channels=10, kernel_size=(1, 1), padding=0, bias=False),\n",
        "            # nn.BatchNorm2d(10),\n",
        "            # nn.ReLU(),\n",
        "            # nn.Dropout(dropout_value)\n",
        "        ) \n",
        "\n",
        "        self.dropout = nn.Dropout(dropout_value)\n",
        "\n",
        "    def forward(self, x):\n",
        "        x = self.convblock1(x)\n",
        "        x = self.convblock2(x)\n",
        "        x = self.convblock3(x)\n",
        "        x = self.pool1(x)\n",
        "        x = self.convblock4(x)\n",
        "        x = self.convblock5(x)\n",
        "        x = self.convblock6(x)\n",
        "        x = self.convblock7(x)\n",
        "        x = self.gap(x)        \n",
        "        x = self.convblock8(x)\n",
        "\n",
        "        x = x.view(-1, 10)\n",
        "        return F.log_softmax(x, dim=-1)"
      ],
      "execution_count": 6,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "J5BNiZvn_4Bn",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#Model Block 1\n",
        "class Net(nn.Module):\n",
        "    def __init__(self, gbn, gbn_split):\n",
        "        super(Net, self).__init__()\n",
        "        self.gbn = gbn\n",
        "        self.gbn_split = gbn_split\n",
        "\n",
        "        print(\"Is GBN run:\", gbn)\n",
        "        # Add BatchNorm or GhostBatchNorm\n",
        "        def BNLayer(n_features):\n",
        "            bn = nn.BatchNorm2d(n_features) if not self.gbn else GhostBatchNorm(n_features, self.gbn_split) \n",
        "            return bn\n",
        "\n",
        "        # Input Block\n",
        "        self.convblock1 = nn.Sequential(\n",
        "            nn.Conv2d(in_channels=1, out_channels=10, kernel_size=(3, 3), padding=0, bias=False),\n",
        "            nn.ReLU(),\n",
        "            #nn.BatchNorm2d(10)\n",
        "            BNLayer(10)\n",
        "        ) # output_size = 26\n",
        "\n",
        "        # CONVOLUTION BLOCK 1\n",
        "        self.convblock2 = nn.Sequential(\n",
        "            nn.Conv2d(in_channels=10, out_channels=10, kernel_size=(3, 3), padding=0, bias=False),\n",
        "            nn.ReLU(),\n",
        "            #nn.BatchNorm2d(10)\n",
        "            BNLayer(10)\n",
        "        ) # output_size = 24\n",
        "        self.convblock3 = nn.Sequential(\n",
        "            nn.Conv2d(in_channels=10, out_channels=16, kernel_size=(3, 3), padding=0, bias=False),\n",
        "            nn.ReLU(),\n",
        "            #nn.BatchNorm2d(16)\n",
        "            BNLayer(16)\n",
        "        ) # output_size = 22\n",
        "\n",
        "        # TRANSITION BLOCK 1\n",
        "        self.pool1 = nn.MaxPool2d(2, 2) # output_size = 11\n",
        "        self.convblock4 = nn.Sequential(\n",
        "            nn.Conv2d(in_channels=16, out_channels=12, kernel_size=(1, 1), padding=0, bias=False),\n",
        "            nn.ReLU(),\n",
        "            #nn.BatchNorm2d(12)\n",
        "            BNLayer(12)\n",
        "        ) # output_size = 11\n",
        "\n",
        "        # CONVOLUTION BLOCK 2\n",
        "        self.convblock5 = nn.Sequential(\n",
        "            nn.Conv2d(in_channels=12, out_channels=16, kernel_size=(3, 3), padding=0, bias=False),\n",
        "            nn.ReLU(),\n",
        "            #nn.BatchNorm2d(16)\n",
        "            BNLayer(16)\n",
        "        ) # output_size = 9\n",
        "        self.convblock6 = nn.Sequential(\n",
        "            nn.Conv2d(in_channels=16, out_channels=20, kernel_size=(3, 3), padding=0, bias=False),\n",
        "            nn.ReLU(),\n",
        "            #nn.BatchNorm2d(20)\n",
        "            BNLayer(20)\n",
        "        ) # output_size = 7\n",
        "\n",
        "        self.gap = nn.Sequential(\n",
        "            nn.AvgPool2d(kernel_size=7)\n",
        "        ) # output_size = 1\n",
        "\n",
        "        self.convblock7 = nn.Sequential(\n",
        "            nn.Conv2d(in_channels=20, out_channels=16, kernel_size=(1, 1), padding=0, bias=False),\n",
        "            nn.ReLU(),\n",
        "            #nn.BatchNorm2d(16)\n",
        "            BNLayer(16)\n",
        "        ) # output_size = 7\n",
        "        self.convblock8 = nn.Sequential(\n",
        "            nn.Conv2d(in_channels=16, out_channels=10, kernel_size=(1, 1), padding=0, bias=False),\n",
        "        ) # output_size = 7\n",
        "\n",
        "        self.dropout = nn.Dropout2d(0.25)\n",
        "\n",
        "    def forward(self, x):\n",
        "        x = self.convblock1(x)\n",
        "        x = self.convblock2(x)\n",
        "        x = self.convblock3(x)\n",
        "        x = self.pool1(x)\n",
        "        x = self.convblock4(x)\n",
        "        x = self.convblock5(x)\n",
        "        x = self.convblock6(x)\n",
        "        x = self.gap(x)\n",
        "        x = self.convblock7(x)\n",
        "        x = self.convblock8(x)\n",
        "        x = x.view(-1, 10)\n",
        "        return F.log_softmax(x, dim=-1)"
      ],
      "execution_count": 7,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "QlckNRvDdoD5",
        "colab_type": "text"
      },
      "source": [
        "## **Model Params**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hVyKicBtdrEL",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 697
        },
        "outputId": "c096fab3-232e-4243-c811-556e42243bd2"
      },
      "source": [
        "!pip install torchsummary\n",
        "from torchsummary import summary\n",
        "use_cuda = torch.cuda.is_available()\n",
        "device = torch.device(\"cuda\" if use_cuda else \"cpu\")\n",
        "print(device)\n",
        "model = Net(False, 1).to(device)\n",
        "summary(model, input_size=(1, 28, 28))"
      ],
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Requirement already satisfied: torchsummary in /usr/local/lib/python3.6/dist-packages (1.5.1)\n",
            "cuda\n",
            "Is GBN run: False\n",
            "----------------------------------------------------------------\n",
            "        Layer (type)               Output Shape         Param #\n",
            "================================================================\n",
            "            Conv2d-1           [-1, 10, 26, 26]              90\n",
            "              ReLU-2           [-1, 10, 26, 26]               0\n",
            "       BatchNorm2d-3           [-1, 10, 26, 26]              20\n",
            "            Conv2d-4           [-1, 10, 24, 24]             900\n",
            "              ReLU-5           [-1, 10, 24, 24]               0\n",
            "       BatchNorm2d-6           [-1, 10, 24, 24]              20\n",
            "            Conv2d-7           [-1, 16, 22, 22]           1,440\n",
            "              ReLU-8           [-1, 16, 22, 22]               0\n",
            "       BatchNorm2d-9           [-1, 16, 22, 22]              32\n",
            "        MaxPool2d-10           [-1, 16, 11, 11]               0\n",
            "           Conv2d-11           [-1, 12, 11, 11]             192\n",
            "             ReLU-12           [-1, 12, 11, 11]               0\n",
            "      BatchNorm2d-13           [-1, 12, 11, 11]              24\n",
            "           Conv2d-14             [-1, 16, 9, 9]           1,728\n",
            "             ReLU-15             [-1, 16, 9, 9]               0\n",
            "      BatchNorm2d-16             [-1, 16, 9, 9]              32\n",
            "           Conv2d-17             [-1, 20, 7, 7]           2,880\n",
            "             ReLU-18             [-1, 20, 7, 7]               0\n",
            "      BatchNorm2d-19             [-1, 20, 7, 7]              40\n",
            "        AvgPool2d-20             [-1, 20, 1, 1]               0\n",
            "           Conv2d-21             [-1, 16, 1, 1]             320\n",
            "             ReLU-22             [-1, 16, 1, 1]               0\n",
            "      BatchNorm2d-23             [-1, 16, 1, 1]              32\n",
            "           Conv2d-24             [-1, 10, 1, 1]             160\n",
            "================================================================\n",
            "Total params: 7,910\n",
            "Trainable params: 7,910\n",
            "Non-trainable params: 0\n",
            "----------------------------------------------------------------\n",
            "Input size (MB): 0.00\n",
            "Forward/backward pass size (MB): 0.56\n",
            "Params size (MB): 0.03\n",
            "Estimated Total Size (MB): 0.60\n",
            "----------------------------------------------------------------\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "fZcGpMCrTo_h",
        "colab_type": "text"
      },
      "source": [
        "## **Setting up parameters for 5 runs**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "UWns5AjjTtcF",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "run_params = [\n",
        "  {'model': 'L1_with_BN', 'l1': True, 'l2': False, 'gbn': False, 'gbn_split': 1, 'l_l1': 0.0000085},\n",
        "  {'model': 'L2_with_BN', 'l1': False, 'l2': True, 'gbn': False, 'gbn_split': 1, 'l_l1': 0},\n",
        "  {'model': 'L1_and_L2_with_BN', 'l1': True, 'l2': True, 'gbn': False, 'gbn_split': 1, 'l_l1': 0.00000000009},\n",
        "  {'model': 'GBN', 'l1': False, 'l2': False, 'gbn': True, 'gbn_split': 2, 'l_l1': 0},\n",
        "  {'model': 'L1_and_L2_with_GBN', 'l1': True, 'l2': True, 'gbn': True, 'gbn_split': 2, 'l_l1': 0.00000000000001}\n",
        "]\n",
        "    "
      ],
      "execution_count": 9,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Dq3t-rSPp0tb",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "\n",
        "train_losses = {}\n",
        "train_acc = {}\n",
        "test_losses = {}\n",
        "test_acc = {}\n",
        "for rp in run_params:\n",
        "  train_losses[rp['model']] = []\n",
        "  train_acc[rp['model']] = []\n",
        "  test_losses[rp['model']] = []\n",
        "  test_acc[rp['model']] = []\n",
        "\n"
      ],
      "execution_count": 10,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "VWo5Ny6aduIC",
        "colab_type": "text"
      },
      "source": [
        "## **Training and Testing Loops**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "57yWDgg6BBHP",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from tqdm import tqdm\n",
        "\n",
        "# train_losses = []\n",
        "# train_acc = []\n",
        "# test_losses = []\n",
        "# test_acc = []\n",
        "\n",
        "misclassified = []\n",
        "\n",
        "#{'model': 'L1_with_BN', 'l1': True, 'l2': False, 'gbn': False, 'gbn_split': 1},\n",
        "def train(model, device, train_loader, optimizer, epoch, params):\n",
        "  name = params['model']\n",
        "\n",
        "  model.train()\n",
        "  pbar = tqdm(train_loader, position=0)\n",
        "  correct = 0\n",
        "  processed = 0\n",
        "  for batch_idx, (data, target) in enumerate(pbar):\n",
        "    # get samples\n",
        "    data, target = data.to(device), target.to(device)\n",
        "\n",
        "    # Init\n",
        "    optimizer.zero_grad()\n",
        "    # In PyTorch, we need to set the gradients to zero before starting to do backpropragation because PyTorch accumulates the gradients on subsequent backward passes. \n",
        "    # Because of this, when you start your training loop, ideally you should zero out the gradients so that you do the parameter update correctly.\n",
        "\n",
        "    # Predict\n",
        "    y_pred = model(data)\n",
        "\n",
        "    l1_reg = 0\n",
        "    l2_reg = 0\n",
        "\n",
        "    if params['l1']:\n",
        "      for param in model.parameters():\n",
        "        l1_reg += param.abs().sum()\n",
        "\n",
        "    # if params['l2']:\n",
        "    #   for param in model.parameters():\n",
        "    #     l2_reg += param.norm(2)\n",
        "\n",
        "    # Calculate loss\n",
        "    loss = F.nll_loss(y_pred, target) + params['l_l1'] * l1_reg #+ lambda_l2 * l2_reg\n",
        "\n",
        "    #train_losses.append(loss)\n",
        "    train_losses[name].append(loss)\n",
        "\n",
        "\n",
        "    # Backpropagation\n",
        "    loss.backward()\n",
        "    optimizer.step()\n",
        "\n",
        "    # Update pbar-tqdm\n",
        "    \n",
        "    pred = y_pred.argmax(dim=1, keepdim=True)  # get the index of the max log-probability\n",
        "    correct += pred.eq(target.view_as(pred)).sum().item()\n",
        "    processed += len(data)\n",
        "\n",
        "    pbar.set_description(desc= f'Loss={loss.item()} Batch_id={batch_idx} Accuracy={100*correct/processed:0.2f}')\n",
        "    #train_acc.append(100*correct/processed)\n",
        "    train_acc[name].append(100*correct/processed)\n",
        "  \n",
        "\n",
        "def test(model, device, test_loader, params):\n",
        "    name = params['model']\n",
        "\n",
        "    model.eval()\n",
        "    test_loss = 0\n",
        "    correct = 0\n",
        "    with torch.no_grad():\n",
        "        for data, target in test_loader:\n",
        "            data, target = data.to(device), target.to(device)\n",
        "            output = model(data)\n",
        "            test_loss += F.nll_loss(output, target, reduction='sum').item()  # sum up batch loss\n",
        "            pred = output.argmax(dim=1, keepdim=True)  # get the index of the max log-probability\n",
        "            correct += pred.eq(target.view_as(pred)).sum().item()\n",
        "\n",
        "    test_loss /= len(test_loader.dataset)\n",
        "    #test_losses.append(test_loss)\n",
        "    test_losses[name].append(test_loss)\n",
        "\n",
        "    print('\\nTest set: Average loss: {:.4f}, Accuracy: {}/{} ({:.2f}%)\\n'.format(\n",
        "        test_loss, correct, len(test_loader.dataset),\n",
        "        100. * correct / len(test_loader.dataset)))\n",
        "    \n",
        "    #test_acc.append(100. * correct / len(test_loader.dataset))\n",
        "    test_acc[name].append(100. * correct / len(test_loader.dataset))\n",
        "\n",
        "    "
      ],
      "execution_count": 11,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "k5HdrVLYT2l5",
        "colab_type": "text"
      },
      "source": [
        "## **Running the models**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cw0UflFUYSno",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "794d0470-aa76-454d-9efb-3a32056df8c3"
      },
      "source": [
        "models = {}\n",
        "\n",
        "optim_params = {\n",
        "  'L1_with_BN':        {'lr':0.025, 'step_size': 12,'gamma': 0.2, 'weight_decay': 0},\n",
        "  'L2_with_BN':        {'lr':0.025, 'step_size': 6, 'gamma': 0.1, 'weight_decay': 0.000095},\n",
        "  'L1_and_L2_with_BN': {'lr':0.025, 'step_size': 6, 'gamma': 0.1, 'weight_decay': 0.000009},\n",
        "  'GBN':               {'lr':0.025, 'step_size': 6, 'gamma': 0.1, 'weight_decay': 0},\n",
        "  'L1_and_L2_with_GBN':{'lr':0.025, 'step_size': 6, 'gamma': 0.1, 'weight_decay': 0.0000001} \n",
        "}\n",
        "\n",
        "\n",
        "for i, rp in enumerate(run_params):\n",
        "    name = rp['model']\n",
        "    opt = optim_params[name]\n",
        "    print(\"\\n\\n ******** Running for: \", name, \"====================================================\")\n",
        "    model =  Net(rp['gbn'], rp['gbn_split']).to(device)\n",
        "    \n",
        "    #optimizer = optim.SGD(model.parameters(), lr=0.0180, momentum=0.92)\n",
        "    optimizer = optim.SGD(model.parameters(), lr=opt['lr'], momentum=0.9, weight_decay=opt['weight_decay'])\n",
        "    scheduler = StepLR(optimizer, step_size=opt['step_size'], gamma=opt['gamma'])\n",
        "\n",
        "    models[name] = model\n",
        "\n",
        "    EPOCHS = 25\n",
        "    for epoch in range(EPOCHS):\n",
        "        print(\"EPOCH:\", epoch+1)\n",
        "        train(model, device, train_loader, optimizer, epoch, rp)\n",
        "        scheduler.step()\n",
        "        test(model, device, test_loader, rp)\n"
      ],
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\r  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "\n",
            " ******** Running for:  L1_with_BN ====================================================\n",
            "Is GBN run: False\n",
            "EPOCH: 1\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.047586590051651 Batch_id=468 Accuracy=93.45: 100%|██████████| 469/469 [00:15<00:00, 30.37it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0720, Accuracy: 9780/10000 (97.80%)\n",
            "\n",
            "EPOCH: 2\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.08392079919576645 Batch_id=468 Accuracy=97.90: 100%|██████████| 469/469 [00:15<00:00, 30.61it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0424, Accuracy: 9868/10000 (98.68%)\n",
            "\n",
            "EPOCH: 3\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.012756433337926865 Batch_id=468 Accuracy=98.44: 100%|██████████| 469/469 [00:15<00:00, 30.74it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0362, Accuracy: 9882/10000 (98.82%)\n",
            "\n",
            "EPOCH: 4\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.09352466464042664 Batch_id=468 Accuracy=98.58: 100%|██████████| 469/469 [00:15<00:00, 31.25it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0315, Accuracy: 9897/10000 (98.97%)\n",
            "\n",
            "EPOCH: 5\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.022475695237517357 Batch_id=468 Accuracy=98.67: 100%|██████████| 469/469 [00:14<00:00, 31.27it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0286, Accuracy: 9905/10000 (99.05%)\n",
            "\n",
            "EPOCH: 6\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.017834655940532684 Batch_id=468 Accuracy=98.84: 100%|██████████| 469/469 [00:14<00:00, 31.56it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0297, Accuracy: 9901/10000 (99.01%)\n",
            "\n",
            "EPOCH: 7\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.1211862564086914 Batch_id=468 Accuracy=98.91: 100%|██████████| 469/469 [00:14<00:00, 31.34it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0319, Accuracy: 9902/10000 (99.02%)\n",
            "\n",
            "EPOCH: 8\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.15520676970481873 Batch_id=468 Accuracy=98.95: 100%|██████████| 469/469 [00:14<00:00, 31.34it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0293, Accuracy: 9903/10000 (99.03%)\n",
            "\n",
            "EPOCH: 9\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.032127901911735535 Batch_id=468 Accuracy=99.00: 100%|██████████| 469/469 [00:14<00:00, 32.05it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0262, Accuracy: 9917/10000 (99.17%)\n",
            "\n",
            "EPOCH: 10\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.033229488879442215 Batch_id=468 Accuracy=99.03: 100%|██████████| 469/469 [00:14<00:00, 32.34it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0250, Accuracy: 9916/10000 (99.16%)\n",
            "\n",
            "EPOCH: 11\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.06850796937942505 Batch_id=468 Accuracy=99.09: 100%|██████████| 469/469 [00:14<00:00, 32.20it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0259, Accuracy: 9912/10000 (99.12%)\n",
            "\n",
            "EPOCH: 12\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.05499899014830589 Batch_id=468 Accuracy=99.14: 100%|██████████| 469/469 [00:14<00:00, 32.51it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0287, Accuracy: 9904/10000 (99.04%)\n",
            "\n",
            "EPOCH: 13\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.012434953823685646 Batch_id=468 Accuracy=99.42: 100%|██████████| 469/469 [00:14<00:00, 32.35it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0188, Accuracy: 9937/10000 (99.37%)\n",
            "\n",
            "EPOCH: 14\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.019339516758918762 Batch_id=468 Accuracy=99.49: 100%|██████████| 469/469 [00:14<00:00, 32.61it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0203, Accuracy: 9926/10000 (99.26%)\n",
            "\n",
            "EPOCH: 15\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.020838607102632523 Batch_id=468 Accuracy=99.50: 100%|██████████| 469/469 [00:14<00:00, 32.93it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0194, Accuracy: 9933/10000 (99.33%)\n",
            "\n",
            "EPOCH: 16\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.01096355076879263 Batch_id=468 Accuracy=99.52: 100%|██████████| 469/469 [00:14<00:00, 33.05it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0192, Accuracy: 9930/10000 (99.30%)\n",
            "\n",
            "EPOCH: 17\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.04817938059568405 Batch_id=468 Accuracy=99.54: 100%|██████████| 469/469 [00:14<00:00, 32.41it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0194, Accuracy: 9934/10000 (99.34%)\n",
            "\n",
            "EPOCH: 18\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.01735781505703926 Batch_id=468 Accuracy=99.53: 100%|██████████| 469/469 [00:14<00:00, 32.73it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0196, Accuracy: 9924/10000 (99.24%)\n",
            "\n",
            "EPOCH: 19\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.009636909700930119 Batch_id=468 Accuracy=99.54: 100%|██████████| 469/469 [00:14<00:00, 32.92it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0204, Accuracy: 9928/10000 (99.28%)\n",
            "\n",
            "EPOCH: 20\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.01106911338865757 Batch_id=468 Accuracy=99.55: 100%|██████████| 469/469 [00:14<00:00, 32.81it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0202, Accuracy: 9928/10000 (99.28%)\n",
            "\n",
            "EPOCH: 21\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.011740226298570633 Batch_id=468 Accuracy=99.55: 100%|██████████| 469/469 [00:14<00:00, 32.69it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0201, Accuracy: 9929/10000 (99.29%)\n",
            "\n",
            "EPOCH: 22\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.03936079144477844 Batch_id=468 Accuracy=99.53: 100%|██████████| 469/469 [00:13<00:00, 33.51it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0202, Accuracy: 9932/10000 (99.32%)\n",
            "\n",
            "EPOCH: 23\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.0230497308075428 Batch_id=468 Accuracy=99.59: 100%|██████████| 469/469 [00:14<00:00, 33.22it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0191, Accuracy: 9932/10000 (99.32%)\n",
            "\n",
            "EPOCH: 24\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.010507270693778992 Batch_id=468 Accuracy=99.61: 100%|██████████| 469/469 [00:14<00:00, 32.41it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0226, Accuracy: 9924/10000 (99.24%)\n",
            "\n",
            "EPOCH: 25\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.015050999820232391 Batch_id=468 Accuracy=99.65: 100%|██████████| 469/469 [00:14<00:00, 31.35it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0196, Accuracy: 9929/10000 (99.29%)\n",
            "\n",
            "\n",
            "\n",
            " ******** Running for:  L2_with_BN ====================================================\n",
            "Is GBN run: False\n",
            "EPOCH: 1\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.04160846024751663 Batch_id=468 Accuracy=93.60: 100%|██████████| 469/469 [00:14<00:00, 32.79it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0514, Accuracy: 9833/10000 (98.33%)\n",
            "\n",
            "EPOCH: 2\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.028750034049153328 Batch_id=468 Accuracy=98.08: 100%|██████████| 469/469 [00:14<00:00, 32.14it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0458, Accuracy: 9857/10000 (98.57%)\n",
            "\n",
            "EPOCH: 3\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.1284441500902176 Batch_id=468 Accuracy=98.33: 100%|██████████| 469/469 [00:14<00:00, 31.70it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0460, Accuracy: 9844/10000 (98.44%)\n",
            "\n",
            "EPOCH: 4\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.025593185797333717 Batch_id=468 Accuracy=98.57: 100%|██████████| 469/469 [00:14<00:00, 33.01it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0440, Accuracy: 9872/10000 (98.72%)\n",
            "\n",
            "EPOCH: 5\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.007269354537129402 Batch_id=468 Accuracy=98.71: 100%|██████████| 469/469 [00:14<00:00, 32.93it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0320, Accuracy: 9908/10000 (99.08%)\n",
            "\n",
            "EPOCH: 6\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.05260096490383148 Batch_id=468 Accuracy=98.86: 100%|██████████| 469/469 [00:14<00:00, 33.39it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0303, Accuracy: 9914/10000 (99.14%)\n",
            "\n",
            "EPOCH: 7\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.0015585358487442136 Batch_id=468 Accuracy=99.22: 100%|██████████| 469/469 [00:14<00:00, 33.27it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0206, Accuracy: 9940/10000 (99.40%)\n",
            "\n",
            "EPOCH: 8\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.013495209626853466 Batch_id=468 Accuracy=99.28: 100%|██████████| 469/469 [00:14<00:00, 32.93it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0209, Accuracy: 9938/10000 (99.38%)\n",
            "\n",
            "EPOCH: 9\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.001932071871124208 Batch_id=468 Accuracy=99.31: 100%|██████████| 469/469 [00:14<00:00, 33.25it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0208, Accuracy: 9937/10000 (99.37%)\n",
            "\n",
            "EPOCH: 10\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.015611216425895691 Batch_id=468 Accuracy=99.30: 100%|██████████| 469/469 [00:13<00:00, 33.57it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0206, Accuracy: 9941/10000 (99.41%)\n",
            "\n",
            "EPOCH: 11\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.06115948036313057 Batch_id=468 Accuracy=99.34: 100%|██████████| 469/469 [00:13<00:00, 33.59it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0207, Accuracy: 9941/10000 (99.41%)\n",
            "\n",
            "EPOCH: 12\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.007530607748776674 Batch_id=468 Accuracy=99.35: 100%|██████████| 469/469 [00:13<00:00, 33.60it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0200, Accuracy: 9943/10000 (99.43%)\n",
            "\n",
            "EPOCH: 13\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.006074739620089531 Batch_id=468 Accuracy=99.41: 100%|██████████| 469/469 [00:14<00:00, 33.30it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0195, Accuracy: 9947/10000 (99.47%)\n",
            "\n",
            "EPOCH: 14\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.020930113270878792 Batch_id=468 Accuracy=99.36: 100%|██████████| 469/469 [00:13<00:00, 33.97it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0199, Accuracy: 9944/10000 (99.44%)\n",
            "\n",
            "EPOCH: 15\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.013531186617910862 Batch_id=468 Accuracy=99.38: 100%|██████████| 469/469 [00:14<00:00, 33.24it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0196, Accuracy: 9944/10000 (99.44%)\n",
            "\n",
            "EPOCH: 16\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.011616977863013744 Batch_id=468 Accuracy=99.41: 100%|██████████| 469/469 [00:13<00:00, 34.12it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0196, Accuracy: 9943/10000 (99.43%)\n",
            "\n",
            "EPOCH: 17\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.0018994769779965281 Batch_id=468 Accuracy=99.37: 100%|██████████| 469/469 [00:13<00:00, 34.12it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0196, Accuracy: 9943/10000 (99.43%)\n",
            "\n",
            "EPOCH: 18\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.03515801206231117 Batch_id=468 Accuracy=99.38: 100%|██████████| 469/469 [00:13<00:00, 33.92it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0199, Accuracy: 9945/10000 (99.45%)\n",
            "\n",
            "EPOCH: 19\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.08505924791097641 Batch_id=468 Accuracy=99.40: 100%|██████████| 469/469 [00:13<00:00, 34.42it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0196, Accuracy: 9944/10000 (99.44%)\n",
            "\n",
            "EPOCH: 20\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.00266945268958807 Batch_id=468 Accuracy=99.42: 100%|██████████| 469/469 [00:13<00:00, 34.54it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0193, Accuracy: 9945/10000 (99.45%)\n",
            "\n",
            "EPOCH: 21\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.003296604612842202 Batch_id=468 Accuracy=99.42: 100%|██████████| 469/469 [00:13<00:00, 34.90it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0194, Accuracy: 9943/10000 (99.43%)\n",
            "\n",
            "EPOCH: 22\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.009916013106703758 Batch_id=468 Accuracy=99.43: 100%|██████████| 469/469 [00:13<00:00, 35.15it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0192, Accuracy: 9944/10000 (99.44%)\n",
            "\n",
            "EPOCH: 23\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.0010738997953012586 Batch_id=468 Accuracy=99.42: 100%|██████████| 469/469 [00:13<00:00, 34.43it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0195, Accuracy: 9943/10000 (99.43%)\n",
            "\n",
            "EPOCH: 24\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.005276810377836227 Batch_id=468 Accuracy=99.38: 100%|██████████| 469/469 [00:13<00:00, 34.80it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0191, Accuracy: 9946/10000 (99.46%)\n",
            "\n",
            "EPOCH: 25\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.07128507643938065 Batch_id=468 Accuracy=99.38: 100%|██████████| 469/469 [00:13<00:00, 35.06it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0198, Accuracy: 9944/10000 (99.44%)\n",
            "\n",
            "\n",
            "\n",
            " ******** Running for:  L1_and_L2_with_BN ====================================================\n",
            "Is GBN run: False\n",
            "EPOCH: 1\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.1659460812807083 Batch_id=468 Accuracy=93.06: 100%|██████████| 469/469 [00:14<00:00, 32.81it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0776, Accuracy: 9766/10000 (97.66%)\n",
            "\n",
            "EPOCH: 2\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.033706530928611755 Batch_id=468 Accuracy=97.94: 100%|██████████| 469/469 [00:14<00:00, 32.70it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.1177, Accuracy: 9638/10000 (96.38%)\n",
            "\n",
            "EPOCH: 3\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.0776861310005188 Batch_id=468 Accuracy=98.35: 100%|██████████| 469/469 [00:14<00:00, 32.73it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0457, Accuracy: 9850/10000 (98.50%)\n",
            "\n",
            "EPOCH: 4\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.009049490094184875 Batch_id=468 Accuracy=98.57: 100%|██████████| 469/469 [00:14<00:00, 32.85it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0464, Accuracy: 9851/10000 (98.51%)\n",
            "\n",
            "EPOCH: 5\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.04798010736703873 Batch_id=468 Accuracy=98.67: 100%|██████████| 469/469 [00:14<00:00, 33.08it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0334, Accuracy: 9900/10000 (99.00%)\n",
            "\n",
            "EPOCH: 6\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.016591744497418404 Batch_id=468 Accuracy=98.79: 100%|██████████| 469/469 [00:14<00:00, 32.62it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0305, Accuracy: 9902/10000 (99.02%)\n",
            "\n",
            "EPOCH: 7\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.01582890748977661 Batch_id=468 Accuracy=99.20: 100%|██████████| 469/469 [00:14<00:00, 32.96it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0203, Accuracy: 9937/10000 (99.37%)\n",
            "\n",
            "EPOCH: 8\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.02499879151582718 Batch_id=468 Accuracy=99.25: 100%|██████████| 469/469 [00:14<00:00, 32.43it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0211, Accuracy: 9935/10000 (99.35%)\n",
            "\n",
            "EPOCH: 9\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.047403376549482346 Batch_id=468 Accuracy=99.28: 100%|██████████| 469/469 [00:14<00:00, 32.39it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0203, Accuracy: 9935/10000 (99.35%)\n",
            "\n",
            "EPOCH: 10\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.03722913935780525 Batch_id=468 Accuracy=99.31: 100%|██████████| 469/469 [00:14<00:00, 32.40it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0209, Accuracy: 9928/10000 (99.28%)\n",
            "\n",
            "EPOCH: 11\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.026354990899562836 Batch_id=468 Accuracy=99.28: 100%|██████████| 469/469 [00:14<00:00, 32.97it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0195, Accuracy: 9947/10000 (99.47%)\n",
            "\n",
            "EPOCH: 12\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.00535610830411315 Batch_id=468 Accuracy=99.31: 100%|██████████| 469/469 [00:14<00:00, 33.12it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0197, Accuracy: 9939/10000 (99.39%)\n",
            "\n",
            "EPOCH: 13\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.0050308676436543465 Batch_id=468 Accuracy=99.39: 100%|██████████| 469/469 [00:14<00:00, 32.37it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0194, Accuracy: 9940/10000 (99.40%)\n",
            "\n",
            "EPOCH: 14\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.015661122277379036 Batch_id=468 Accuracy=99.40: 100%|██████████| 469/469 [00:14<00:00, 32.73it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0194, Accuracy: 9943/10000 (99.43%)\n",
            "\n",
            "EPOCH: 15\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.0398571640253067 Batch_id=468 Accuracy=99.37: 100%|██████████| 469/469 [00:14<00:00, 33.11it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0196, Accuracy: 9937/10000 (99.37%)\n",
            "\n",
            "EPOCH: 16\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.07965303957462311 Batch_id=468 Accuracy=99.36: 100%|██████████| 469/469 [00:14<00:00, 32.59it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0193, Accuracy: 9939/10000 (99.39%)\n",
            "\n",
            "EPOCH: 17\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.004904213361442089 Batch_id=468 Accuracy=99.36: 100%|██████████| 469/469 [00:14<00:00, 32.86it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0193, Accuracy: 9939/10000 (99.39%)\n",
            "\n",
            "EPOCH: 18\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.0027339356020092964 Batch_id=468 Accuracy=99.38: 100%|██████████| 469/469 [00:14<00:00, 32.73it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0198, Accuracy: 9935/10000 (99.35%)\n",
            "\n",
            "EPOCH: 19\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.018705397844314575 Batch_id=468 Accuracy=99.37: 100%|██████████| 469/469 [00:14<00:00, 32.92it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0197, Accuracy: 9936/10000 (99.36%)\n",
            "\n",
            "EPOCH: 20\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.01053981576114893 Batch_id=468 Accuracy=99.42: 100%|██████████| 469/469 [00:14<00:00, 33.01it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0197, Accuracy: 9936/10000 (99.36%)\n",
            "\n",
            "EPOCH: 21\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.005441613495349884 Batch_id=468 Accuracy=99.39: 100%|██████████| 469/469 [00:14<00:00, 33.25it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0198, Accuracy: 9938/10000 (99.38%)\n",
            "\n",
            "EPOCH: 22\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.027603236958384514 Batch_id=468 Accuracy=99.36: 100%|██████████| 469/469 [00:14<00:00, 33.39it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0195, Accuracy: 9936/10000 (99.36%)\n",
            "\n",
            "EPOCH: 23\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.020395582541823387 Batch_id=468 Accuracy=99.36: 100%|██████████| 469/469 [00:14<00:00, 32.59it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0193, Accuracy: 9941/10000 (99.41%)\n",
            "\n",
            "EPOCH: 24\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.001686316798441112 Batch_id=468 Accuracy=99.37: 100%|██████████| 469/469 [00:14<00:00, 32.84it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0195, Accuracy: 9937/10000 (99.37%)\n",
            "\n",
            "EPOCH: 25\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.010109425522387028 Batch_id=468 Accuracy=99.40: 100%|██████████| 469/469 [00:14<00:00, 33.25it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0193, Accuracy: 9938/10000 (99.38%)\n",
            "\n",
            "\n",
            "\n",
            " ******** Running for:  GBN ====================================================\n",
            "Is GBN run: True\n",
            "EPOCH: 1\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.04080190137028694 Batch_id=468 Accuracy=93.23: 100%|██████████| 469/469 [00:13<00:00, 34.69it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0636, Accuracy: 9801/10000 (98.01%)\n",
            "\n",
            "EPOCH: 2\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.02769782952964306 Batch_id=468 Accuracy=97.86: 100%|██████████| 469/469 [00:13<00:00, 35.26it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0393, Accuracy: 9869/10000 (98.69%)\n",
            "\n",
            "EPOCH: 3\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.08926365524530411 Batch_id=468 Accuracy=98.30: 100%|██████████| 469/469 [00:13<00:00, 35.29it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0379, Accuracy: 9887/10000 (98.87%)\n",
            "\n",
            "EPOCH: 4\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.015244563110172749 Batch_id=468 Accuracy=98.45: 100%|██████████| 469/469 [00:13<00:00, 34.98it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0362, Accuracy: 9871/10000 (98.71%)\n",
            "\n",
            "EPOCH: 5\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.054897189140319824 Batch_id=468 Accuracy=98.74: 100%|██████████| 469/469 [00:13<00:00, 34.71it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0269, Accuracy: 9918/10000 (99.18%)\n",
            "\n",
            "EPOCH: 6\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.036353908479213715 Batch_id=468 Accuracy=98.86: 100%|██████████| 469/469 [00:13<00:00, 35.49it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0280, Accuracy: 9909/10000 (99.09%)\n",
            "\n",
            "EPOCH: 7\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.012620653957128525 Batch_id=468 Accuracy=99.15: 100%|██████████| 469/469 [00:13<00:00, 35.04it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0202, Accuracy: 9930/10000 (99.30%)\n",
            "\n",
            "EPOCH: 8\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.0784020945429802 Batch_id=468 Accuracy=99.24: 100%|██████████| 469/469 [00:13<00:00, 34.91it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0203, Accuracy: 9931/10000 (99.31%)\n",
            "\n",
            "EPOCH: 9\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.016356175765395164 Batch_id=468 Accuracy=99.27: 100%|██████████| 469/469 [00:13<00:00, 34.50it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0193, Accuracy: 9929/10000 (99.29%)\n",
            "\n",
            "EPOCH: 10\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.006443571299314499 Batch_id=468 Accuracy=99.32: 100%|██████████| 469/469 [00:13<00:00, 35.17it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0196, Accuracy: 9931/10000 (99.31%)\n",
            "\n",
            "EPOCH: 11\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.005486130248755217 Batch_id=468 Accuracy=99.28: 100%|██████████| 469/469 [00:13<00:00, 34.97it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0200, Accuracy: 9930/10000 (99.30%)\n",
            "\n",
            "EPOCH: 12\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.026189029216766357 Batch_id=468 Accuracy=99.31: 100%|██████████| 469/469 [00:13<00:00, 35.53it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0196, Accuracy: 9932/10000 (99.32%)\n",
            "\n",
            "EPOCH: 13\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.008274701423943043 Batch_id=468 Accuracy=99.33: 100%|██████████| 469/469 [00:13<00:00, 35.45it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0196, Accuracy: 9933/10000 (99.33%)\n",
            "\n",
            "EPOCH: 14\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.08821997046470642 Batch_id=468 Accuracy=99.28: 100%|██████████| 469/469 [00:13<00:00, 35.21it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0193, Accuracy: 9936/10000 (99.36%)\n",
            "\n",
            "EPOCH: 15\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.013462062925100327 Batch_id=468 Accuracy=99.33: 100%|██████████| 469/469 [00:13<00:00, 34.99it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0197, Accuracy: 9933/10000 (99.33%)\n",
            "\n",
            "EPOCH: 16\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.03298424929380417 Batch_id=468 Accuracy=99.38: 100%|██████████| 469/469 [00:13<00:00, 35.39it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0192, Accuracy: 9934/10000 (99.34%)\n",
            "\n",
            "EPOCH: 17\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.009989330545067787 Batch_id=468 Accuracy=99.33: 100%|██████████| 469/469 [00:13<00:00, 35.16it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0193, Accuracy: 9932/10000 (99.32%)\n",
            "\n",
            "EPOCH: 18\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.004246934317052364 Batch_id=468 Accuracy=99.35: 100%|██████████| 469/469 [00:13<00:00, 35.28it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0189, Accuracy: 9934/10000 (99.34%)\n",
            "\n",
            "EPOCH: 19\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.08720750361680984 Batch_id=468 Accuracy=99.37: 100%|██████████| 469/469 [00:13<00:00, 34.96it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0192, Accuracy: 9934/10000 (99.34%)\n",
            "\n",
            "EPOCH: 20\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.010664432309567928 Batch_id=468 Accuracy=99.37: 100%|██████████| 469/469 [00:13<00:00, 34.83it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0192, Accuracy: 9933/10000 (99.33%)\n",
            "\n",
            "EPOCH: 21\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.009560640901327133 Batch_id=468 Accuracy=99.37: 100%|██████████| 469/469 [00:13<00:00, 35.07it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0191, Accuracy: 9933/10000 (99.33%)\n",
            "\n",
            "EPOCH: 22\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.008923898451030254 Batch_id=468 Accuracy=99.32: 100%|██████████| 469/469 [00:13<00:00, 35.57it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0192, Accuracy: 9933/10000 (99.33%)\n",
            "\n",
            "EPOCH: 23\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.01262510847300291 Batch_id=468 Accuracy=99.35: 100%|██████████| 469/469 [00:12<00:00, 36.55it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0190, Accuracy: 9932/10000 (99.32%)\n",
            "\n",
            "EPOCH: 24\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.06746739894151688 Batch_id=468 Accuracy=99.36: 100%|██████████| 469/469 [00:12<00:00, 36.16it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0191, Accuracy: 9935/10000 (99.35%)\n",
            "\n",
            "EPOCH: 25\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.011365833692252636 Batch_id=468 Accuracy=99.36: 100%|██████████| 469/469 [00:12<00:00, 36.09it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0192, Accuracy: 9931/10000 (99.31%)\n",
            "\n",
            "\n",
            "\n",
            " ******** Running for:  L1_and_L2_with_GBN ====================================================\n",
            "Is GBN run: True\n",
            "EPOCH: 1\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.031726133078336716 Batch_id=468 Accuracy=93.13: 100%|██████████| 469/469 [00:14<00:00, 32.71it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0669, Accuracy: 9799/10000 (97.99%)\n",
            "\n",
            "EPOCH: 2\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.07376376539468765 Batch_id=468 Accuracy=97.84: 100%|██████████| 469/469 [00:14<00:00, 33.00it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0452, Accuracy: 9857/10000 (98.57%)\n",
            "\n",
            "EPOCH: 3\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.03546133637428284 Batch_id=468 Accuracy=98.30: 100%|██████████| 469/469 [00:14<00:00, 32.83it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0384, Accuracy: 9877/10000 (98.77%)\n",
            "\n",
            "EPOCH: 4\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.05998022481799126 Batch_id=468 Accuracy=98.52: 100%|██████████| 469/469 [00:14<00:00, 32.79it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0320, Accuracy: 9890/10000 (98.90%)\n",
            "\n",
            "EPOCH: 5\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.058358240872621536 Batch_id=468 Accuracy=98.69: 100%|██████████| 469/469 [00:14<00:00, 32.83it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0287, Accuracy: 9915/10000 (99.15%)\n",
            "\n",
            "EPOCH: 6\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.06924328953027725 Batch_id=468 Accuracy=98.82: 100%|██████████| 469/469 [00:14<00:00, 31.69it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0286, Accuracy: 9901/10000 (99.01%)\n",
            "\n",
            "EPOCH: 7\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.0505179725587368 Batch_id=468 Accuracy=99.09: 100%|██████████| 469/469 [00:15<00:00, 30.61it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0212, Accuracy: 9930/10000 (99.30%)\n",
            "\n",
            "EPOCH: 8\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.08004868030548096 Batch_id=468 Accuracy=99.23: 100%|██████████| 469/469 [00:15<00:00, 30.05it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0213, Accuracy: 9927/10000 (99.27%)\n",
            "\n",
            "EPOCH: 9\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.12025503069162369 Batch_id=468 Accuracy=99.25: 100%|██████████| 469/469 [00:16<00:00, 28.74it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0206, Accuracy: 9929/10000 (99.29%)\n",
            "\n",
            "EPOCH: 10\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.0027310699224472046 Batch_id=468 Accuracy=99.24: 100%|██████████| 469/469 [00:16<00:00, 28.95it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0209, Accuracy: 9932/10000 (99.32%)\n",
            "\n",
            "EPOCH: 11\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.004232124425470829 Batch_id=468 Accuracy=99.28: 100%|██████████| 469/469 [00:15<00:00, 30.40it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0204, Accuracy: 9928/10000 (99.28%)\n",
            "\n",
            "EPOCH: 12\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.01447733398526907 Batch_id=468 Accuracy=99.26: 100%|██████████| 469/469 [00:15<00:00, 30.71it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0202, Accuracy: 9931/10000 (99.31%)\n",
            "\n",
            "EPOCH: 13\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.051063522696495056 Batch_id=468 Accuracy=99.34: 100%|██████████| 469/469 [00:15<00:00, 30.92it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0204, Accuracy: 9930/10000 (99.30%)\n",
            "\n",
            "EPOCH: 14\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.027480928227305412 Batch_id=468 Accuracy=99.32: 100%|██████████| 469/469 [00:15<00:00, 30.89it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0200, Accuracy: 9932/10000 (99.32%)\n",
            "\n",
            "EPOCH: 15\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.003992496989667416 Batch_id=468 Accuracy=99.39: 100%|██████████| 469/469 [00:15<00:00, 30.87it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0199, Accuracy: 9933/10000 (99.33%)\n",
            "\n",
            "EPOCH: 16\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.013884241692721844 Batch_id=468 Accuracy=99.38: 100%|██████████| 469/469 [00:15<00:00, 31.26it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0197, Accuracy: 9934/10000 (99.34%)\n",
            "\n",
            "EPOCH: 17\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.017451053485274315 Batch_id=468 Accuracy=99.35: 100%|██████████| 469/469 [00:14<00:00, 31.80it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0198, Accuracy: 9933/10000 (99.33%)\n",
            "\n",
            "EPOCH: 18\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.025466205552220345 Batch_id=468 Accuracy=99.34: 100%|██████████| 469/469 [00:14<00:00, 31.64it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0197, Accuracy: 9932/10000 (99.32%)\n",
            "\n",
            "EPOCH: 19\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.10791542381048203 Batch_id=468 Accuracy=99.37: 100%|██████████| 469/469 [00:15<00:00, 31.09it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0199, Accuracy: 9933/10000 (99.33%)\n",
            "\n",
            "EPOCH: 20\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.06060445308685303 Batch_id=468 Accuracy=99.33: 100%|██████████| 469/469 [00:15<00:00, 31.15it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0197, Accuracy: 9931/10000 (99.31%)\n",
            "\n",
            "EPOCH: 21\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.0073476326651871204 Batch_id=468 Accuracy=99.33: 100%|██████████| 469/469 [00:14<00:00, 31.72it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0202, Accuracy: 9931/10000 (99.31%)\n",
            "\n",
            "EPOCH: 22\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.03385685011744499 Batch_id=468 Accuracy=99.36: 100%|██████████| 469/469 [00:14<00:00, 31.37it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0198, Accuracy: 9934/10000 (99.34%)\n",
            "\n",
            "EPOCH: 23\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.02324589155614376 Batch_id=468 Accuracy=99.31: 100%|██████████| 469/469 [00:14<00:00, 32.02it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0197, Accuracy: 9935/10000 (99.35%)\n",
            "\n",
            "EPOCH: 24\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.037820447236299515 Batch_id=468 Accuracy=99.33: 100%|██████████| 469/469 [00:14<00:00, 31.39it/s]\n",
            "  0%|          | 0/469 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0198, Accuracy: 9934/10000 (99.34%)\n",
            "\n",
            "EPOCH: 25\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Loss=0.059954870492219925 Batch_id=468 Accuracy=99.35: 100%|██████████| 469/469 [00:14<00:00, 32.12it/s]\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Test set: Average loss: 0.0197, Accuracy: 9931/10000 (99.31%)\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "73zQg2kRHn33",
        "colab_type": "text"
      },
      "source": [
        "## **Plotting functions**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-4u9w-HV6swj",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "%matplotlib inline\n",
        "import matplotlib.pyplot as plt\n",
        "from google.colab import files\n",
        "\n",
        "def plot_with_legends(run_params, data, title):\n",
        "    figure = plt.figure()\n",
        "    #figure.suptitle(title, fontsize=16)\n",
        "    plt.figure(figsize=(15,8))\n",
        "    \n",
        "    for i, pld in enumerate(run_params):\n",
        "        name = pld['model']\n",
        "        plt.plot(data[name], label=name)\n",
        "\n",
        "    plt.legend()\n",
        "    plt.title(title)\n",
        "    fname = name + \"_\" + title  \n",
        "    plt.savefig(fname + \".png\")\n",
        "    files.download(fname + \".png\") \n",
        "    plt.show()  \n"
      ],
      "execution_count": 13,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "AMcA9PTAck82",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 516
        },
        "outputId": "34dacde6-cf8c-431f-8971-31b67ca90ccc"
      },
      "source": [
        "plot_with_legends(run_params, train_acc, 'Training Accuracy')\n"
      ],
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "\n",
              "    async function download(id, filename, size) {\n",
              "      if (!google.colab.kernel.accessAllowed) {\n",
              "        return;\n",
              "      }\n",
              "      const div = document.createElement('div');\n",
              "      const label = document.createElement('label');\n",
              "      label.textContent = `Downloading \"${filename}\": `;\n",
              "      div.appendChild(label);\n",
              "      const progress = document.createElement('progress');\n",
              "      progress.max = size;\n",
              "      div.appendChild(progress);\n",
              "      document.body.appendChild(div);\n",
              "\n",
              "      const buffers = [];\n",
              "      let downloaded = 0;\n",
              "\n",
              "      const channel = await google.colab.kernel.comms.open(id);\n",
              "      // Send a message to notify the kernel that we're ready.\n",
              "      channel.send({})\n",
              "\n",
              "      for await (const message of channel.messages) {\n",
              "        // Send a message to notify the kernel that we're ready.\n",
              "        channel.send({})\n",
              "        if (message.buffers) {\n",
              "          for (const buffer of message.buffers) {\n",
              "            buffers.push(buffer);\n",
              "            downloaded += buffer.byteLength;\n",
              "            progress.value = downloaded;\n",
              "          }\n",
              "        }\n",
              "      }\n",
              "      const blob = new Blob(buffers, {type: 'application/binary'});\n",
              "      const a = document.createElement('a');\n",
              "      a.href = window.URL.createObjectURL(blob);\n",
              "      a.download = filename;\n",
              "      div.appendChild(a);\n",
              "      a.click();\n",
              "      div.remove();\n",
              "    }\n",
              "  "
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "download(\"download_1c0ae982-51dc-4a8d-8a76-d197129648b1\", \"L1_and_L2_with_GBN_Training Accuracy.png\", 28182)"
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 0 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1080x576 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "n0rnAu7WcoT7",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 516
        },
        "outputId": "24ad0dbb-3504-4df4-cf56-b1306cf06298"
      },
      "source": [
        "\n",
        "plot_with_legends(run_params, test_acc, 'Test Accuracy')\n"
      ],
      "execution_count": 15,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "\n",
              "    async function download(id, filename, size) {\n",
              "      if (!google.colab.kernel.accessAllowed) {\n",
              "        return;\n",
              "      }\n",
              "      const div = document.createElement('div');\n",
              "      const label = document.createElement('label');\n",
              "      label.textContent = `Downloading \"${filename}\": `;\n",
              "      div.appendChild(label);\n",
              "      const progress = document.createElement('progress');\n",
              "      progress.max = size;\n",
              "      div.appendChild(progress);\n",
              "      document.body.appendChild(div);\n",
              "\n",
              "      const buffers = [];\n",
              "      let downloaded = 0;\n",
              "\n",
              "      const channel = await google.colab.kernel.comms.open(id);\n",
              "      // Send a message to notify the kernel that we're ready.\n",
              "      channel.send({})\n",
              "\n",
              "      for await (const message of channel.messages) {\n",
              "        // Send a message to notify the kernel that we're ready.\n",
              "        channel.send({})\n",
              "        if (message.buffers) {\n",
              "          for (const buffer of message.buffers) {\n",
              "            buffers.push(buffer);\n",
              "            downloaded += buffer.byteLength;\n",
              "            progress.value = downloaded;\n",
              "          }\n",
              "        }\n",
              "      }\n",
              "      const blob = new Blob(buffers, {type: 'application/binary'});\n",
              "      const a = document.createElement('a');\n",
              "      a.href = window.URL.createObjectURL(blob);\n",
              "      a.download = filename;\n",
              "      div.appendChild(a);\n",
              "      a.click();\n",
              "      div.remove();\n",
              "    }\n",
              "  "
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "download(\"download_3fcf29c1-4373-4c59-bfb8-45e86a7a9f61\", \"L1_and_L2_with_GBN_Test Accuracy.png\", 46990)"
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 0 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1080x576 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Mo_8gN0Ucqws",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 516
        },
        "outputId": "47fd0708-85fa-4b3e-cf65-c20dc31c19c9"
      },
      "source": [
        "plot_with_legends(run_params, test_losses, 'Test Loss')"
      ],
      "execution_count": 16,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "\n",
              "    async function download(id, filename, size) {\n",
              "      if (!google.colab.kernel.accessAllowed) {\n",
              "        return;\n",
              "      }\n",
              "      const div = document.createElement('div');\n",
              "      const label = document.createElement('label');\n",
              "      label.textContent = `Downloading \"${filename}\": `;\n",
              "      div.appendChild(label);\n",
              "      const progress = document.createElement('progress');\n",
              "      progress.max = size;\n",
              "      div.appendChild(progress);\n",
              "      document.body.appendChild(div);\n",
              "\n",
              "      const buffers = [];\n",
              "      let downloaded = 0;\n",
              "\n",
              "      const channel = await google.colab.kernel.comms.open(id);\n",
              "      // Send a message to notify the kernel that we're ready.\n",
              "      channel.send({})\n",
              "\n",
              "      for await (const message of channel.messages) {\n",
              "        // Send a message to notify the kernel that we're ready.\n",
              "        channel.send({})\n",
              "        if (message.buffers) {\n",
              "          for (const buffer of message.buffers) {\n",
              "            buffers.push(buffer);\n",
              "            downloaded += buffer.byteLength;\n",
              "            progress.value = downloaded;\n",
              "          }\n",
              "        }\n",
              "      }\n",
              "      const blob = new Blob(buffers, {type: 'application/binary'});\n",
              "      const a = document.createElement('a');\n",
              "      a.href = window.URL.createObjectURL(blob);\n",
              "      a.download = filename;\n",
              "      div.appendChild(a);\n",
              "      a.click();\n",
              "      div.remove();\n",
              "    }\n",
              "  "
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "download(\"download_6e4d20b6-d144-42e1-a484-858d1d62f0b8\", \"L1_and_L2_with_GBN_Test Loss.png\", 40812)"
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 0 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1080x576 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lUZzYV5aGHMI",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# %matplotlib inline\n",
        "# import matplotlib.pyplot as plt\n",
        "\n",
        "# def plot_combined_graphs(run_params, data, title):\n",
        "#     #figure = plt.figure()\n",
        "#     fig, axs = plt.subplots(1, 5,figsize=(25,5))\n",
        "#     fig.suptitle(title, fontsize=16)\n",
        "#     for i, pld in enumerate(run_params):\n",
        "#         name = pld['model']\n",
        "#         axs[i].plot(data[name])\n",
        "#         axs[i].set_title(name)\n",
        "\n",
        "# #plot_combined_graphs(run_params, train_losses, 'Train Losses')\n",
        "# plot_combined_graphs(run_params, train_acc, 'Train Acc')\n",
        "# #plot_combined_graphs(run_params, test_losses, 'Test Losses')\n",
        "# plot_combined_graphs(run_params, test_acc, 'Test Acc')"
      ],
      "execution_count": 17,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "vrcLOHlevPp2",
        "colab_type": "text"
      },
      "source": [
        "## **Misclassified Images with GBN**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "b6FpdO6EvFVJ",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        },
        "outputId": "ab2beabd-d9a1-423a-bb0b-8faf45621e6c"
      },
      "source": [
        "gbn_model = models['GBN']\n",
        "\n",
        "\n",
        "incorrect_indexes = {} # {23: {'actual': 1, 'predicted': 4}}\n",
        "gbn_model.eval()\n",
        "count = 0\n",
        "with torch.no_grad():\n",
        "    for data, target in test_loader:\n",
        "        data, target = data.to(device), target.to(device)\n",
        "        output = gbn_model(data)\n",
        "        pred = output.argmax(dim=1, keepdim=True)  # get the index of the max log-probability\n",
        "\n",
        "        for sampleno in range(data.shape[0]):\n",
        "                if(target[sampleno]!=pred[sampleno]):\n",
        "                    count += 1\n",
        "                    #print(\"Index=\", sampleno, \", Actual=\", target[sampleno].cpu().numpy(), \", Predicted: \", pred[sampleno].cpu().numpy()[0])\n",
        "                    incorrect_indexes[sampleno] = {'actual': target[sampleno].cpu().numpy(), \n",
        "                                                   'predicted': pred[sampleno].cpu().numpy()[0], \n",
        "                                                   'data': data[sampleno].cpu().numpy()}\n",
        "\n",
        "#print(incorrect_indexes) \n",
        "print(len(incorrect_indexes))\n",
        "print(count)"
      ],
      "execution_count": 18,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "54\n",
            "69\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "SZ08UZXm2BWZ",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 848
        },
        "outputId": "7038c414-5518-48f1-904c-10bf5fd47c04"
      },
      "source": [
        "from matplotlib import pyplot as plt\n",
        "%matplotlib inline\n",
        "import numpy as np\n",
        "\n",
        "#{23: {'actual': 1, 'predicted': 4}}\n",
        "x = 0\n",
        "y = 0\n",
        "fig, axs = plt.subplots(5, 5, figsize = (15, 15))\n",
        "plt.setp(axs, xticks=[], yticks=[])\n",
        "fig.subplots_adjust(wspace=0.7)\n",
        "images = list(incorrect_indexes.items())[:25]\n",
        "for index, results in images:\n",
        "  #print(index)\n",
        "  img = results['data'] \n",
        "  img = np.squeeze(img)\n",
        "  actual_class = results['actual'] \n",
        "  predicted_class =  results['predicted'] \n",
        "\n",
        "  plt.savefig(\"misclassified.png\")\n",
        "  files.download(\"misclassified.png\")  \n",
        "  axs[x, y].imshow(img)\n",
        "  axs[x, y].set_title('Actual Class:' + str(actual_class) + \"\\nPredicted class: \" + str(predicted_class))\n",
        "  \n",
        "  if y == 4:\n",
        "    x += 1\n",
        "    y = 0\n",
        "  else:\n",
        "    y += 1\n",
        "\n",
        "plt.savefig(\"misclassified.png\")\n",
        "files.download(\"misclassified.png\")     "
      ],
      "execution_count": 19,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "\n",
              "    async function download(id, filename, size) {\n",
              "      if (!google.colab.kernel.accessAllowed) {\n",
              "        return;\n",
              "      }\n",
              "      const div = document.createElement('div');\n",
              "      const label = document.createElement('label');\n",
              "      label.textContent = `Downloading \"${filename}\": `;\n",
              "      div.appendChild(label);\n",
              "      const progress = document.createElement('progress');\n",
              "      progress.max = size;\n",
              "      div.appendChild(progress);\n",
              "      document.body.appendChild(div);\n",
              "\n",
              "      const buffers = [];\n",
              "      let downloaded = 0;\n",
              "\n",
              "      const channel = await google.colab.kernel.comms.open(id);\n",
              "      // Send a message to notify the kernel that we're ready.\n",
              "      channel.send({})\n",
              "\n",
              "      for await (const message of channel.messages) {\n",
              "        // Send a message to notify the kernel that we're ready.\n",
              "        channel.send({})\n",
              "        if (message.buffers) {\n",
              "          for (const buffer of message.buffers) {\n",
              "            buffers.push(buffer);\n",
              "            downloaded += buffer.byteLength;\n",
              "            progress.value = downloaded;\n",
              "          }\n",
              "        }\n",
              "      }\n",
              "      const blob = new Blob(buffers, {type: 'application/binary'});\n",
              "      const a = document.createElement('a');\n",
              "      a.href = window.URL.createObjectURL(blob);\n",
              "      a.download = filename;\n",
              "      div.appendChild(a);\n",
              "      a.click();\n",
              "      div.remove();\n",
              "    }\n",
              "  "
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "download(\"download_f9393c0c-e540-4a77-af85-ee1b80068fbb\", \"misclassified.png\", 7794)"
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "\n",
              "    async function download(id, filename, size) {\n",
              "      if (!google.colab.kernel.accessAllowed) {\n",
              "        return;\n",
              "      }\n",
              "      const div = document.createElement('div');\n",
              "      const label = document.createElement('label');\n",
              "      label.textContent = `Downloading \"${filename}\": `;\n",
              "      div.appendChild(label);\n",
              "      const progress = document.createElement('progress');\n",
              "      progress.max = size;\n",
              "      div.appendChild(progress);\n",
              "      document.body.appendChild(div);\n",
              "\n",
              "      const buffers = [];\n",
              "      let downloaded = 0;\n",
              "\n",
              "      const channel = await google.colab.kernel.comms.open(id);\n",
              "      // Send a message to notify the kernel that we're ready.\n",
              "      channel.send({})\n",
              "\n",
              "      for await (const message of channel.messages) {\n",
              "        // Send a message to notify the kernel that we're ready.\n",
              "        channel.send({})\n",
              "        if (message.buffers) {\n",
              "          for (const buffer of message.buffers) {\n",
              "            buffers.push(buffer);\n",
              "            downloaded += buffer.byteLength;\n",
              "            progress.value = downloaded;\n",
              "          }\n",
              "        }\n",
              "      }\n",
              "      const blob = new Blob(buffers, {type: 'application/binary'});\n",
              "      const a = document.createElement('a');\n",
              "      a.href = window.URL.createObjectURL(blob);\n",
              "      a.download = filename;\n",
              "      div.appendChild(a);\n",
              "      a.click();\n",
              "      div.remove();\n",
              "    }\n",
              "  "
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "download(\"download_665e6bfe-68f8-423f-8081-3e9ef4a1ffc3\", \"misclassified.png\", 12883)"
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "\n",
              "    async function download(id, filename, size) {\n",
              "      if (!google.colab.kernel.accessAllowed) {\n",
              "        return;\n",
              "      }\n",
              "      const div = document.createElement('div');\n",
              "      const label = document.createElement('label');\n",
              "      label.textContent = `Downloading \"${filename}\": `;\n",
              "      div.appendChild(label);\n",
              "      const progress = document.createElement('progress');\n",
              "      progress.max = size;\n",
              "      div.appendChild(progress);\n",
              "      document.body.appendChild(div);\n",
              "\n",
              "      const buffers = [];\n",
              "      let downloaded = 0;\n",
              "\n",
              "      const channel = await google.colab.kernel.comms.open(id);\n",
              "      // Send a message to notify the kernel that we're ready.\n",
              "      channel.send({})\n",
              "\n",
              "      for await (const message of channel.messages) {\n",
              "        // Send a message to notify the kernel that we're ready.\n",
              "        channel.send({})\n",
              "        if (message.buffers) {\n",
              "          for (const buffer of message.buffers) {\n",
              "            buffers.push(buffer);\n",
              "            downloaded += buffer.byteLength;\n",
              "            progress.value = downloaded;\n",
              "          }\n",
              "        }\n",
              "      }\n",
              "      const blob = new Blob(buffers, {type: 'application/binary'});\n",
              "      const a = document.createElement('a');\n",
              "      a.href = window.URL.createObjectURL(blob);\n",
              "      a.download = filename;\n",
              "      div.appendChild(a);\n",
              "      a.click();\n",
              "      div.remove();\n",
              "    }\n",
              "  "
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "download(\"download_5787255d-2914-4478-9148-4ade34dacfb5\", \"misclassified.png\", 14144)"
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "\n",
              "    async function download(id, filename, size) {\n",
              "      if (!google.colab.kernel.accessAllowed) {\n",
              "        return;\n",
              "      }\n",
              "      const div = document.createElement('div');\n",
              "      const label = document.createElement('label');\n",
              "      label.textContent = `Downloading \"${filename}\": `;\n",
              "      div.appendChild(label);\n",
              "      const progress = document.createElement('progress');\n",
              "      progress.max = size;\n",
              "      div.appendChild(progress);\n",
              "      document.body.appendChild(div);\n",
              "\n",
              "      const buffers = [];\n",
              "      let downloaded = 0;\n",
              "\n",
              "      const channel = await google.colab.kernel.comms.open(id);\n",
              "      // Send a message to notify the kernel that we're ready.\n",
              "      channel.send({})\n",
              "\n",
              "      for await (const message of channel.messages) {\n",
              "        // Send a message to notify the kernel that we're ready.\n",
              "        channel.send({})\n",
              "        if (message.buffers) {\n",
              "          for (const buffer of message.buffers) {\n",
              "            buffers.push(buffer);\n",
              "            downloaded += buffer.byteLength;\n",
              "            progress.value = downloaded;\n",
              "          }\n",
              "        }\n",
              "      }\n",
              "      const blob = new Blob(buffers, {type: 'application/binary'});\n",
              "      const a = document.createElement('a');\n",
              "      a.href = window.URL.createObjectURL(blob);\n",
              "      a.download = filename;\n",
              "      div.appendChild(a);\n",
              "      a.click();\n",
              "      div.remove();\n",
              "    }\n",
              "  "
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "download(\"download_66ad0d49-edc3-4b7c-b9f9-bde861877170\", \"misclassified.png\", 15071)"
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "\n",
              "    async function download(id, filename, size) {\n",
              "      if (!google.colab.kernel.accessAllowed) {\n",
              "        return;\n",
              "      }\n",
              "      const div = document.createElement('div');\n",
              "      const label = document.createElement('label');\n",
              "      label.textContent = `Downloading \"${filename}\": `;\n",
              "      div.appendChild(label);\n",
              "      const progress = document.createElement('progress');\n",
              "      progress.max = size;\n",
              "      div.appendChild(progress);\n",
              "      document.body.appendChild(div);\n",
              "\n",
              "      const buffers = [];\n",
              "      let downloaded = 0;\n",
              "\n",
              "      const channel = await google.colab.kernel.comms.open(id);\n",
              "      // Send a message to notify the kernel that we're ready.\n",
              "      channel.send({})\n",
              "\n",
              "      for await (const message of channel.messages) {\n",
              "        // Send a message to notify the kernel that we're ready.\n",
              "        channel.send({})\n",
              "        if (message.buffers) {\n",
              "          for (const buffer of message.buffers) {\n",
              "            buffers.push(buffer);\n",
              "            downloaded += buffer.byteLength;\n",
              "            progress.value = downloaded;\n",
              "          }\n",
              "        }\n",
              "      }\n",
              "      const blob = new Blob(buffers, {type: 'application/binary'});\n",
              "      const a = document.createElement('a');\n",
              "      a.href = window.URL.createObjectURL(blob);\n",
              "      a.download = filename;\n",
              "      div.appendChild(a);\n",
              "      a.click();\n",
              "      div.remove();\n",
              "    }\n",
              "  "
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "download(\"download_37a77728-60b1-4f0f-8e4a-b2e2eac41718\", \"misclassified.png\", 16322)"
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "\n",
              "    async function download(id, filename, size) {\n",
              "      if (!google.colab.kernel.accessAllowed) {\n",
              "        return;\n",
              "      }\n",
              "      const div = document.createElement('div');\n",
              "      const label = document.createElement('label');\n",
              "      label.textContent = `Downloading \"${filename}\": `;\n",
              "      div.appendChild(label);\n",
              "      const progress = document.createElement('progress');\n",
              "      progress.max = size;\n",
              "      div.appendChild(progress);\n",
              "      document.body.appendChild(div);\n",
              "\n",
              "      const buffers = [];\n",
              "      let downloaded = 0;\n",
              "\n",
              "      const channel = await google.colab.kernel.comms.open(id);\n",
              "      // Send a message to notify the kernel that we're ready.\n",
              "      channel.send({})\n",
              "\n",
              "      for await (const message of channel.messages) {\n",
              "        // Send a message to notify the kernel that we're ready.\n",
              "        channel.send({})\n",
              "        if (message.buffers) {\n",
              "          for (const buffer of message.buffers) {\n",
              "            buffers.push(buffer);\n",
              "            downloaded += buffer.byteLength;\n",
              "            progress.value = downloaded;\n",
              "          }\n",
              "        }\n",
              "      }\n",
              "      const blob = new Blob(buffers, {type: 'application/binary'});\n",
              "      const a = document.createElement('a');\n",
              "      a.href = window.URL.createObjectURL(blob);\n",
              "      a.download = filename;\n",
              "      div.appendChild(a);\n",
              "      a.click();\n",
              "      div.remove();\n",
              "    }\n",
              "  "
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "download(\"download_166f2408-caae-4ec7-8311-cff6d9ab8b9d\", \"misclassified.png\", 17738)"
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "\n",
              "    async function download(id, filename, size) {\n",
              "      if (!google.colab.kernel.accessAllowed) {\n",
              "        return;\n",
              "      }\n",
              "      const div = document.createElement('div');\n",
              "      const label = document.createElement('label');\n",
              "      label.textContent = `Downloading \"${filename}\": `;\n",
              "      div.appendChild(label);\n",
              "      const progress = document.createElement('progress');\n",
              "      progress.max = size;\n",
              "      div.appendChild(progress);\n",
              "      document.body.appendChild(div);\n",
              "\n",
              "      const buffers = [];\n",
              "      let downloaded = 0;\n",
              "\n",
              "      const channel = await google.colab.kernel.comms.open(id);\n",
              "      // Send a message to notify the kernel that we're ready.\n",
              "      channel.send({})\n",
              "\n",
              "      for await (const message of channel.messages) {\n",
              "        // Send a message to notify the kernel that we're ready.\n",
              "        channel.send({})\n",
              "        if (message.buffers) {\n",
              "          for (const buffer of message.buffers) {\n",
              "            buffers.push(buffer);\n",
              "            downloaded += buffer.byteLength;\n",
              "            progress.value = downloaded;\n",
              "          }\n",
              "        }\n",
              "      }\n",
              "      const blob = new Blob(buffers, {type: 'application/binary'});\n",
              "      const a = document.createElement('a');\n",
              "      a.href = window.URL.createObjectURL(blob);\n",
              "      a.download = filename;\n",
              "      div.appendChild(a);\n",
              "      a.click();\n",
              "      div.remove();\n",
              "    }\n",
              "  "
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "download(\"download_a608f47f-c6a6-40e9-996d-42f66a19b930\", \"misclassified.png\", 22367)"
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "\n",
              "    async function download(id, filename, size) {\n",
              "      if (!google.colab.kernel.accessAllowed) {\n",
              "        return;\n",
              "      }\n",
              "      const div = document.createElement('div');\n",
              "      const label = document.createElement('label');\n",
              "      label.textContent = `Downloading \"${filename}\": `;\n",
              "      div.appendChild(label);\n",
              "      const progress = document.createElement('progress');\n",
              "      progress.max = size;\n",
              "      div.appendChild(progress);\n",
              "      document.body.appendChild(div);\n",
              "\n",
              "      const buffers = [];\n",
              "      let downloaded = 0;\n",
              "\n",
              "      const channel = await google.colab.kernel.comms.open(id);\n",
              "      // Send a message to notify the kernel that we're ready.\n",
              "      channel.send({})\n",
              "\n",
              "      for await (const message of channel.messages) {\n",
              "        // Send a message to notify the kernel that we're ready.\n",
              "        channel.send({})\n",
              "        if (message.buffers) {\n",
              "          for (const buffer of message.buffers) {\n",
              "            buffers.push(buffer);\n",
              "            downloaded += buffer.byteLength;\n",
              "            progress.value = downloaded;\n",
              "          }\n",
              "        }\n",
              "      }\n",
              "      const blob = new Blob(buffers, {type: 'application/binary'});\n",
              "      const a = document.createElement('a');\n",
              "      a.href = window.URL.createObjectURL(blob);\n",
              "      a.download = filename;\n",
              "      div.appendChild(a);\n",
              "      a.click();\n",
              "      div.remove();\n",
              "    }\n",
              "  "
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "download(\"download_1f682cb2-2f22-4631-b0f8-73fe64c8cecb\", \"misclassified.png\", 23485)"
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "\n",
              "    async function download(id, filename, size) {\n",
              "      if (!google.colab.kernel.accessAllowed) {\n",
              "        return;\n",
              "      }\n",
              "      const div = document.createElement('div');\n",
              "      const label = document.createElement('label');\n",
              "      label.textContent = `Downloading \"${filename}\": `;\n",
              "      div.appendChild(label);\n",
              "      const progress = document.createElement('progress');\n",
              "      progress.max = size;\n",
              "      div.appendChild(progress);\n",
              "      document.body.appendChild(div);\n",
              "\n",
              "      const buffers = [];\n",
              "      let downloaded = 0;\n",
              "\n",
              "      const channel = await google.colab.kernel.comms.open(id);\n",
              "      // Send a message to notify the kernel that we're ready.\n",
              "      channel.send({})\n",
              "\n",
              "      for await (const message of channel.messages) {\n",
              "        // Send a message to notify the kernel that we're ready.\n",
              "        channel.send({})\n",
              "        if (message.buffers) {\n",
              "          for (const buffer of message.buffers) {\n",
              "            buffers.push(buffer);\n",
              "            downloaded += buffer.byteLength;\n",
              "            progress.value = downloaded;\n",
              "          }\n",
              "        }\n",
              "      }\n",
              "      const blob = new Blob(buffers, {type: 'application/binary'});\n",
              "      const a = document.createElement('a');\n",
              "      a.href = window.URL.createObjectURL(blob);\n",
              "      a.download = filename;\n",
              "      div.appendChild(a);\n",
              "      a.click();\n",
              "      div.remove();\n",
              "    }\n",
              "  "
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "download(\"download_b1d77172-2fe4-41d3-8fa9-d9067fbe827b\", \"misclassified.png\", 24531)"
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "\n",
              "    async function download(id, filename, size) {\n",
              "      if (!google.colab.kernel.accessAllowed) {\n",
              "        return;\n",
              "      }\n",
              "      const div = document.createElement('div');\n",
              "      const label = document.createElement('label');\n",
              "      label.textContent = `Downloading \"${filename}\": `;\n",
              "      div.appendChild(label);\n",
              "      const progress = document.createElement('progress');\n",
              "      progress.max = size;\n",
              "      div.appendChild(progress);\n",
              "      document.body.appendChild(div);\n",
              "\n",
              "      const buffers = [];\n",
              "      let downloaded = 0;\n",
              "\n",
              "      const channel = await google.colab.kernel.comms.open(id);\n",
              "      // Send a message to notify the kernel that we're ready.\n",
              "      channel.send({})\n",
              "\n",
              "      for await (const message of channel.messages) {\n",
              "        // Send a message to notify the kernel that we're ready.\n",
              "        channel.send({})\n",
              "        if (message.buffers) {\n",
              "          for (const buffer of message.buffers) {\n",
              "            buffers.push(buffer);\n",
              "            downloaded += buffer.byteLength;\n",
              "            progress.value = downloaded;\n",
              "          }\n",
              "        }\n",
              "      }\n",
              "      const blob = new Blob(buffers, {type: 'application/binary'});\n",
              "      const a = document.createElement('a');\n",
              "      a.href = window.URL.createObjectURL(blob);\n",
              "      a.download = filename;\n",
              "      div.appendChild(a);\n",
              "      a.click();\n",
              "      div.remove();\n",
              "    }\n",
              "  "
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "download(\"download_cc9567db-1a2b-4817-8d2c-cf4524b8df13\", \"misclassified.png\", 25408)"
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "\n",
              "    async function download(id, filename, size) {\n",
              "      if (!google.colab.kernel.accessAllowed) {\n",
              "        return;\n",
              "      }\n",
              "      const div = document.createElement('div');\n",
              "      const label = document.createElement('label');\n",
              "      label.textContent = `Downloading \"${filename}\": `;\n",
              "      div.appendChild(label);\n",
              "      const progress = document.createElement('progress');\n",
              "      progress.max = size;\n",
              "      div.appendChild(progress);\n",
              "      document.body.appendChild(div);\n",
              "\n",
              "      const buffers = [];\n",
              "      let downloaded = 0;\n",
              "\n",
              "      const channel = await google.colab.kernel.comms.open(id);\n",
              "      // Send a message to notify the kernel that we're ready.\n",
              "      channel.send({})\n",
              "\n",
              "      for await (const message of channel.messages) {\n",
              "        // Send a message to notify the kernel that we're ready.\n",
              "        channel.send({})\n",
              "        if (message.buffers) {\n",
              "          for (const buffer of message.buffers) {\n",
              "            buffers.push(buffer);\n",
              "            downloaded += buffer.byteLength;\n",
              "            progress.value = downloaded;\n",
              "          }\n",
              "        }\n",
              "      }\n",
              "      const blob = new Blob(buffers, {type: 'application/binary'});\n",
              "      const a = document.createElement('a');\n",
              "      a.href = window.URL.createObjectURL(blob);\n",
              "      a.download = filename;\n",
              "      div.appendChild(a);\n",
              "      a.click();\n",
              "      div.remove();\n",
              "    }\n",
              "  "
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "download(\"download_5547b584-299b-4ea6-8d34-806c7bd7e779\", \"misclassified.png\", 26731)"
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "\n",
              "    async function download(id, filename, size) {\n",
              "      if (!google.colab.kernel.accessAllowed) {\n",
              "        return;\n",
              "      }\n",
              "      const div = document.createElement('div');\n",
              "      const label = document.createElement('label');\n",
              "      label.textContent = `Downloading \"${filename}\": `;\n",
              "      div.appendChild(label);\n",
              "      const progress = document.createElement('progress');\n",
              "      progress.max = size;\n",
              "      div.appendChild(progress);\n",
              "      document.body.appendChild(div);\n",
              "\n",
              "      const buffers = [];\n",
              "      let downloaded = 0;\n",
              "\n",
              "      const channel = await google.colab.kernel.comms.open(id);\n",
              "      // Send a message to notify the kernel that we're ready.\n",
              "      channel.send({})\n",
              "\n",
              "      for await (const message of channel.messages) {\n",
              "        // Send a message to notify the kernel that we're ready.\n",
              "        channel.send({})\n",
              "        if (message.buffers) {\n",
              "          for (const buffer of message.buffers) {\n",
              "            buffers.push(buffer);\n",
              "            downloaded += buffer.byteLength;\n",
              "            progress.value = downloaded;\n",
              "          }\n",
              "        }\n",
              "      }\n",
              "      const blob = new Blob(buffers, {type: 'application/binary'});\n",
              "      const a = document.createElement('a');\n",
              "      a.href = window.URL.createObjectURL(blob);\n",
              "      a.download = filename;\n",
              "      div.appendChild(a);\n",
              "      a.click();\n",
              "      div.remove();\n",
              "    }\n",
              "  "
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "download(\"download_a445fb30-cbeb-4395-b6d9-b19f7ee3fb8d\", \"misclassified.png\", 31366)"
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "\n",
              "    async function download(id, filename, size) {\n",
              "      if (!google.colab.kernel.accessAllowed) {\n",
              "        return;\n",
              "      }\n",
              "      const div = document.createElement('div');\n",
              "      const label = document.createElement('label');\n",
              "      label.textContent = `Downloading \"${filename}\": `;\n",
              "      div.appendChild(label);\n",
              "      const progress = document.createElement('progress');\n",
              "      progress.max = size;\n",
              "      div.appendChild(progress);\n",
              "      document.body.appendChild(div);\n",
              "\n",
              "      const buffers = [];\n",
              "      let downloaded = 0;\n",
              "\n",
              "      const channel = await google.colab.kernel.comms.open(id);\n",
              "      // Send a message to notify the kernel that we're ready.\n",
              "      channel.send({})\n",
              "\n",
              "      for await (const message of channel.messages) {\n",
              "        // Send a message to notify the kernel that we're ready.\n",
              "        channel.send({})\n",
              "        if (message.buffers) {\n",
              "          for (const buffer of message.buffers) {\n",
              "            buffers.push(buffer);\n",
              "            downloaded += buffer.byteLength;\n",
              "            progress.value = downloaded;\n",
              "          }\n",
              "        }\n",
              "      }\n",
              "      const blob = new Blob(buffers, {type: 'application/binary'});\n",
              "      const a = document.createElement('a');\n",
              "      a.href = window.URL.createObjectURL(blob);\n",
              "      a.download = filename;\n",
              "      div.appendChild(a);\n",
              "      a.click();\n",
              "      div.remove();\n",
              "    }\n",
              "  "
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "download(\"download_9f896bc6-e7bc-4209-97cf-4eb1c19c3272\", \"misclassified.png\", 32669)"
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "\n",
              "    async function download(id, filename, size) {\n",
              "      if (!google.colab.kernel.accessAllowed) {\n",
              "        return;\n",
              "      }\n",
              "      const div = document.createElement('div');\n",
              "      const label = document.createElement('label');\n",
              "      label.textContent = `Downloading \"${filename}\": `;\n",
              "      div.appendChild(label);\n",
              "      const progress = document.createElement('progress');\n",
              "      progress.max = size;\n",
              "      div.appendChild(progress);\n",
              "      document.body.appendChild(div);\n",
              "\n",
              "      const buffers = [];\n",
              "      let downloaded = 0;\n",
              "\n",
              "      const channel = await google.colab.kernel.comms.open(id);\n",
              "      // Send a message to notify the kernel that we're ready.\n",
              "      channel.send({})\n",
              "\n",
              "      for await (const message of channel.messages) {\n",
              "        // Send a message to notify the kernel that we're ready.\n",
              "        channel.send({})\n",
              "        if (message.buffers) {\n",
              "          for (const buffer of message.buffers) {\n",
              "            buffers.push(buffer);\n",
              "            downloaded += buffer.byteLength;\n",
              "            progress.value = downloaded;\n",
              "          }\n",
              "        }\n",
              "      }\n",
              "      const blob = new Blob(buffers, {type: 'application/binary'});\n",
              "      const a = document.createElement('a');\n",
              "      a.href = window.URL.createObjectURL(blob);\n",
              "      a.download = filename;\n",
              "      div.appendChild(a);\n",
              "      a.click();\n",
              "      div.remove();\n",
              "    }\n",
              "  "
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "download(\"download_110b8546-7ab4-4ceb-b91a-0b286ac77cc5\", \"misclassified.png\", 33692)"
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "\n",
              "    async function download(id, filename, size) {\n",
              "      if (!google.colab.kernel.accessAllowed) {\n",
              "        return;\n",
              "      }\n",
              "      const div = document.createElement('div');\n",
              "      const label = document.createElement('label');\n",
              "      label.textContent = `Downloading \"${filename}\": `;\n",
              "      div.appendChild(label);\n",
              "      const progress = document.createElement('progress');\n",
              "      progress.max = size;\n",
              "      div.appendChild(progress);\n",
              "      document.body.appendChild(div);\n",
              "\n",
              "      const buffers = [];\n",
              "      let downloaded = 0;\n",
              "\n",
              "      const channel = await google.colab.kernel.comms.open(id);\n",
              "      // Send a message to notify the kernel that we're ready.\n",
              "      channel.send({})\n",
              "\n",
              "      for await (const message of channel.messages) {\n",
              "        // Send a message to notify the kernel that we're ready.\n",
              "        channel.send({})\n",
              "        if (message.buffers) {\n",
              "          for (const buffer of message.buffers) {\n",
              "            buffers.push(buffer);\n",
              "            downloaded += buffer.byteLength;\n",
              "            progress.value = downloaded;\n",
              "          }\n",
              "        }\n",
              "      }\n",
              "      const blob = new Blob(buffers, {type: 'application/binary'});\n",
              "      const a = document.createElement('a');\n",
              "      a.href = window.URL.createObjectURL(blob);\n",
              "      a.download = filename;\n",
              "      div.appendChild(a);\n",
              "      a.click();\n",
              "      div.remove();\n",
              "    }\n",
              "  "
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "download(\"download_7778183f-de65-4d9b-aea6-3491c1feed57\", \"misclassified.png\", 34508)"
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "\n",
              "    async function download(id, filename, size) {\n",
              "      if (!google.colab.kernel.accessAllowed) {\n",
              "        return;\n",
              "      }\n",
              "      const div = document.createElement('div');\n",
              "      const label = document.createElement('label');\n",
              "      label.textContent = `Downloading \"${filename}\": `;\n",
              "      div.appendChild(label);\n",
              "      const progress = document.createElement('progress');\n",
              "      progress.max = size;\n",
              "      div.appendChild(progress);\n",
              "      document.body.appendChild(div);\n",
              "\n",
              "      const buffers = [];\n",
              "      let downloaded = 0;\n",
              "\n",
              "      const channel = await google.colab.kernel.comms.open(id);\n",
              "      // Send a message to notify the kernel that we're ready.\n",
              "      channel.send({})\n",
              "\n",
              "      for await (const message of channel.messages) {\n",
              "        // Send a message to notify the kernel that we're ready.\n",
              "        channel.send({})\n",
              "        if (message.buffers) {\n",
              "          for (const buffer of message.buffers) {\n",
              "            buffers.push(buffer);\n",
              "            downloaded += buffer.byteLength;\n",
              "            progress.value = downloaded;\n",
              "          }\n",
              "        }\n",
              "      }\n",
              "      const blob = new Blob(buffers, {type: 'application/binary'});\n",
              "      const a = document.createElement('a');\n",
              "      a.href = window.URL.createObjectURL(blob);\n",
              "      a.download = filename;\n",
              "      div.appendChild(a);\n",
              "      a.click();\n",
              "      div.remove();\n",
              "    }\n",
              "  "
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "download(\"download_0b2b1a6c-f0e5-4edc-9fd1-cc2c28263d94\", \"misclassified.png\", 35759)"
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "\n",
              "    async function download(id, filename, size) {\n",
              "      if (!google.colab.kernel.accessAllowed) {\n",
              "        return;\n",
              "      }\n",
              "      const div = document.createElement('div');\n",
              "      const label = document.createElement('label');\n",
              "      label.textContent = `Downloading \"${filename}\": `;\n",
              "      div.appendChild(label);\n",
              "      const progress = document.createElement('progress');\n",
              "      progress.max = size;\n",
              "      div.appendChild(progress);\n",
              "      document.body.appendChild(div);\n",
              "\n",
              "      const buffers = [];\n",
              "      let downloaded = 0;\n",
              "\n",
              "      const channel = await google.colab.kernel.comms.open(id);\n",
              "      // Send a message to notify the kernel that we're ready.\n",
              "      channel.send({})\n",
              "\n",
              "      for await (const message of channel.messages) {\n",
              "        // Send a message to notify the kernel that we're ready.\n",
              "        channel.send({})\n",
              "        if (message.buffers) {\n",
              "          for (const buffer of message.buffers) {\n",
              "            buffers.push(buffer);\n",
              "            downloaded += buffer.byteLength;\n",
              "            progress.value = downloaded;\n",
              "          }\n",
              "        }\n",
              "      }\n",
              "      const blob = new Blob(buffers, {type: 'application/binary'});\n",
              "      const a = document.createElement('a');\n",
              "      a.href = window.URL.createObjectURL(blob);\n",
              "      a.download = filename;\n",
              "      div.appendChild(a);\n",
              "      a.click();\n",
              "      div.remove();\n",
              "    }\n",
              "  "
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "download(\"download_d02b01e6-c8ff-49f2-b3d0-96e6df072959\", \"misclassified.png\", 39971)"
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "\n",
              "    async function download(id, filename, size) {\n",
              "      if (!google.colab.kernel.accessAllowed) {\n",
              "        return;\n",
              "      }\n",
              "      const div = document.createElement('div');\n",
              "      const label = document.createElement('label');\n",
              "      label.textContent = `Downloading \"${filename}\": `;\n",
              "      div.appendChild(label);\n",
              "      const progress = document.createElement('progress');\n",
              "      progress.max = size;\n",
              "      div.appendChild(progress);\n",
              "      document.body.appendChild(div);\n",
              "\n",
              "      const buffers = [];\n",
              "      let downloaded = 0;\n",
              "\n",
              "      const channel = await google.colab.kernel.comms.open(id);\n",
              "      // Send a message to notify the kernel that we're ready.\n",
              "      channel.send({})\n",
              "\n",
              "      for await (const message of channel.messages) {\n",
              "        // Send a message to notify the kernel that we're ready.\n",
              "        channel.send({})\n",
              "        if (message.buffers) {\n",
              "          for (const buffer of message.buffers) {\n",
              "            buffers.push(buffer);\n",
              "            downloaded += buffer.byteLength;\n",
              "            progress.value = downloaded;\n",
              "          }\n",
              "        }\n",
              "      }\n",
              "      const blob = new Blob(buffers, {type: 'application/binary'});\n",
              "      const a = document.createElement('a');\n",
              "      a.href = window.URL.createObjectURL(blob);\n",
              "      a.download = filename;\n",
              "      div.appendChild(a);\n",
              "      a.click();\n",
              "      div.remove();\n",
              "    }\n",
              "  "
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "download(\"download_25ced2b0-619f-4c90-9f1d-25894d53809a\", \"misclassified.png\", 41239)"
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "\n",
              "    async function download(id, filename, size) {\n",
              "      if (!google.colab.kernel.accessAllowed) {\n",
              "        return;\n",
              "      }\n",
              "      const div = document.createElement('div');\n",
              "      const label = document.createElement('label');\n",
              "      label.textContent = `Downloading \"${filename}\": `;\n",
              "      div.appendChild(label);\n",
              "      const progress = document.createElement('progress');\n",
              "      progress.max = size;\n",
              "      div.appendChild(progress);\n",
              "      document.body.appendChild(div);\n",
              "\n",
              "      const buffers = [];\n",
              "      let downloaded = 0;\n",
              "\n",
              "      const channel = await google.colab.kernel.comms.open(id);\n",
              "      // Send a message to notify the kernel that we're ready.\n",
              "      channel.send({})\n",
              "\n",
              "      for await (const message of channel.messages) {\n",
              "        // Send a message to notify the kernel that we're ready.\n",
              "        channel.send({})\n",
              "        if (message.buffers) {\n",
              "          for (const buffer of message.buffers) {\n",
              "            buffers.push(buffer);\n",
              "            downloaded += buffer.byteLength;\n",
              "            progress.value = downloaded;\n",
              "          }\n",
              "        }\n",
              "      }\n",
              "      const blob = new Blob(buffers, {type: 'application/binary'});\n",
              "      const a = document.createElement('a');\n",
              "      a.href = window.URL.createObjectURL(blob);\n",
              "      a.download = filename;\n",
              "      div.appendChild(a);\n",
              "      a.click();\n",
              "      div.remove();\n",
              "    }\n",
              "  "
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "download(\"download_be484a78-f1e0-4fae-9578-98416224090d\", \"misclassified.png\", 42534)"
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "\n",
              "    async function download(id, filename, size) {\n",
              "      if (!google.colab.kernel.accessAllowed) {\n",
              "        return;\n",
              "      }\n",
              "      const div = document.createElement('div');\n",
              "      const label = document.createElement('label');\n",
              "      label.textContent = `Downloading \"${filename}\": `;\n",
              "      div.appendChild(label);\n",
              "      const progress = document.createElement('progress');\n",
              "      progress.max = size;\n",
              "      div.appendChild(progress);\n",
              "      document.body.appendChild(div);\n",
              "\n",
              "      const buffers = [];\n",
              "      let downloaded = 0;\n",
              "\n",
              "      const channel = await google.colab.kernel.comms.open(id);\n",
              "      // Send a message to notify the kernel that we're ready.\n",
              "      channel.send({})\n",
              "\n",
              "      for await (const message of channel.messages) {\n",
              "        // Send a message to notify the kernel that we're ready.\n",
              "        channel.send({})\n",
              "        if (message.buffers) {\n",
              "          for (const buffer of message.buffers) {\n",
              "            buffers.push(buffer);\n",
              "            downloaded += buffer.byteLength;\n",
              "            progress.value = downloaded;\n",
              "          }\n",
              "        }\n",
              "      }\n",
              "      const blob = new Blob(buffers, {type: 'application/binary'});\n",
              "      const a = document.createElement('a');\n",
              "      a.href = window.URL.createObjectURL(blob);\n",
              "      a.download = filename;\n",
              "      div.appendChild(a);\n",
              "      a.click();\n",
              "      div.remove();\n",
              "    }\n",
              "  "
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "download(\"download_3fc4b8a8-8ee9-4f7d-86dc-b9dcb2061303\", \"misclassified.png\", 44071)"
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "\n",
              "    async function download(id, filename, size) {\n",
              "      if (!google.colab.kernel.accessAllowed) {\n",
              "        return;\n",
              "      }\n",
              "      const div = document.createElement('div');\n",
              "      const label = document.createElement('label');\n",
              "      label.textContent = `Downloading \"${filename}\": `;\n",
              "      div.appendChild(label);\n",
              "      const progress = document.createElement('progress');\n",
              "      progress.max = size;\n",
              "      div.appendChild(progress);\n",
              "      document.body.appendChild(div);\n",
              "\n",
              "      const buffers = [];\n",
              "      let downloaded = 0;\n",
              "\n",
              "      const channel = await google.colab.kernel.comms.open(id);\n",
              "      // Send a message to notify the kernel that we're ready.\n",
              "      channel.send({})\n",
              "\n",
              "      for await (const message of channel.messages) {\n",
              "        // Send a message to notify the kernel that we're ready.\n",
              "        channel.send({})\n",
              "        if (message.buffers) {\n",
              "          for (const buffer of message.buffers) {\n",
              "            buffers.push(buffer);\n",
              "            downloaded += buffer.byteLength;\n",
              "            progress.value = downloaded;\n",
              "          }\n",
              "        }\n",
              "      }\n",
              "      const blob = new Blob(buffers, {type: 'application/binary'});\n",
              "      const a = document.createElement('a');\n",
              "      a.href = window.URL.createObjectURL(blob);\n",
              "      a.download = filename;\n",
              "      div.appendChild(a);\n",
              "      a.click();\n",
              "      div.remove();\n",
              "    }\n",
              "  "
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "download(\"download_8318a967-335e-4c38-b0e6-eded27955415\", \"misclassified.png\", 45022)"
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "\n",
              "    async function download(id, filename, size) {\n",
              "      if (!google.colab.kernel.accessAllowed) {\n",
              "        return;\n",
              "      }\n",
              "      const div = document.createElement('div');\n",
              "      const label = document.createElement('label');\n",
              "      label.textContent = `Downloading \"${filename}\": `;\n",
              "      div.appendChild(label);\n",
              "      const progress = document.createElement('progress');\n",
              "      progress.max = size;\n",
              "      div.appendChild(progress);\n",
              "      document.body.appendChild(div);\n",
              "\n",
              "      const buffers = [];\n",
              "      let downloaded = 0;\n",
              "\n",
              "      const channel = await google.colab.kernel.comms.open(id);\n",
              "      // Send a message to notify the kernel that we're ready.\n",
              "      channel.send({})\n",
              "\n",
              "      for await (const message of channel.messages) {\n",
              "        // Send a message to notify the kernel that we're ready.\n",
              "        channel.send({})\n",
              "        if (message.buffers) {\n",
              "          for (const buffer of message.buffers) {\n",
              "            buffers.push(buffer);\n",
              "            downloaded += buffer.byteLength;\n",
              "            progress.value = downloaded;\n",
              "          }\n",
              "        }\n",
              "      }\n",
              "      const blob = new Blob(buffers, {type: 'application/binary'});\n",
              "      const a = document.createElement('a');\n",
              "      a.href = window.URL.createObjectURL(blob);\n",
              "      a.download = filename;\n",
              "      div.appendChild(a);\n",
              "      a.click();\n",
              "      div.remove();\n",
              "    }\n",
              "  "
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "download(\"download_c7b4f414-40ff-4b47-a7e3-f5a427772fe5\", \"misclassified.png\", 49244)"
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "\n",
              "    async function download(id, filename, size) {\n",
              "      if (!google.colab.kernel.accessAllowed) {\n",
              "        return;\n",
              "      }\n",
              "      const div = document.createElement('div');\n",
              "      const label = document.createElement('label');\n",
              "      label.textContent = `Downloading \"${filename}\": `;\n",
              "      div.appendChild(label);\n",
              "      const progress = document.createElement('progress');\n",
              "      progress.max = size;\n",
              "      div.appendChild(progress);\n",
              "      document.body.appendChild(div);\n",
              "\n",
              "      const buffers = [];\n",
              "      let downloaded = 0;\n",
              "\n",
              "      const channel = await google.colab.kernel.comms.open(id);\n",
              "      // Send a message to notify the kernel that we're ready.\n",
              "      channel.send({})\n",
              "\n",
              "      for await (const message of channel.messages) {\n",
              "        // Send a message to notify the kernel that we're ready.\n",
              "        channel.send({})\n",
              "        if (message.buffers) {\n",
              "          for (const buffer of message.buffers) {\n",
              "            buffers.push(buffer);\n",
              "            downloaded += buffer.byteLength;\n",
              "            progress.value = downloaded;\n",
              "          }\n",
              "        }\n",
              "      }\n",
              "      const blob = new Blob(buffers, {type: 'application/binary'});\n",
              "      const a = document.createElement('a');\n",
              "      a.href = window.URL.createObjectURL(blob);\n",
              "      a.download = filename;\n",
              "      div.appendChild(a);\n",
              "      a.click();\n",
              "      div.remove();\n",
              "    }\n",
              "  "
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "download(\"download_092bfdfe-8c6d-4601-aa5c-404959a8c14a\", \"misclassified.png\", 50264)"
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "\n",
              "    async function download(id, filename, size) {\n",
              "      if (!google.colab.kernel.accessAllowed) {\n",
              "        return;\n",
              "      }\n",
              "      const div = document.createElement('div');\n",
              "      const label = document.createElement('label');\n",
              "      label.textContent = `Downloading \"${filename}\": `;\n",
              "      div.appendChild(label);\n",
              "      const progress = document.createElement('progress');\n",
              "      progress.max = size;\n",
              "      div.appendChild(progress);\n",
              "      document.body.appendChild(div);\n",
              "\n",
              "      const buffers = [];\n",
              "      let downloaded = 0;\n",
              "\n",
              "      const channel = await google.colab.kernel.comms.open(id);\n",
              "      // Send a message to notify the kernel that we're ready.\n",
              "      channel.send({})\n",
              "\n",
              "      for await (const message of channel.messages) {\n",
              "        // Send a message to notify the kernel that we're ready.\n",
              "        channel.send({})\n",
              "        if (message.buffers) {\n",
              "          for (const buffer of message.buffers) {\n",
              "            buffers.push(buffer);\n",
              "            downloaded += buffer.byteLength;\n",
              "            progress.value = downloaded;\n",
              "          }\n",
              "        }\n",
              "      }\n",
              "      const blob = new Blob(buffers, {type: 'application/binary'});\n",
              "      const a = document.createElement('a');\n",
              "      a.href = window.URL.createObjectURL(blob);\n",
              "      a.download = filename;\n",
              "      div.appendChild(a);\n",
              "      a.click();\n",
              "      div.remove();\n",
              "    }\n",
              "  "
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "download(\"download_ea644dcd-0866-4ca6-8255-bda056f61575\", \"misclassified.png\", 51404)"
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "\n",
              "    async function download(id, filename, size) {\n",
              "      if (!google.colab.kernel.accessAllowed) {\n",
              "        return;\n",
              "      }\n",
              "      const div = document.createElement('div');\n",
              "      const label = document.createElement('label');\n",
              "      label.textContent = `Downloading \"${filename}\": `;\n",
              "      div.appendChild(label);\n",
              "      const progress = document.createElement('progress');\n",
              "      progress.max = size;\n",
              "      div.appendChild(progress);\n",
              "      document.body.appendChild(div);\n",
              "\n",
              "      const buffers = [];\n",
              "      let downloaded = 0;\n",
              "\n",
              "      const channel = await google.colab.kernel.comms.open(id);\n",
              "      // Send a message to notify the kernel that we're ready.\n",
              "      channel.send({})\n",
              "\n",
              "      for await (const message of channel.messages) {\n",
              "        // Send a message to notify the kernel that we're ready.\n",
              "        channel.send({})\n",
              "        if (message.buffers) {\n",
              "          for (const buffer of message.buffers) {\n",
              "            buffers.push(buffer);\n",
              "            downloaded += buffer.byteLength;\n",
              "            progress.value = downloaded;\n",
              "          }\n",
              "        }\n",
              "      }\n",
              "      const blob = new Blob(buffers, {type: 'application/binary'});\n",
              "      const a = document.createElement('a');\n",
              "      a.href = window.URL.createObjectURL(blob);\n",
              "      a.download = filename;\n",
              "      div.appendChild(a);\n",
              "      a.click();\n",
              "      div.remove();\n",
              "    }\n",
              "  "
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "download(\"download_27a590d1-7201-4e5f-b814-a6bc864a947e\", \"misclassified.png\", 52175)"
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "\n",
              "    async function download(id, filename, size) {\n",
              "      if (!google.colab.kernel.accessAllowed) {\n",
              "        return;\n",
              "      }\n",
              "      const div = document.createElement('div');\n",
              "      const label = document.createElement('label');\n",
              "      label.textContent = `Downloading \"${filename}\": `;\n",
              "      div.appendChild(label);\n",
              "      const progress = document.createElement('progress');\n",
              "      progress.max = size;\n",
              "      div.appendChild(progress);\n",
              "      document.body.appendChild(div);\n",
              "\n",
              "      const buffers = [];\n",
              "      let downloaded = 0;\n",
              "\n",
              "      const channel = await google.colab.kernel.comms.open(id);\n",
              "      // Send a message to notify the kernel that we're ready.\n",
              "      channel.send({})\n",
              "\n",
              "      for await (const message of channel.messages) {\n",
              "        // Send a message to notify the kernel that we're ready.\n",
              "        channel.send({})\n",
              "        if (message.buffers) {\n",
              "          for (const buffer of message.buffers) {\n",
              "            buffers.push(buffer);\n",
              "            downloaded += buffer.byteLength;\n",
              "            progress.value = downloaded;\n",
              "          }\n",
              "        }\n",
              "      }\n",
              "      const blob = new Blob(buffers, {type: 'application/binary'});\n",
              "      const a = document.createElement('a');\n",
              "      a.href = window.URL.createObjectURL(blob);\n",
              "      a.download = filename;\n",
              "      div.appendChild(a);\n",
              "      a.click();\n",
              "      div.remove();\n",
              "    }\n",
              "  "
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "download(\"download_b4874ed6-c98f-4327-945f-3e2ad3a460df\", \"misclassified.png\", 53212)"
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1080x1080 with 25 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    }
  ]
}