07_food_vision_milestone_project_1.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "07_food_vision_milestone_project_1.ipynb",
      "provenance": [],
      "collapsed_sections": [
        "abkqNyLtTNeg",
        "vtgDiXpja-ot",
        "bM34xgP3dHuc",
        "STMO4jHiiAZ3"
      ],
      "toc_visible": true,
      "mount_file_id": "1ubO_evp5_pQh10BB-D9tILQUYy-3bWdl",
      "authorship_tag": "ABX9TyMoOFWHtrrXZdIls5yD/qMd",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU",
    "widgets": {
      "application/vnd.jupyter.widget-state+json": {
        "40602fc20b4d4605853ed0faf97e2e0e": {
          "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_0158f9dd514141a7b1ddab842f332a18",
            "_model_module": "@jupyter-widgets/controls",
            "children": [
              "IPY_MODEL_14d74cd26c5341bf815b523b908aaf92",
              "IPY_MODEL_d1a36386fdd24ad492d05ad866164a64"
            ]
          }
        },
        "0158f9dd514141a7b1ddab842f332a18": {
          "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
          }
        },
        "14d74cd26c5341bf815b523b908aaf92": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "state": {
            "_view_name": "ProgressView",
            "style": "IPY_MODEL_03c6156d6998490aae887f2d92f4963d",
            "_dom_classes": [],
            "description": "Dl Completed...: 100%",
            "_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_a225930f0be14b87a45194a51791817b"
          }
        },
        "d1a36386fdd24ad492d05ad866164a64": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_9bdb4563943d42e0bb55dd4f16618701",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": " 1/1 [04:28&lt;00:00, 268.84s/ url]",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_c69c211937f44be6a9d78f94f126ae7d"
          }
        },
        "03c6156d6998490aae887f2d92f4963d": {
          "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"
          }
        },
        "a225930f0be14b87a45194a51791817b": {
          "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
          }
        },
        "9bdb4563943d42e0bb55dd4f16618701": {
          "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"
          }
        },
        "c69c211937f44be6a9d78f94f126ae7d": {
          "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
          }
        },
        "cdf16990d08947f4ace37859e6a96af0": {
          "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_fa9236bfd3e14e9cbb321b003b608819",
            "_model_module": "@jupyter-widgets/controls",
            "children": [
              "IPY_MODEL_950371e34a4e4d76b1747d09c56c75ea",
              "IPY_MODEL_92e7697f4c6a42dfb6d3b37d32f86ab9"
            ]
          }
        },
        "fa9236bfd3e14e9cbb321b003b608819": {
          "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
          }
        },
        "950371e34a4e4d76b1747d09c56c75ea": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "state": {
            "_view_name": "ProgressView",
            "style": "IPY_MODEL_f9b31a3a01fa423aaaab1e9067917f25",
            "_dom_classes": [],
            "description": "Dl Size...: 100%",
            "_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_8f910e6f3f194982bbe52e85e4a02ff1"
          }
        },
        "92e7697f4c6a42dfb6d3b37d32f86ab9": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_c17f5f4e9f674facb06e40b5441f7e36",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": " 4764/4764 [04:28&lt;00:00, 17.72 MiB/s]",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_b9aede1d1d304f9f9665ea4db7acd258"
          }
        },
        "f9b31a3a01fa423aaaab1e9067917f25": {
          "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"
          }
        },
        "8f910e6f3f194982bbe52e85e4a02ff1": {
          "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
          }
        },
        "c17f5f4e9f674facb06e40b5441f7e36": {
          "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"
          }
        },
        "b9aede1d1d304f9f9665ea4db7acd258": {
          "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
          }
        },
        "b86540e3b06f4e80a47014985ce629d1": {
          "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_eff9872c3de94d728b082c747d8e848c",
            "_model_module": "@jupyter-widgets/controls",
            "children": [
              "IPY_MODEL_b512b87c671a4a7fb18fef1229604266",
              "IPY_MODEL_d3364328b5504e23ae48dc105fd1a539"
            ]
          }
        },
        "eff9872c3de94d728b082c747d8e848c": {
          "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
          }
        },
        "b512b87c671a4a7fb18fef1229604266": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "state": {
            "_view_name": "ProgressView",
            "style": "IPY_MODEL_3804c882a3ae42019ca333bcbaf58ef5",
            "_dom_classes": [],
            "description": "Extraction completed...: 100%",
            "_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_2c09c404ae304bbea7a621948bd0564e"
          }
        },
        "d3364328b5504e23ae48dc105fd1a539": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_aeedcdfc18b348e7a0a60ffe24d2018b",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": " 1/1 [04:28&lt;00:00, 268.75s/ file]",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_6aa735755bd04e3ca1bed195a82fdd85"
          }
        },
        "3804c882a3ae42019ca333bcbaf58ef5": {
          "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"
          }
        },
        "2c09c404ae304bbea7a621948bd0564e": {
          "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
          }
        },
        "aeedcdfc18b348e7a0a60ffe24d2018b": {
          "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"
          }
        },
        "6aa735755bd04e3ca1bed195a82fdd85": {
          "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
          }
        },
        "b5fc0fe99bf945cd8b34d6306a92d1bf": {
          "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_f75325f868504b6cb68dd17f740201e3",
            "_model_module": "@jupyter-widgets/controls",
            "children": [
              "IPY_MODEL_15dd8ed1d3fb4092b0b368bcd407da7d",
              "IPY_MODEL_3d7f4186421f4d739e07fb60d4db32fa"
            ]
          }
        },
        "f75325f868504b6cb68dd17f740201e3": {
          "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
          }
        },
        "15dd8ed1d3fb4092b0b368bcd407da7d": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "state": {
            "_view_name": "ProgressView",
            "style": "IPY_MODEL_70893e8dc2624c0eb3239e573f630814",
            "_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_66a2e5a432564d8399a857c6ddb92745"
          }
        },
        "3d7f4186421f4d739e07fb60d4db32fa": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_947c5f07f095414cb58e5e24f394335c",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": " 75750/0 [01:52&lt;00:00, 634.63 examples/s]",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_236cb5b12125417b8b3d838054131e1e"
          }
        },
        "70893e8dc2624c0eb3239e573f630814": {
          "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"
          }
        },
        "66a2e5a432564d8399a857c6ddb92745": {
          "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
          }
        },
        "947c5f07f095414cb58e5e24f394335c": {
          "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"
          }
        },
        "236cb5b12125417b8b3d838054131e1e": {
          "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
          }
        },
        "8be1726c9f7f48db93a64f1733af1a81": {
          "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_1b4e3abcb390482eb9c734c92585f461",
            "_model_module": "@jupyter-widgets/controls",
            "children": [
              "IPY_MODEL_5cdef67f62e74f50b2b7fd66e026e3fa",
              "IPY_MODEL_14e4b4f7bd2b45a2b3382d834d97a25c"
            ]
          }
        },
        "1b4e3abcb390482eb9c734c92585f461": {
          "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
          }
        },
        "5cdef67f62e74f50b2b7fd66e026e3fa": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "state": {
            "_view_name": "ProgressView",
            "style": "IPY_MODEL_9966f9a263d742d3a55a0c91b1187cc8",
            "_dom_classes": [],
            "description": "100%",
            "_model_name": "FloatProgressModel",
            "bar_style": "danger",
            "max": 75750,
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": 75601,
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "orientation": "horizontal",
            "min": 0,
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_75171aeca5504699bdc651151955ffff"
          }
        },
        "14e4b4f7bd2b45a2b3382d834d97a25c": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_d9bbe81e95b74333bbdc097b1f1ada9c",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": " 75601/75750 [01:40&lt;00:00, 549.51 examples/s]",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_c71afcd76afc4796a03b1e7cf874ce44"
          }
        },
        "9966f9a263d742d3a55a0c91b1187cc8": {
          "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"
          }
        },
        "75171aeca5504699bdc651151955ffff": {
          "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
          }
        },
        "d9bbe81e95b74333bbdc097b1f1ada9c": {
          "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"
          }
        },
        "c71afcd76afc4796a03b1e7cf874ce44": {
          "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
          }
        },
        "abe518c58101448b94fdf73282489bb7": {
          "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_bf7ae08a885f4f44b7a4f927b92b0230",
            "_model_module": "@jupyter-widgets/controls",
            "children": [
              "IPY_MODEL_c8d4c232ead44544ab4bfe783cf3cb21",
              "IPY_MODEL_0c49f2c3e9f348dfa7b8b1e0d9686885"
            ]
          }
        },
        "bf7ae08a885f4f44b7a4f927b92b0230": {
          "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
          }
        },
        "c8d4c232ead44544ab4bfe783cf3cb21": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "state": {
            "_view_name": "ProgressView",
            "style": "IPY_MODEL_87d247d21ca5433eb9836f981fb3da71",
            "_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_78891148fbbf455cb1c19dcdc4e75f08"
          }
        },
        "0c49f2c3e9f348dfa7b8b1e0d9686885": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_983f45c6456d46fb8620d8a4156fccf6",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": " 25250/0 [00:52&lt;00:00, 515.21 examples/s]",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_9eeb036a10ab42a3ab9e7947380fac76"
          }
        },
        "87d247d21ca5433eb9836f981fb3da71": {
          "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"
          }
        },
        "78891148fbbf455cb1c19dcdc4e75f08": {
          "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
          }
        },
        "983f45c6456d46fb8620d8a4156fccf6": {
          "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"
          }
        },
        "9eeb036a10ab42a3ab9e7947380fac76": {
          "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
          }
        },
        "c41f6dbbd8844335a0d066c623e23e29": {
          "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_399774dfa9984e7ca51f5d95c963d655",
            "_model_module": "@jupyter-widgets/controls",
            "children": [
              "IPY_MODEL_2022003efbc44e09a0f8c785181424d5",
              "IPY_MODEL_7a3d634f65b6404483a1b7a5801a68ae"
            ]
          }
        },
        "399774dfa9984e7ca51f5d95c963d655": {
          "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
          }
        },
        "2022003efbc44e09a0f8c785181424d5": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "state": {
            "_view_name": "ProgressView",
            "style": "IPY_MODEL_2687a81f54fc4655be5bd42cb4c3254a",
            "_dom_classes": [],
            "description": "100%",
            "_model_name": "FloatProgressModel",
            "bar_style": "danger",
            "max": 25250,
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": 25229,
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "orientation": "horizontal",
            "min": 0,
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_3b2faeaad7424cf6aa517dd53315eeb5"
          }
        },
        "7a3d634f65b6404483a1b7a5801a68ae": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_3f59491d2ebc4fe79d1412e87eceee5e",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": " 25229/25250 [00:28&lt;00:00, 884.86 examples/s]",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_fe2da2b2748f46e5a800eb0b3f64c64f"
          }
        },
        "2687a81f54fc4655be5bd42cb4c3254a": {
          "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"
          }
        },
        "3b2faeaad7424cf6aa517dd53315eeb5": {
          "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
          }
        },
        "3f59491d2ebc4fe79d1412e87eceee5e": {
          "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"
          }
        },
        "fe2da2b2748f46e5a800eb0b3f64c64f": {
          "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/prachuryanath/TF-Learning/blob/main/07_food_vision_milestone_project_1.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "VD27Vo8tK1Fb"
      },
      "source": [
        "# **Milestone Project 1: Food Vision Big**\n",
        "\n",
        "What we're going to cover :\n",
        "* Using TensorFlow Datasets to download and explore data\n",
        "* Creating preprocessing function for our data\n",
        "* Batching & preparing datasets for modelling (making our datasets run fast)\n",
        "* Creating modelling callbacks\n",
        "* Setting up mixed precision training\n",
        "* Building a feature extraction model (see transfer learning part 1: feature extraction)\n",
        "* Fine-tuning the feature extraction model (see transfer learning part 2: fine-tuning)\n",
        "* Viewing training results on TensorBoard"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "skj4fEJrLSeB"
      },
      "source": [
        "## **Check GPU**\n",
        "We're going to be using mixed precision training.\n",
        "\n",
        "Google Colab offers P100, K80 and T4 GPUs, however, the P100 and K80 aren't compatible with mixed precision training.\n",
        "\n",
        "Therefore before we proceed we need to make sure we have access to a Tesla T4 GPU in our Google Colab instance.\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "LlNw1tg9JkHX",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "070a4d2f-bd25-44ac-e8dc-ae78cc998a78"
      },
      "source": [
        "# If the following line doesn't output \"Tesla T4\", you can try getting access to\n",
        "# another GPU by going to Runtime -> Factory Reset Runtime -> \"Yes\" and then\n",
        "# re-running this cell.\n",
        "!nvidia-smi -L"
      ],
      "execution_count": 42,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "GPU 0: Tesla T4 (UUID: GPU-8d095188-b983-a415-6d60-00eccd323001)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ezuJyWVGKBMR",
        "outputId": "03ff5c82-970d-46b4-ef17-a8ee6d85e91f"
      },
      "source": [
        "# Check TensorFlow version (should be 2.4.0+)\n",
        "import tensorflow as tf\n",
        "print(tf.__version__)"
      ],
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "2.4.1\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0uVuF1bGLqKW"
      },
      "source": [
        "## **Get helper functions**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "GqFO-_bbL0N-",
        "outputId": "76cb5e4d-2cf0-45b2-b38c-35e0e54bdca8"
      },
      "source": [
        "# Get helper functions file\n",
        "!wget https://raw.githubusercontent.com/prachuryanath/TF-Learning/main/extras/helper_functions.py"
      ],
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "--2021-05-20 13:24:04--  https://raw.githubusercontent.com/prachuryanath/TF-Learning/main/extras/helper_functions.py\n",
            "Resolving raw.githubusercontent.com (raw.githubusercontent.com)... 185.199.108.133, 185.199.109.133, 185.199.110.133, ...\n",
            "Connecting to raw.githubusercontent.com (raw.githubusercontent.com)|185.199.108.133|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 10235 (10.0K) [text/plain]\n",
            "Saving to: ‘helper_functions.py’\n",
            "\n",
            "helper_functions.py 100%[===================>]  10.00K  --.-KB/s    in 0s      \n",
            "\n",
            "2021-05-20 13:24:04 (85.3 MB/s) - ‘helper_functions.py’ saved [10235/10235]\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "AyteLaNCL1zx"
      },
      "source": [
        "# Import series of helper functions for the notebook (we've created/used these in previous notebooks)\n",
        "from helper_functions import create_tensorboard_callback, plot_loss_curves, compare_historys"
      ],
      "execution_count": 5,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "tcJPVyQnMQPg"
      },
      "source": [
        "## **Use TensorFlow Datasets to Download Data**\n",
        "\n",
        "For many of the most popular datasets in the machine learning world (often referred to and used as benchmarks), you can access them through TensorFlow Datasets (TFDS)."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "WOu_PoOdM3ZM"
      },
      "source": [
        "# Get TensorFlow Datasets\n",
        "import tensorflow_datasets as tfds"
      ],
      "execution_count": 6,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "KgjfiAlOM7OO",
        "outputId": "eb7d5c95-8d54-41a5-d295-0b4d400b8dba"
      },
      "source": [
        "# List available datasets\n",
        "datasets_list = tfds.list_builders() # get all available datasets in TFDS\n",
        "print(\"food101\" in datasets_list) # is the dataset we're after available?"
      ],
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "True\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "p9f1x2oCQW1_"
      },
      "source": [
        "In particular, we'll have to pass it a few parameters to let it know what we're after:\n",
        "\n",
        "* `name (str)` : the target dataset (e.g. \"food101\")\n",
        "* `split (list, optional)` : what splits of the dataset we're after (e.g. [\"train\", \"validation\"])\n",
        "  * the split parameter is quite tricky. See the documentation for more.\n",
        "* `shuffle_files (bool)` : whether or not to shuffle the files on download, defaults to False\n",
        "* `as_supervised (bool)` : True to download data samples in tuple format ((data, label)) or False for dictionary format\n",
        "* `with_info (bool)` : True to download dataset metadata (labels, number of samples, etc)"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 355,
          "referenced_widgets": [
            "40602fc20b4d4605853ed0faf97e2e0e",
            "0158f9dd514141a7b1ddab842f332a18",
            "14d74cd26c5341bf815b523b908aaf92",
            "d1a36386fdd24ad492d05ad866164a64",
            "03c6156d6998490aae887f2d92f4963d",
            "a225930f0be14b87a45194a51791817b",
            "9bdb4563943d42e0bb55dd4f16618701",
            "c69c211937f44be6a9d78f94f126ae7d",
            "cdf16990d08947f4ace37859e6a96af0",
            "fa9236bfd3e14e9cbb321b003b608819",
            "950371e34a4e4d76b1747d09c56c75ea",
            "92e7697f4c6a42dfb6d3b37d32f86ab9",
            "f9b31a3a01fa423aaaab1e9067917f25",
            "8f910e6f3f194982bbe52e85e4a02ff1",
            "c17f5f4e9f674facb06e40b5441f7e36",
            "b9aede1d1d304f9f9665ea4db7acd258",
            "b86540e3b06f4e80a47014985ce629d1",
            "eff9872c3de94d728b082c747d8e848c",
            "b512b87c671a4a7fb18fef1229604266",
            "d3364328b5504e23ae48dc105fd1a539",
            "3804c882a3ae42019ca333bcbaf58ef5",
            "2c09c404ae304bbea7a621948bd0564e",
            "aeedcdfc18b348e7a0a60ffe24d2018b",
            "6aa735755bd04e3ca1bed195a82fdd85",
            "b5fc0fe99bf945cd8b34d6306a92d1bf",
            "f75325f868504b6cb68dd17f740201e3",
            "15dd8ed1d3fb4092b0b368bcd407da7d",
            "3d7f4186421f4d739e07fb60d4db32fa",
            "70893e8dc2624c0eb3239e573f630814",
            "66a2e5a432564d8399a857c6ddb92745",
            "947c5f07f095414cb58e5e24f394335c",
            "236cb5b12125417b8b3d838054131e1e",
            "8be1726c9f7f48db93a64f1733af1a81",
            "1b4e3abcb390482eb9c734c92585f461",
            "5cdef67f62e74f50b2b7fd66e026e3fa",
            "14e4b4f7bd2b45a2b3382d834d97a25c",
            "9966f9a263d742d3a55a0c91b1187cc8",
            "75171aeca5504699bdc651151955ffff",
            "d9bbe81e95b74333bbdc097b1f1ada9c",
            "c71afcd76afc4796a03b1e7cf874ce44",
            "abe518c58101448b94fdf73282489bb7",
            "bf7ae08a885f4f44b7a4f927b92b0230",
            "c8d4c232ead44544ab4bfe783cf3cb21",
            "0c49f2c3e9f348dfa7b8b1e0d9686885",
            "87d247d21ca5433eb9836f981fb3da71",
            "78891148fbbf455cb1c19dcdc4e75f08",
            "983f45c6456d46fb8620d8a4156fccf6",
            "9eeb036a10ab42a3ab9e7947380fac76",
            "c41f6dbbd8844335a0d066c623e23e29",
            "399774dfa9984e7ca51f5d95c963d655",
            "2022003efbc44e09a0f8c785181424d5",
            "7a3d634f65b6404483a1b7a5801a68ae",
            "2687a81f54fc4655be5bd42cb4c3254a",
            "3b2faeaad7424cf6aa517dd53315eeb5",
            "3f59491d2ebc4fe79d1412e87eceee5e",
            "fe2da2b2748f46e5a800eb0b3f64c64f"
          ]
        },
        "id": "pL7I8L9dM_eN",
        "outputId": "91b16dca-ac5a-4f81-c644-be758ba135cd"
      },
      "source": [
        "# Load in the data (takes 5-6 minutes in Google Colab)\n",
        "(train_data,test_data), ds_info = tfds.load(name=\"food101\", \n",
        "                                            split=[\"train\",\"validation\"],\n",
        "                                            shuffle_files=True,\n",
        "                                            as_supervised=True, # data gets returned in tuple format(data, label),\n",
        "                                            with_info=True)"
      ],
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\u001b[1mDownloading and preparing dataset food101/2.0.0 (download: 4.65 GiB, generated: Unknown size, total: 4.65 GiB) to /root/tensorflow_datasets/food101/2.0.0...\u001b[0m\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "40602fc20b4d4605853ed0faf97e2e0e",
              "version_minor": 0,
              "version_major": 2
            },
            "text/plain": [
              "HBox(children=(FloatProgress(value=1.0, bar_style='info', description='Dl Completed...', max=1.0, style=Progre…"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "cdf16990d08947f4ace37859e6a96af0",
              "version_minor": 0,
              "version_major": 2
            },
            "text/plain": [
              "HBox(children=(FloatProgress(value=1.0, bar_style='info', description='Dl Size...', max=1.0, style=ProgressSty…"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "b86540e3b06f4e80a47014985ce629d1",
              "version_minor": 0,
              "version_major": 2
            },
            "text/plain": [
              "HBox(children=(FloatProgress(value=1.0, bar_style='info', description='Extraction completed...', max=1.0, styl…"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "b5fc0fe99bf945cd8b34d6306a92d1bf",
              "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": [
            "\rShuffling and writing examples to /root/tensorflow_datasets/food101/2.0.0.incompleteXCOKPC/food101-train.tfrecord\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "8be1726c9f7f48db93a64f1733af1a81",
              "version_minor": 0,
              "version_major": 2
            },
            "text/plain": [
              "HBox(children=(FloatProgress(value=0.0, max=75750.0), HTML(value='')))"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "abe518c58101448b94fdf73282489bb7",
              "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": [
            "Shuffling and writing examples to /root/tensorflow_datasets/food101/2.0.0.incompleteXCOKPC/food101-validation.tfrecord\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "c41f6dbbd8844335a0d066c623e23e29",
              "version_minor": 0,
              "version_major": 2
            },
            "text/plain": [
              "HBox(children=(FloatProgress(value=0.0, max=25250.0), HTML(value='')))"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "\u001b[1mDataset food101 downloaded and prepared to /root/tensorflow_datasets/food101/2.0.0. Subsequent calls will reuse this data.\u001b[0m\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "-D6SOifdNApm",
        "outputId": "82da7e3c-377f-4d05-e8a9-6243f2e9a613"
      },
      "source": [
        "# Features of Food101 TFDS\n",
        "ds_info.features"
      ],
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "FeaturesDict({\n",
              "    'image': Image(shape=(None, None, 3), dtype=tf.uint8),\n",
              "    'label': ClassLabel(shape=(), dtype=tf.int64, num_classes=101),\n",
              "})"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 9
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "1cvf2Z_tR8ao",
        "outputId": "73a7bd6e-f7ee-4c21-b0ab-9077e6a90360"
      },
      "source": [
        "# Get class names\n",
        "class_names = ds_info.features[\"label\"].names\n",
        "class_names[:10]"
      ],
      "execution_count": 10,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "['apple_pie',\n",
              " 'baby_back_ribs',\n",
              " 'baklava',\n",
              " 'beef_carpaccio',\n",
              " 'beef_tartare',\n",
              " 'beet_salad',\n",
              " 'beignets',\n",
              " 'bibimbap',\n",
              " 'bread_pudding',\n",
              " 'breakfast_burrito']"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 10
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "YjeZLXWsR_XU"
      },
      "source": [
        "### Exploring the Food101 data from TensorFlow Datasets\n",
        "Let's find out a few details about our dataset:\n",
        "\n",
        "* The shape of our input data (image tensors)\n",
        "* The datatype of our input data\n",
        "* What the labels of our input data look like (e.g. one-hot encoded versus label-encoded)\n",
        "* Do the labels match up with the class names?"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "usllEaEgSenP"
      },
      "source": [
        "# Take one sample off the training data\n",
        "train_one_sample = train_data.take(1) # samples are in format (image_tensor, label)"
      ],
      "execution_count": 11,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "K-VpvBg4SlEX"
      },
      "source": [
        "Because we used the as_supervised=True parameter in our tfds.load() method above, data samples come in the tuple format structure (data, label) or in our case (image_tensor, label)."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "M2Q2JBF-SnUy",
        "outputId": "b1f38b22-d2f6-4451-ed01-f9dfeeaab1fc"
      },
      "source": [
        "# What does one sample of our training data look like?\n",
        "train_one_sample"
      ],
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<TakeDataset shapes: ((None, None, 3), ()), types: (tf.uint8, tf.int64)>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 12
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "YGFq9vnISoy1",
        "outputId": "ce4d11a5-486e-4570-8fa6-e83df800ef91"
      },
      "source": [
        "# Output info about our training sample\n",
        "for image, label in train_one_sample:\n",
        "  print(f\"\"\"\n",
        "  Image shape: {image.shape}\n",
        "  Image dtype: {image.dtype}\n",
        "  Target class from Food101 (tensor form): {label}\n",
        "  Class name (str form): {class_names[label.numpy()]}\n",
        "        \"\"\")"
      ],
      "execution_count": 13,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\n",
            "  Image shape: (512, 512, 3)\n",
            "  Image dtype: <dtype: 'uint8'>\n",
            "  Target class from Food101 (tensor form): 16\n",
            "  Class name (str form): cheesecake\n",
            "        \n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "OvQgCW4RSsct"
      },
      "source": [
        "Because we set the shuffle_files=True parameter in our tfds.load() method above, running the cell above a few times will give a different result each time.\n",
        "\n",
        "Checking these you might notice some of the images have different shapes, for example (512, 342, 3) and (512, 512, 3) (height, width, color_channels).\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "oPa-DS3KS7Ih",
        "outputId": "e3f5d91c-e568-4946-920a-da0a94a3be6f"
      },
      "source": [
        "# What does an image tensor from TFS's Food101 look like ?\n",
        "image"
      ],
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tf.Tensor: shape=(512, 512, 3), dtype=uint8, numpy=\n",
              "array([[[ 85,  84,  92],\n",
              "        [ 86,  85,  93],\n",
              "        [ 93,  92, 100],\n",
              "        ...,\n",
              "        [120, 196, 245],\n",
              "        [120, 198, 246],\n",
              "        [121, 199, 247]],\n",
              "\n",
              "       [[ 88,  88,  96],\n",
              "        [ 82,  82,  90],\n",
              "        [ 88,  88,  96],\n",
              "        ...,\n",
              "        [116, 192, 244],\n",
              "        [118, 195, 249],\n",
              "        [119, 196, 248]],\n",
              "\n",
              "       [[ 77,  80,  89],\n",
              "        [ 75,  78,  87],\n",
              "        [ 82,  86,  95],\n",
              "        ...,\n",
              "        [115, 189, 252],\n",
              "        [117, 192, 255],\n",
              "        [118, 194, 255]],\n",
              "\n",
              "       ...,\n",
              "\n",
              "       [[159, 166, 194],\n",
              "        [154, 161, 187],\n",
              "        [171, 178, 204],\n",
              "        ...,\n",
              "        [219, 216, 235],\n",
              "        [223, 220, 239],\n",
              "        [227, 226, 244]],\n",
              "\n",
              "       [[170, 179, 208],\n",
              "        [169, 179, 206],\n",
              "        [173, 183, 210],\n",
              "        ...,\n",
              "        [218, 217, 235],\n",
              "        [223, 222, 240],\n",
              "        [226, 227, 245]],\n",
              "\n",
              "       [[168, 177, 206],\n",
              "        [170, 179, 208],\n",
              "        [152, 162, 189],\n",
              "        ...,\n",
              "        [224, 223, 241],\n",
              "        [228, 229, 247],\n",
              "        [231, 232, 250]]], dtype=uint8)>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 14
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Lwj2u80CTFqK",
        "outputId": "c7794076-a48a-4c3b-87f3-2584ec1a1048"
      },
      "source": [
        "# What are the min and max values?\n",
        "tf.reduce_min(image), tf.reduce_max(image)"
      ],
      "execution_count": 15,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(<tf.Tensor: shape=(), dtype=uint8, numpy=0>,\n",
              " <tf.Tensor: shape=(), dtype=uint8, numpy=255>)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 15
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "abkqNyLtTNeg"
      },
      "source": [
        "### Plot an image from TensorFlow Datasets"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        },
        "id": "BVG3Tb-CTbiD",
        "outputId": "376dd974-40d3-4a4b-d7bb-b32a990e6d3b"
      },
      "source": [
        "# Plot an image tensor\n",
        "import matplotlib.pyplot as plt\n",
        "plt.imshow(image)\n",
        "plt.title(class_names[label.numpy()]) # add title to image by indexing on class_names list\n",
        "plt.axis(False);"
      ],
      "execution_count": 16,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "UW-2nAIYTdXp"
      },
      "source": [
        "## **Create preprocessing functions for our data**\n",
        "What we know about our data:\n",
        "* In uint8 datatype\n",
        "* Comprised of all different size tensors (different sized images)\n",
        "* Not scaled (the pixel values are between 0 & 255)\n",
        "\n",
        "What we know models like:\n",
        "* Data in float32 dtype (or for mixed precision float16 and float32)\n",
        "* For batches, TensorFlow likes all of the tensors within a batch to be of the same size\n",
        "* Scaled (values between 0 & 1) also called normalized tensors generally perform better\n",
        "\n",
        "This means our functions needs to:\n",
        "\n",
        "* Reshape our images to all the same size\n",
        "* Convert the dtype of our image tensors from unit8 to float32"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "iYwMt0UWWRW8"
      },
      "source": [
        "# Make a function for preprocessing images\n",
        "def preprocess_img(image, label, img_shape=224):\n",
        "  \"\"\"\n",
        "  Converts image datatype from 'uint8' -> 'float32' and reshapes\n",
        "  image to [img_shape, img_shape, colour_channels]\n",
        "  \"\"\"\n",
        "  image = tf.image.resize(image, [img_shape, img_shape]) # reshape target image\n",
        "  # image = image/255. # scale image values (not required with EfficientNetBX models from tf.keras.applications)\n",
        "  return tf.cast(image, tf.float32), label # return (float32_image, label) tuple"
      ],
      "execution_count": 17,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "vMxnAEmUWtDH",
        "outputId": "c7f54ed6-308d-4d7e-d1d9-89a1f9819f67"
      },
      "source": [
        "# Preprocess a single sample image and check the outputs\n",
        "preprocessed_img = preprocess_img(image, label)[0]\n",
        "print(f\"Image before preprocessing:\\n {image[:2]}..., \\nShape: {image.shape},\\nDatatype: {image.dtype}\\n\")\n",
        "print(f\"Image after preprocessing:\\n{preprocessed_img[:2]}...,\\nShape: {preprocessed_img.shape},\\nDatatype: {preprocessed_img.dtype}\")"
      ],
      "execution_count": 18,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Image before preprocessing:\n",
            " [[[ 85  84  92]\n",
            "  [ 86  85  93]\n",
            "  [ 93  92 100]\n",
            "  ...\n",
            "  [120 196 245]\n",
            "  [120 198 246]\n",
            "  [121 199 247]]\n",
            "\n",
            " [[ 88  88  96]\n",
            "  [ 82  82  90]\n",
            "  [ 88  88  96]\n",
            "  ...\n",
            "  [116 192 244]\n",
            "  [118 195 249]\n",
            "  [119 196 248]]]..., \n",
            "Shape: (512, 512, 3),\n",
            "Datatype: <dtype: 'uint8'>\n",
            "\n",
            "Image after preprocessing:\n",
            "[[[ 84.67857   84.321434  92.321434]\n",
            "  [ 87.19898   86.841835  94.841835]\n",
            "  [ 84.72959   84.581635  94.72449 ]\n",
            "  ...\n",
            "  [119.50503  194.93877  245.85202 ]\n",
            "  [114.974525 190.97452  241.90309 ]\n",
            "  [119.071465 196.4286   247.82652 ]]\n",
            "\n",
            " [[ 75.647964  84.21939   93.21939 ]\n",
            "  [ 71.45407   80.959175  89.959175]\n",
            "  [ 78.03061   87.20408   97.47449 ]\n",
            "  ...\n",
            "  [113.31632  182.62755  249.35715 ]\n",
            "  [106.57655  176.86227  245.4337  ]\n",
            "  [108.601944 179.9796   248.16833 ]]]...,\n",
            "Shape: (224, 224, 3),\n",
            "Datatype: <dtype: 'float32'>\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "reFgfxYQWzfi"
      },
      "source": [
        "## **Batch & prepare datasets**\n",
        "Before we can model our data, we have to turn it into batches as computing on batches is memory efficient.\n",
        "\n",
        "We turn our data from 101,000 image tensors and labels (train and test combined) into batches of 32 image and label pairs, thus enabling it to fit into the memory of our GPU.\n",
        "\n",
        "Specifically, we're going to be using:\n",
        "\n",
        "* `map()` - maps a predefined function to a target dataset (e.g. preprocess_img() to our image tensors)\n",
        "* `shuffle()` - randomly shuffles the elements of a target dataset up buffer_size (ideally, the buffer_size is equal to the size of the dataset, however, this may have implications on memory)\n",
        "* `batch()` - turns elements of a target dataset into batches (size defined by parameter batch_size)\n",
        "* `prefetch()` - prepares subsequent batches of data whilst other batches of data are being computed on (improves data loading speed but costs memory)\n",
        "* Extra: `cache()` - caches (saves them for later) elements in a target dataset, saving loading time (will only work if your dataset is small enough to fit in memory, standard Colab instances only have 12GB of memory)\n",
        "---\n",
        "Things to note:\n",
        "* Can't batch tensors of different shapes (e.g. different image sizes, need to reshape images first, hence our `preprocess_img()` function)\n",
        "* `shuffle()` keeps a buffer of the number you pass it images shuffled, ideally this number would be all of the samples in your training set, however, if your training set is large, this buffer might not fit in memory (a fairly large number like 1000 or 10000 is usually suffice for shuffling)\n",
        "* For methods with the num_parallel_calls parameter available (such as `map()`), setting it to `num_parallel_calls=tf.data.AUTOTUNE` will parallelize preprocessing and significantly improve speed\n",
        "* Can't use `cache()` unless your dataset can \n",
        "fit in memory\n",
        "\n",
        "Woah, the above is alot. But once we've coded below, it'll start to make sense.\n",
        "\n",
        "We're going to through things in the following order:\n",
        "\n",
        "Original dataset (e.g. train_data) -> map() -> shuffle() -> batch() -> prefetch() -> PrefetchDataset\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Zq-UzQCRXV-1"
      },
      "source": [
        "# Map preprocessing function to training data (and parallelize)\n",
        "train_data = train_data.map(map_func=preprocess_img, num_parallel_calls=tf.data.AUTOTUNE)\n",
        "\n",
        "# Shuffle train_data and turn it into batches and prefetch it (load it faster)\n",
        "train_data = train_data.shuffle(buffer_size=1000).batch(batch_size=32).prefetch(buffer_size=tf.data.AUTOTUNE)\n",
        "\n",
        "# Map preprocessing function to test data\n",
        "test_data = test_data.map(preprocess_img, num_parallel_calls=tf.data.AUTOTUNE)\n",
        "\n",
        "# Turn test data into batches (don't need to shuffle)\n",
        "test_data = test_data.batch(32).prefetch(tf.data.AUTOTUNE)"
      ],
      "execution_count": 19,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "DAq2GFRRa8Jf",
        "outputId": "5ac3aa1a-46f0-4579-f389-6abfa034023d"
      },
      "source": [
        "train_data, test_data"
      ],
      "execution_count": 20,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(<PrefetchDataset shapes: ((None, 224, 224, 3), (None,)), types: (tf.float32, tf.int64)>,\n",
              " <PrefetchDataset shapes: ((None, 224, 224, 3), (None,)), types: (tf.float32, tf.int64)>)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 20
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "vtgDiXpja-ot"
      },
      "source": [
        "### Create modelling callbacks\n",
        "* `tf.keras.callbacks.TensorBoard()` - allows us to keep track of our model's training history so we can inspect it later (note: we've created this callback before have imported it from `helper_functions.py` as `create_tensorboard_callback()`)\n",
        "* `tf.keras.callbacks.ModelCheckpoint()` - saves our model's progress at various intervals so we can load it and resuse it later without having to retrain it\n",
        "  * Checkpointing is also helpful so we can start fine-tuning our model at a particular epoch and revert back to a previous state if fine-tuning offers no benefits"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "aenZWn8kcfAQ"
      },
      "source": [
        "# Create TensorBoard callback (already have \"create_tensorboard_callback()\" from a previous notebook)\n",
        "from helper_functions import create_tensorboard_callback\n",
        "\n",
        "# Create ModelCheckpoint callback to save model's progress\n",
        "checkpoint_path = \"model_checkpoints/cp.ckpt\" # saving weights requires \".ckpt\" extension\n",
        "model_checkpoint = tf.keras.callbacks.ModelCheckpoint(checkpoint_path,\n",
        "                                                      montior=\"val_acc\", # save the model weights with best validation accuracy\n",
        "                                                      save_best_only=True, # only save the best weights\n",
        "                                                      save_weights_only=True, # only save model weights (not whole model)\n",
        "                                                      verbose=0) # don't print out whether or not model is being saved"
      ],
      "execution_count": 21,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "bM34xgP3dHuc"
      },
      "source": [
        "### Setup mixed precision training\n",
        "\n",
        "When using mixed precision training, your model will make use of float32 and float16 data types to use less memory where possible and in turn run faster (using less memory per tensor means more tensors can be computed on simultaneously).\n",
        "\n",
        "As a result, using mixed precision training can improve your performance on modern GPUs (those with a compute capability score of 7.0+) by up to 3x.\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "oakbylpIghE2",
        "outputId": "15d912a2-7083-4bb2-b2ce-3e94358cecd9"
      },
      "source": [
        "# Turn on mixed precision training\n",
        "from tensorflow.keras import mixed_precision\n",
        "mixed_precision.set_global_policy(policy=\"mixed_float16\") # set global policy to mixed precision"
      ],
      "execution_count": 22,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:Mixed precision compatibility check (mixed_float16): OK\n",
            "Your GPU will likely run quickly with dtype policy mixed_float16 as it has compute capability of at least 7.0. Your GPU: Tesla T4, compute capability 7.5\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:Mixed precision compatibility check (mixed_float16): OK\n",
            "Your GPU will likely run quickly with dtype policy mixed_float16 as it has compute capability of at least 7.0. Your GPU: Tesla T4, compute capability 7.5\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "kGNjrPMvgqNx",
        "outputId": "25090f9f-54ee-4f23-8364-3b2875aa18cc"
      },
      "source": [
        "mixed_precision.global_policy() # should output \"mixed_float16\""
      ],
      "execution_count": 23,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<Policy \"mixed_float16\">"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 23
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "6JVO9X_kguwv"
      },
      "source": [
        "## **Build feature extraction model**\n",
        "\n",
        "To build the feature extraction model (covered in Transfer Learning in TensorFlow Part 1: Feature extraction), we'll:\n",
        "* Use EfficientNetB0 from tf.keras.applications pre-trained on ImageNet as our base model\n",
        "* We'll download this without the top layers using include_top=False parameter so we can create our own output layers\n",
        "* Freeze the base model layers so we can use the pre-learned patterns the base model has found on ImageNet\n",
        "* Put together the input, base model, pooling and output layers in a Functional model\n",
        "* Compile the Functional model using the Adam optimizer and sparse categorical crossentropy as the loss function (since our labels aren't one-hot encoded)\n",
        "* Fit the model for 3 epochs using the TensorBoard and ModelCheckpoint callbacks\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "NVXC9S5Zgywa",
        "outputId": "8ccc8f38-e4f2-4b7e-8776-74faae80c6f6"
      },
      "source": [
        "from tensorflow.keras import layers\n",
        "from tensorflow.keras.layers.experimental import preprocessing\n",
        "\n",
        "# Create base model\n",
        "input_shape = (224, 224, 3)\n",
        "base_model = tf.keras.applications.EfficientNetB0(include_top=False)\n",
        "base_model.trainable = False # freeze base model layers\n",
        "\n",
        "# Create Functional model \n",
        "inputs = layers.Input(shape=input_shape, name=\"input_layer\")\n",
        "# Note: EfficientNetBX models have rescaling built-in but if your model didn't you could have a layer like below\n",
        "# x = preprocessing.Rescaling(1./255)(x)\n",
        "x = base_model(inputs, training=False) # set base_model to inference mode only\n",
        "x = layers.GlobalAveragePooling2D(name=\"pooling_layer\")(x)\n",
        "x = layers.Dense(len(class_names))(x) # want one output neuron per class \n",
        "# Separate activation of output layer so we can output float32 activations\n",
        "outputs = layers.Activation(\"softmax\", dtype=tf.float32, name=\"softmax_float32\")(x) \n",
        "model = tf.keras.Model(inputs, outputs)\n",
        "\n",
        "# Compile the model\n",
        "model.compile(loss=\"sparse_categorical_crossentropy\", # Use sparse_categorical_crossentropy when labels are *not* one-hot\n",
        "              optimizer=tf.keras.optimizers.Adam(),\n",
        "              metrics=[\"accuracy\"])"
      ],
      "execution_count": 24,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Downloading data from https://storage.googleapis.com/keras-applications/efficientnetb0_notop.h5\n",
            "16711680/16705208 [==============================] - 0s 0us/step\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "KiZyB7iehjq5",
        "outputId": "0825de0f-d4ad-4bee-f11a-e30b33940636"
      },
      "source": [
        "# Check out our model\n",
        "model.summary()"
      ],
      "execution_count": 25,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"model\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "input_layer (InputLayer)     [(None, 224, 224, 3)]     0         \n",
            "_________________________________________________________________\n",
            "efficientnetb0 (Functional)  (None, None, None, 1280)  4049571   \n",
            "_________________________________________________________________\n",
            "pooling_layer (GlobalAverage (None, 1280)              0         \n",
            "_________________________________________________________________\n",
            "dense (Dense)                (None, 101)               129381    \n",
            "_________________________________________________________________\n",
            "softmax_float32 (Activation) (None, 101)               0         \n",
            "=================================================================\n",
            "Total params: 4,178,952\n",
            "Trainable params: 129,381\n",
            "Non-trainable params: 4,049,571\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "STMO4jHiiAZ3"
      },
      "source": [
        "### Checking layer dtype policies (are we using mixed precision ?)"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Lt_hh5gghnfH",
        "outputId": "d97e98e7-2cb5-43d2-afb3-9cae6e22c48a"
      },
      "source": [
        "# Check the dtype_policy attributes of layers in our model\n",
        "for layer in model.layers:\n",
        "  print(layer.name, layer.trainable, layer.dtype, layer.dtype_policy) # Check the dtype policy of layers"
      ],
      "execution_count": 26,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "input_layer True float32 <Policy \"float32\">\n",
            "efficientnetb0 False float32 <Policy \"mixed_float16\">\n",
            "pooling_layer True float32 <Policy \"mixed_float16\">\n",
            "dense True float32 <Policy \"mixed_float16\">\n",
            "softmax_float32 True float32 <Policy \"float32\">\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "EKrYGZmEiLWd"
      },
      "source": [
        "Going through the above we see:\n",
        "* `layer.name` (str) : a layer's human-readable name, can be defined by the name parameter on construction\n",
        "* `layer.trainable` (bool) : whether or not a layer is trainable (all of our layers are trainable except the efficientnetb0 layer since we set it's trainable attribute to False\n",
        "* `layer.dtype` : the data type a layer stores its variables in\n",
        "* layer.dtype_policy : the data type a layer computes in"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "5n9wqRS6hr-E",
        "outputId": "c2fc2ae5-6669-4e85-c362-628a65a002a7"
      },
      "source": [
        "# Check the layers in the base model and see what dtype policy they're using\n",
        "for layer in model.layers[1].layers[:20]: # only check the first 20 layers to save output space\n",
        "  print(layer.name, layer.trainable, layer.dtype, layer.dtype_policy)"
      ],
      "execution_count": 27,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "input_1 False float32 <Policy \"float32\">\n",
            "rescaling False float32 <Policy \"mixed_float16\">\n",
            "normalization False float32 <Policy \"float32\">\n",
            "stem_conv_pad False float32 <Policy \"mixed_float16\">\n",
            "stem_conv False float32 <Policy \"mixed_float16\">\n",
            "stem_bn False float32 <Policy \"mixed_float16\">\n",
            "stem_activation False float32 <Policy \"mixed_float16\">\n",
            "block1a_dwconv False float32 <Policy \"mixed_float16\">\n",
            "block1a_bn False float32 <Policy \"mixed_float16\">\n",
            "block1a_activation False float32 <Policy \"mixed_float16\">\n",
            "block1a_se_squeeze False float32 <Policy \"mixed_float16\">\n",
            "block1a_se_reshape False float32 <Policy \"mixed_float16\">\n",
            "block1a_se_reduce False float32 <Policy \"mixed_float16\">\n",
            "block1a_se_expand False float32 <Policy \"mixed_float16\">\n",
            "block1a_se_excite False float32 <Policy \"mixed_float16\">\n",
            "block1a_project_conv False float32 <Policy \"mixed_float16\">\n",
            "block1a_project_bn False float32 <Policy \"mixed_float16\">\n",
            "block2a_expand_conv False float32 <Policy \"mixed_float16\">\n",
            "block2a_expand_bn False float32 <Policy \"mixed_float16\">\n",
            "block2a_expand_activation False float32 <Policy \"mixed_float16\">\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "etw5nepGh0wi"
      },
      "source": [
        "## **Fit the feature extraction model**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "TPdsrYcJhxE1",
        "outputId": "ce2edde5-1c71-42ac-928c-86c44684c7ee"
      },
      "source": [
        "# Fit the model with callbacks\n",
        "history_101_food_classes_feature_extract = model.fit(train_data, \n",
        "                                                     epochs=3,\n",
        "                                                     steps_per_epoch=len(train_data),\n",
        "                                                     validation_data=test_data,\n",
        "                                                     validation_steps=int(0.15 * len(test_data)),\n",
        "                                                     callbacks=[create_tensorboard_callback(\"training_logs\", \n",
        "                                                                                            \"efficientnetb0_101_classes_all_data_feature_extract\"),\n",
        "                                                                model_checkpoint])"
      ],
      "execution_count": 28,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Saving TensorBoard log files to: training_logs/efficientnetb0_101_classes_all_data_feature_extract/20210520-133328\n",
            "Epoch 1/3\n",
            "2368/2368 [==============================] - ETA: 0s - loss: 2.3171 - accuracy: 0.4675Epoch 2/3\n",
            "2368/2368 [==============================] - 159s 66ms/step - loss: 1.3139 - accuracy: 0.6646 - val_loss: 1.1243 - val_accuracy: 0.7021\n",
            "Epoch 3/3\n",
            "2368/2368 [==============================] - 157s 66ms/step - loss: 1.1480 - accuracy: 0.7019 - val_loss: 1.0819 - val_accuracy: 0.7119\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "aedp2-gOh-tj",
        "outputId": "1de53289-9a8e-4e0f-ca55-9cad30a08249"
      },
      "source": [
        "# Evaluate model (unsaved version) on whole test dataset\n",
        "results_feature_extract_model = model.evaluate(test_data)\n",
        "results_feature_extract_model"
      ],
      "execution_count": 29,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "790/790 [==============================] - 54s 68ms/step - loss: 1.0862 - accuracy: 0.7088\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[1.0861735343933105, 0.708752453327179]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 29
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "P93-GogEig5B"
      },
      "source": [
        "### Load and evaluate checkpoint weights\n",
        "We can load in and evaluate our model's checkpoints by:\n",
        "\n",
        "1. Cloning our model using `tf.keras.models.clone_model()` to make a copy of our feature extraction model with reset weights.\n",
        "2. Calling the `load_weights()` method on our cloned model passing it the path to where our checkpointed weights are stored.\n",
        "3. Calling `evaluate()` on the cloned model with loaded weights."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "1cjr7CAyi-dh",
        "outputId": "e62faa94-85ec-4a03-a5cd-b51fff41d152"
      },
      "source": [
        "# Clone the model we created (this resets all weights)\n",
        "cloned_model = tf.keras.models.clone_model(model)\n",
        "cloned_model.summary()"
      ],
      "execution_count": 30,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"model\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "input_layer (InputLayer)     [(None, 224, 224, 3)]     0         \n",
            "_________________________________________________________________\n",
            "efficientnetb0 (Functional)  (None, None, None, 1280)  4049571   \n",
            "_________________________________________________________________\n",
            "pooling_layer (GlobalAverage (None, 1280)              0         \n",
            "_________________________________________________________________\n",
            "dense (Dense)                (None, 101)               129381    \n",
            "_________________________________________________________________\n",
            "softmax_float32 (Activation) (None, 101)               0         \n",
            "=================================================================\n",
            "Total params: 4,178,952\n",
            "Trainable params: 129,381\n",
            "Non-trainable params: 4,049,571\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 36
        },
        "id": "bzZDxk7xjEoo",
        "outputId": "53b57853-baba-42c1-eb00-792d7671756d"
      },
      "source": [
        "# Where are our checkpoints stored?\n",
        "checkpoint_path"
      ],
      "execution_count": 31,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "string"
            },
            "text/plain": [
              "'model_checkpoints/cp.ckpt'"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 31
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "l2g8JZFpjGlp",
        "outputId": "f2a4faf7-98a0-4b5b-90f0-489e43040633"
      },
      "source": [
        "# Load checkpointed weights into cloned_model\n",
        "cloned_model.load_weights(checkpoint_path)"
      ],
      "execution_count": 32,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tensorflow.python.training.tracking.util.CheckpointLoadStatus at 0x7f36c7582890>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 32
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "umBJj9I1jIZm"
      },
      "source": [
        "# Compile cloned_model (with same parameters as original model)\n",
        "cloned_model.compile(loss=\"sparse_categorical_crossentropy\",\n",
        "                     optimizer=tf.keras.optimizers.Adam(),\n",
        "                     metrics=[\"accuracy\"])"
      ],
      "execution_count": 33,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "-T9U76uojL6Y",
        "outputId": "b1ffae06-f8bc-4eb1-be24-54804c8e7cd9"
      },
      "source": [
        "# Evalaute cloned model with loaded weights (should be same score as trained model)\n",
        "results_cloned_model_with_loaded_weights = cloned_model.evaluate(test_data)"
      ],
      "execution_count": 34,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "790/790 [==============================] - 48s 57ms/step - loss: 1.0916 - accuracy: 0.7054\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BlVh2Lt2jOgB"
      },
      "source": [
        "# Loaded checkpoint weights should return very similar results to checkpoint weights prior to saving\n",
        "import numpy as np\n",
        "assert np.isclose(results_feature_extract_model, results_cloned_model_with_loaded_weights).all() # check if all elements in array are close"
      ],
      "execution_count": 35,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Pp3nojKcjQi_",
        "outputId": "a5742d0c-3e29-4cbd-fdbe-3e93010aa3e9"
      },
      "source": [
        "# Check the layers in the base model and see what dtype policy they're using\n",
        "for layer in cloned_model.layers[1].layers[:20]: # check only the first 20 layers to save space\n",
        "  print(layer.name, layer.trainable, layer.dtype, layer.dtype_policy)"
      ],
      "execution_count": 36,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "input_1 True float32 <Policy \"float32\">\n",
            "rescaling False float32 <Policy \"mixed_float16\">\n",
            "normalization False float32 <Policy \"float32\">\n",
            "stem_conv_pad False float32 <Policy \"mixed_float16\">\n",
            "stem_conv False float32 <Policy \"mixed_float16\">\n",
            "stem_bn False float32 <Policy \"mixed_float16\">\n",
            "stem_activation False float32 <Policy \"mixed_float16\">\n",
            "block1a_dwconv False float32 <Policy \"mixed_float16\">\n",
            "block1a_bn False float32 <Policy \"mixed_float16\">\n",
            "block1a_activation False float32 <Policy \"mixed_float16\">\n",
            "block1a_se_squeeze False float32 <Policy \"mixed_float16\">\n",
            "block1a_se_reshape False float32 <Policy \"mixed_float16\">\n",
            "block1a_se_reduce False float32 <Policy \"mixed_float16\">\n",
            "block1a_se_expand False float32 <Policy \"mixed_float16\">\n",
            "block1a_se_excite False float32 <Policy \"mixed_float16\">\n",
            "block1a_project_conv False float32 <Policy \"mixed_float16\">\n",
            "block1a_project_bn False float32 <Policy \"mixed_float16\">\n",
            "block2a_expand_conv False float32 <Policy \"mixed_float16\">\n",
            "block2a_expand_bn False float32 <Policy \"mixed_float16\">\n",
            "block2a_expand_activation False float32 <Policy \"mixed_float16\">\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "OOVxMSFBjTn1"
      },
      "source": [
        "## **Save the whole model to file**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "7CszNL6XjgcE",
        "outputId": "d878cb06-fad9-4537-feff-f74af1de0f74"
      },
      "source": [
        "## Saving model to Google Drive (optional)\n",
        "\n",
        "# Create save path to drive \n",
        "save_dir = \"drive/MyDrive/Colab Networks/food_vision/07_efficientnetb0_feature_extract_model_mixed_precision/\"\n",
        "# os.makedirs(save_dir) # Make directory if it doesn't exist\n",
        "\n",
        "# Save model\n",
        "model.save(save_dir)"
      ],
      "execution_count": 37,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:Assets written to: drive/MyDrive/Colab Networks/food_vision/07_efficientnetb0_feature_extract_model_mixed_precision/assets\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:Assets written to: drive/MyDrive/Colab Networks/food_vision/07_efficientnetb0_feature_extract_model_mixed_precision/assets\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "B0X887DNk_FW",
        "outputId": "d0dd0f58-cd6f-4cce-8978-5c50468d64f4"
      },
      "source": [
        "# Save model locally (if you're using Google Colab, your saved model will Colab instance terminates)\n",
        "save_dir_1 = \"07_efficientnetb0_feature_extract_model_mixed_precision\"\n",
        "model.save(save_dir_1)"
      ],
      "execution_count": 38,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:Assets written to: 07_efficientnetb0_feature_extract_model_mixed_precision/assets\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:Assets written to: 07_efficientnetb0_feature_extract_model_mixed_precision/assets\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ZGc15Rwgliqq"
      },
      "source": [
        "# Load model previously saved above\n",
        "loaded_saved_model = tf.keras.models.load_model(save_dir)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "EvR8vTwRl-XS"
      },
      "source": [
        "IGNORE ABOVE WARNINGS !!!"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "kJ-EUWB9lsYl",
        "outputId": "231fe26b-1075-4ddf-9bc4-02337a8144bd"
      },
      "source": [
        "# Check the layers in the base model and see what dtype policy they're using\n",
        "for layer in loaded_saved_model.layers[1].layers[:20]: # check only the first 20 layers to save output space\n",
        "  print(layer.name, layer.trainable, layer.dtype, layer.dtype_policy)"
      ],
      "execution_count": 40,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "input_1 True float32 <Policy \"float32\">\n",
            "rescaling True float32 <Policy \"mixed_float16\">\n",
            "normalization True float32 <Policy \"float32\">\n",
            "stem_conv_pad True float32 <Policy \"mixed_float16\">\n",
            "stem_conv True float32 <Policy \"mixed_float16\">\n",
            "stem_bn True float32 <Policy \"mixed_float16\">\n",
            "stem_activation True float32 <Policy \"mixed_float16\">\n",
            "block1a_dwconv True float32 <Policy \"mixed_float16\">\n",
            "block1a_bn True float32 <Policy \"mixed_float16\">\n",
            "block1a_activation True float32 <Policy \"mixed_float16\">\n",
            "block1a_se_squeeze True float32 <Policy \"mixed_float16\">\n",
            "block1a_se_reshape True float32 <Policy \"mixed_float16\">\n",
            "block1a_se_reduce True float32 <Policy \"mixed_float16\">\n",
            "block1a_se_expand True float32 <Policy \"mixed_float16\">\n",
            "block1a_se_excite True float32 <Policy \"mixed_float16\">\n",
            "block1a_project_conv True float32 <Policy \"mixed_float16\">\n",
            "block1a_project_bn True float32 <Policy \"mixed_float16\">\n",
            "block2a_expand_conv True float32 <Policy \"mixed_float16\">\n",
            "block2a_expand_bn True float32 <Policy \"mixed_float16\">\n",
            "block2a_expand_activation True float32 <Policy \"mixed_float16\">\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ESM6RYOil4sL",
        "outputId": "0427511c-2536-4049-88d6-23841e324ec0"
      },
      "source": [
        "# Check loaded model performance (this should be the same as results_feature_extract_model)\n",
        "results_loaded_saved_model = loaded_saved_model.evaluate(test_data)\n",
        "results_loaded_saved_model"
      ],
      "execution_count": 43,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "790/790 [==============================] - 46s 58ms/step - loss: 1.0862 - accuracy: 0.7088\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[1.0861740112304688, 0.708752453327179]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 43
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "1eemp6lnmCzP"
      },
      "source": [
        "## **Preparing our model's layers for fine-tuning**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "wpYs-m-mIHLw",
        "outputId": "f6150d01-31dc-42fa-bd59-631dffa73afe"
      },
      "source": [
        "# Download the saved model from Google Storage\n",
        "!wget https://storage.googleapis.com/ztm_tf_course/food_vision/07_efficientnetb0_feature_extract_model_mixed_precision.zip"
      ],
      "execution_count": 44,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "--2021-05-20 14:14:30--  https://storage.googleapis.com/ztm_tf_course/food_vision/07_efficientnetb0_feature_extract_model_mixed_precision.zip\n",
            "Resolving storage.googleapis.com (storage.googleapis.com)... 142.250.73.240, 142.250.65.80, 142.250.81.208, ...\n",
            "Connecting to storage.googleapis.com (storage.googleapis.com)|142.250.73.240|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 16976857 (16M) [application/zip]\n",
            "Saving to: ‘07_efficientnetb0_feature_extract_model_mixed_precision.zip’\n",
            "\n",
            "07_efficientnetb0_f 100%[===================>]  16.19M  --.-KB/s    in 0.1s    \n",
            "\n",
            "2021-05-20 14:14:31 (112 MB/s) - ‘07_efficientnetb0_feature_extract_model_mixed_precision.zip’ saved [16976857/16976857]\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "5Mv_J-a9ILcb",
        "outputId": "d10d4342-69ec-4ed4-876e-07c7de6eec9c"
      },
      "source": [
        "# Unzip the SavedModel downloaded from Google Stroage\n",
        "!mkdir downloaded_gs_model # create new dir to store downloaded feature extraction model\n",
        "!unzip 07_efficientnetb0_feature_extract_model_mixed_precision.zip -d downloaded_gs_model"
      ],
      "execution_count": 45,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Archive:  07_efficientnetb0_feature_extract_model_mixed_precision.zip\n",
            "   creating: downloaded_gs_model/07_efficientnetb0_feature_extract_model_mixed_precision/\n",
            "   creating: downloaded_gs_model/07_efficientnetb0_feature_extract_model_mixed_precision/variables/\n",
            "  inflating: downloaded_gs_model/07_efficientnetb0_feature_extract_model_mixed_precision/variables/variables.data-00000-of-00001  \n",
            "  inflating: downloaded_gs_model/07_efficientnetb0_feature_extract_model_mixed_precision/variables/variables.index  \n",
            "  inflating: downloaded_gs_model/07_efficientnetb0_feature_extract_model_mixed_precision/saved_model.pb  \n",
            "   creating: downloaded_gs_model/07_efficientnetb0_feature_extract_model_mixed_precision/assets/\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "rKJ313xtIUh1"
      },
      "source": [
        "# Load and evaluate downloaded GS model\n",
        "loaded_gs_model = tf.keras.models.load_model(\"/content/downloaded_gs_model/07_efficientnetb0_feature_extract_model_mixed_precision\")"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "1s-w_67lIXo3",
        "outputId": "8e3f98ed-4933-4097-d6e3-278ddbbb05db"
      },
      "source": [
        "# Get a summary of our downloaded model\n",
        "loaded_gs_model.summary()"
      ],
      "execution_count": 47,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"model\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "input_layer (InputLayer)     [(None, 224, 224, 3)]     0         \n",
            "_________________________________________________________________\n",
            "efficientnetb0 (Functional)  (None, None, None, 1280)  4049571   \n",
            "_________________________________________________________________\n",
            "pooling_layer (GlobalAverage (None, 1280)              0         \n",
            "_________________________________________________________________\n",
            "dense (Dense)                (None, 101)               129381    \n",
            "_________________________________________________________________\n",
            "softmax_float32 (Activation) (None, 101)               0         \n",
            "=================================================================\n",
            "Total params: 4,178,952\n",
            "Trainable params: 4,136,929\n",
            "Non-trainable params: 42,023\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ZNfHPBxII5HL",
        "outputId": "447853c6-47e5-4eee-e8a7-7d5a93b62e1b"
      },
      "source": [
        "# How does the loaded model perform?\n",
        "results_loaded_gs_model = loaded_gs_model.evaluate(test_data)\n",
        "results_loaded_gs_model"
      ],
      "execution_count": 48,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "790/790 [==============================] - 47s 58ms/step - loss: 1.0881 - accuracy: 0.7065\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[1.088128924369812, 0.7064950466156006]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 48
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "1hATkpFLJB-J"
      },
      "source": [
        "When we first created our model, we froze all of the layers in the base model by setting `base_model.trainable=False` but since we've loaded in our model from file, let's check whether or not the layers are trainable or not."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "hDMFueuBI86Y",
        "outputId": "bf9b22f1-2e8b-4906-cb41-2714b2910976"
      },
      "source": [
        "# Are any of the layers in our model frozen?\n",
        "for layer in loaded_gs_model.layers:\n",
        "  layer.trainable = True # set all layers to trainable\n",
        "  print(layer.name, layer.trainable, layer.dtype, layer.dtype_policy) # make sure loaded model is using mixed precision dtype_policy (\"mixed_float16\")"
      ],
      "execution_count": 49,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "input_layer True float32 <Policy \"float32\">\n",
            "efficientnetb0 True float32 <Policy \"mixed_float16\">\n",
            "pooling_layer True float32 <Policy \"mixed_float16\">\n",
            "dense True float32 <Policy \"mixed_float16\">\n",
            "softmax_float32 True float32 <Policy \"float32\">\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "X-jn-4ZYJKCK"
      },
      "source": [
        "Before saving the Functional model to file, we created it with five layers (layers below are 0-indexed):\n",
        "\n",
        "* The input layer\n",
        "* The pre-trained base model layer (`tf.keras.applications.EfficientNetB0`)\n",
        "* The pooling layer\n",
        "* The fully-connected (dense) layer\n",
        "* The output softmax activation (with float32 dtype)\n",
        "\n",
        "Therefore to inspect our base model layer, we can access the layers attribute of the layer at index 1 in our model.\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "taivpwcWJYix",
        "outputId": "32a08614-0bdd-4e15-b6a2-b47620f17703"
      },
      "source": [
        "# Check the layers in the base model and see what dtype policy they're using\n",
        "for layer in loaded_gs_model.layers[1].layers[:20]:\n",
        "  print(layer.name, layer.trainable, layer.dtype, layer.dtype_policy)"
      ],
      "execution_count": 51,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "input_1 True float32 <Policy \"float32\">\n",
            "rescaling True float32 <Policy \"mixed_float16\">\n",
            "normalization True float32 <Policy \"float32\">\n",
            "stem_conv_pad True float32 <Policy \"mixed_float16\">\n",
            "stem_conv True float32 <Policy \"mixed_float16\">\n",
            "stem_bn True float32 <Policy \"mixed_float16\">\n",
            "stem_activation True float32 <Policy \"mixed_float16\">\n",
            "block1a_dwconv True float32 <Policy \"mixed_float16\">\n",
            "block1a_bn True float32 <Policy \"mixed_float16\">\n",
            "block1a_activation True float32 <Policy \"mixed_float16\">\n",
            "block1a_se_squeeze True float32 <Policy \"mixed_float16\">\n",
            "block1a_se_reshape True float32 <Policy \"mixed_float16\">\n",
            "block1a_se_reduce True float32 <Policy \"mixed_float16\">\n",
            "block1a_se_expand True float32 <Policy \"mixed_float16\">\n",
            "block1a_se_excite True float32 <Policy \"mixed_float16\">\n",
            "block1a_project_conv True float32 <Policy \"mixed_float16\">\n",
            "block1a_project_bn True float32 <Policy \"mixed_float16\">\n",
            "block2a_expand_conv True float32 <Policy \"mixed_float16\">\n",
            "block2a_expand_bn True float32 <Policy \"mixed_float16\">\n",
            "block2a_expand_activation True float32 <Policy \"mixed_float16\">\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "pmByKMsxJfB2"
      },
      "source": [
        "## **A couple more callbacks**\n",
        "\n",
        "We're about to start fine-tuning a deep learning model with over 200 layers using over 100,000 (75k+ training, 25K+ testing) images, which means our model's training time is probably going to be much longer than before.\n",
        "\n",
        "Ideally, when your model stops improving. But again, due to the nature of deep learning, it can be hard to know when exactly a model will stop improving.\n",
        "\n",
        "Luckily, there's a solution: the EarlyStopping callback.\n",
        "\n",
        "The EarlyStopping callback monitors a specified model performance metric (e.g. val_loss) and when it stops improving for a specified number of epochs, automatically stops training.\n",
        "\n",
        "Using the EarlyStopping callback combined with the ModelCheckpoint callback saving the best performing model automatically, we could keep our model training for an unlimited number of epochs until it stops improving.\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "21NtaziGJxaV"
      },
      "source": [
        "# Setup EarlyStopping callback to stop training if model's val_loss doesn't improve for 3 epochs\n",
        "early_stopping = tf.keras.callbacks.EarlyStopping(monitor=\"val_loss\", # watch the val loss metric\n",
        "                                                  patience=3) # if val loss decreases for 3 epochs in a row, stop training\n",
        "\n",
        "# Create ModelCheckpoint callback to save best model during fine-tuning\n",
        "checkpoint_path = \"fine_tune_checkpoints/\"\n",
        "model_checkpoint = tf.keras.callbacks.ModelCheckpoint(checkpoint_path,\n",
        "                                                      save_best_only=True,\n",
        "                                                      monitor=\"val_loss\")"
      ],
      "execution_count": 52,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "L2E_SdzlJ612"
      },
      "source": [
        "If you're planning on training large models, the `ModelCheckpoint` and `EarlyStopping` are two callbacks you'll want to become very familiar with.\n",
        "\n",
        "We're almost ready to start fine-tuning our model but there's one more callback we're going to implement: `ReduceLROnPlateau`.\n",
        "\n",
        "Remember how the learning rate is the most important model hyperparameter you can tune? (if not, treat this as a reminder).\n",
        "\n",
        "Well, the `ReduceLROnPlateau` callback helps to tune the learning rate for you.\n",
        "\n",
        "Like the `ModelCheckpoint` and `EarlyStopping` callbacks, the `ReduceLROnPlateau` callback montiors a specified metric and when that metric stops improving, it reduces the learning rate by a specified factor (e.g. divides the learning rate by 10).\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "83QMULu7KL2E"
      },
      "source": [
        "# Creating learning rate reduction callback\n",
        "reduce_lr = tf.keras.callbacks.ReduceLROnPlateau(monitor=\"val_loss\",  \n",
        "                                                 factor=0.2, # multiply the learning rate by 0.2 (reduce by 5x)\n",
        "                                                 patience=2,\n",
        "                                                 verbose=1, # print out when learning rate goes down \n",
        "                                                 min_lr=1e-7)"
      ],
      "execution_count": 53,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lWXDkOR3KSAI"
      },
      "source": [
        "# Compile the model\n",
        "loaded_gs_model.compile(loss=\"sparse_categorical_crossentropy\", # sparse_categorical_crossentropy for labels that are *not* one-hot\n",
        "                        optimizer=tf.keras.optimizers.Adam(0.0001), # 10x lower learning rate than the default\n",
        "                        metrics=[\"accuracy\"])"
      ],
      "execution_count": 54,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "5n1Sayg6KZig",
        "outputId": "ee03feef-f14d-4f93-bbc9-fa5881373d3e"
      },
      "source": [
        "# Start to fine-tune (all layers)\n",
        "history_101_food_classes_all_data_fine_tune = loaded_gs_model.fit(train_data,\n",
        "                                                        epochs=100, # fine-tune for a maximum of 100 epochs\n",
        "                                                        steps_per_epoch=len(train_data),\n",
        "                                                        validation_data=test_data,\n",
        "                                                        validation_steps=int(0.15 * len(test_data)), # validation during training on 15% of test data\n",
        "                                                        callbacks=[create_tensorboard_callback(\"training_logs\", \"efficientb0_101_classes_all_data_fine_tuning\"), # track the model training logs\n",
        "                                                                   model_checkpoint, # save only the best model during training\n",
        "                                                                   early_stopping, # stop model after X epochs of no improvements\n",
        "                                                                   reduce_lr]) # reduce the learning rate after X epochs of no improvements"
      ],
      "execution_count": 55,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Saving TensorBoard log files to: training_logs/efficientb0_101_classes_all_data_fine_tuning/20210520-142930\n",
            "Epoch 1/100\n",
            "2368/2368 [==============================] - 309s 125ms/step - loss: 0.9725 - accuracy: 0.7411 - val_loss: 0.8089 - val_accuracy: 0.7749\n",
            "INFO:tensorflow:Assets written to: fine_tune_checkpoints/assets\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:Assets written to: fine_tune_checkpoints/assets\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch 2/100\n",
            "2368/2368 [==============================] - 294s 123ms/step - loss: 0.5878 - accuracy: 0.8383 - val_loss: 0.7714 - val_accuracy: 0.7852\n",
            "INFO:tensorflow:Assets written to: fine_tune_checkpoints/assets\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:Assets written to: fine_tune_checkpoints/assets\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch 3/100\n",
            "2368/2368 [==============================] - 294s 124ms/step - loss: 0.3268 - accuracy: 0.9092 - val_loss: 0.8404 - val_accuracy: 0.7860\n",
            "Epoch 4/100\n",
            "2368/2368 [==============================] - 290s 122ms/step - loss: 0.1669 - accuracy: 0.9515 - val_loss: 0.9593 - val_accuracy: 0.7881\n",
            "\n",
            "Epoch 00004: ReduceLROnPlateau reducing learning rate to 1.9999999494757503e-05.\n",
            "Epoch 5/100\n",
            "2368/2368 [==============================] - 292s 123ms/step - loss: 0.0481 - accuracy: 0.9890 - val_loss: 1.0407 - val_accuracy: 0.8032\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nxisTvvZKpUD"
      },
      "source": [
        "# # Save model to Google Drive (optional)\n",
        "# loaded_gs_model.save(\"/content/drive/MyDrive/tensorflow_course/food_vision/07_efficientnetb0_fine_tuned_101_classes_mixed_precision/\")"
      ],
      "execution_count": 56,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "KNJbU5HOKqwV",
        "outputId": "a7d480ba-22a8-47a8-8012-3d746e9de12e"
      },
      "source": [
        "# Save model locally (note: if you're using Google Colab and you save your model locally, it will be deleted when your Google Colab session ends)\n",
        "loaded_gs_model.save(\"07_efficientnetb0_fine_tuned_101_classes_mixed_precision\")"
      ],
      "execution_count": 57,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:Assets written to: 07_efficientnetb0_fine_tuned_101_classes_mixed_precision/assets\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:Assets written to: 07_efficientnetb0_fine_tuned_101_classes_mixed_precision/assets\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "kF84IZqtSkm5",
        "outputId": "e31df2a8-e25f-4fe4-b856-1d45c5e73bb9"
      },
      "source": [
        "# Evaluate mixed precision trained loaded model\n",
        "results_loaded_gs_model_fine_tuned = loaded_gs_model.evaluate(test_data) \n",
        "results_loaded_gs_model_fine_tuned"
      ],
      "execution_count": 66,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "790/790 [==============================] - 62s 76ms/step - loss: 1.0557 - accuracy: 0.7987\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[1.0557451248168945, 0.7986534833908081]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 66
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "bI7gKaKwRswp"
      },
      "source": [
        "Woohoo!!!! It looks like our model beat the results mentioned in the DeepFood paper for Food101 (DeepFood's 77.4% top-1 accuracy versus our ~79% top-1 accuracy)."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "oE-cQ5OpKxku"
      },
      "source": [
        "## **Download fine-tuned model from Google Storage**\n",
        "\n",
        "It's a fine-tuned model exactly like the one we trained above but it's saved to Google Storage so it can be accessed, imported and evaluated.\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "c2pqqsmRSpiD",
        "outputId": "4282dd49-6d35-4a56-ebff-c7cba2ac3e44"
      },
      "source": [
        "# Download and evaluate fine-tuned model from Google Storage\n",
        "!wget https://storage.googleapis.com/ztm_tf_course/food_vision/07_efficientnetb0_fine_tuned_101_classes_mixed_precision.zip"
      ],
      "execution_count": 67,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "--2021-05-20 15:01:03--  https://storage.googleapis.com/ztm_tf_course/food_vision/07_efficientnetb0_fine_tuned_101_classes_mixed_precision.zip\n",
            "Resolving storage.googleapis.com (storage.googleapis.com)... 172.217.7.240, 172.217.8.16, 172.217.13.80, ...\n",
            "Connecting to storage.googleapis.com (storage.googleapis.com)|172.217.7.240|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 46790356 (45M) [application/zip]\n",
            "Saving to: ‘07_efficientnetb0_fine_tuned_101_classes_mixed_precision.zip.1’\n",
            "\n",
            "07_efficientnetb0_f 100%[===================>]  44.62M   260MB/s    in 0.2s    \n",
            "\n",
            "2021-05-20 15:01:03 (260 MB/s) - ‘07_efficientnetb0_fine_tuned_101_classes_mixed_precision.zip.1’ saved [46790356/46790356]\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "RZCSEAbBR1-0"
      },
      "source": [
        "The downloaded model comes in zip format (.zip) so we'll unzip it into the Google Colab instance."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Tz_v8RYTKx0R",
        "outputId": "8b17cb21-c229-4e9c-e3b9-930842f43bf1"
      },
      "source": [
        "# Unzip fine-tuned model\n",
        "!mkdir downloaded_fine_tuned_gs_model # create separate directory for fine-tuned model downloaded from Google Storage\n",
        "!unzip /content/07_efficientnetb0_fine_tuned_101_classes_mixed_precision -d downloaded_fine_tuned_gs_model"
      ],
      "execution_count": 69,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "mkdir: cannot create directory ‘downloaded_fine_tuned_gs_model’: File exists\n",
            "Archive:  /content/07_efficientnetb0_fine_tuned_101_classes_mixed_precision.zip\n",
            "replace downloaded_fine_tuned_gs_model/07_efficientnetb0_fine_tuned_101_classes_mixed_precision/variables/variables.data-00000-of-00001? [y]es, [n]o, [A]ll, [N]one, [r]ename: A\n",
            "  inflating: downloaded_fine_tuned_gs_model/07_efficientnetb0_fine_tuned_101_classes_mixed_precision/variables/variables.data-00000-of-00001  \n",
            "  inflating: downloaded_fine_tuned_gs_model/07_efficientnetb0_fine_tuned_101_classes_mixed_precision/variables/variables.index  \n",
            "  inflating: downloaded_fine_tuned_gs_model/07_efficientnetb0_fine_tuned_101_classes_mixed_precision/saved_model.pb  \n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_bu-oYdOSuXx"
      },
      "source": [
        "# Load in fine-tuned model from Google Storage and evaluate\n",
        "loaded_fine_tuned_gs_model = tf.keras.models.load_model(\"/content/downloaded_fine_tuned_gs_model/07_efficientnetb0_fine_tuned_101_classes_mixed_precision\")"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "M-JkivXGKx7N",
        "outputId": "ade31171-15dd-4cf4-bb7c-21ad09f5b867"
      },
      "source": [
        "# Get a model summary (same model architecture as above)\n",
        "loaded_fine_tuned_gs_model.summary()"
      ],
      "execution_count": 71,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"model\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "input_layer (InputLayer)     [(None, 224, 224, 3)]     0         \n",
            "_________________________________________________________________\n",
            "efficientnetb0 (Functional)  (None, None, None, 1280)  4049571   \n",
            "_________________________________________________________________\n",
            "pooling_layer (GlobalAverage (None, 1280)              0         \n",
            "_________________________________________________________________\n",
            "dense (Dense)                (None, 101)               129381    \n",
            "_________________________________________________________________\n",
            "softmax_float32 (Activation) (None, 101)               0         \n",
            "=================================================================\n",
            "Total params: 4,178,952\n",
            "Trainable params: 4,136,929\n",
            "Non-trainable params: 42,023\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "j6VAgDeYSx6a",
        "outputId": "1707fbc2-3d7b-4d2e-a046-8a2d04d3dbe0"
      },
      "source": [
        "# Note: Even if you're loading in the model from Google Storage, you will still need to load the test_data variable for this cell to work\n",
        "results_downloaded_fine_tuned_gs_model = loaded_fine_tuned_gs_model.evaluate(test_data)\n",
        "results_downloaded_fine_tuned_gs_model"
      ],
      "execution_count": 72,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "790/790 [==============================] - 47s 57ms/step - loss: 0.9073 - accuracy: 0.8017\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.9072550535202026, 0.8017029762268066]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 72
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "MCYyDIpXSG27"
      },
      "source": [
        "Excellent! Our saved model is performing as expected (better results than the DeepFood paper!).\n",
        "\n",
        "Congrautlations! You should be excited! You just trained a computer vision model with competitive performance to a research paper and in far less time (our model took ~20 minutes to train versus DeepFood's quoted 2-3 days)."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ItsKLfzALIPN"
      },
      "source": [
        "## **View training results on TensorBoard**\n",
        "\n",
        "Since we tracked our model's fine-tuning training logs using the TensorBoard callback, let's upload them and inspect them on TensorBoard.dev.\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "bwdmxRY4S0IV",
        "outputId": "3222af9b-9318-4ef7-9022-f382f23a0b51"
      },
      "source": [
        "!tensorboard dev upload --logdir ./training_logs \\\n",
        "  --name \"Fine-tuning EfficientNetB0 on all Food101 Data\" \\\n",
        "  --description \"Training results for fine-tuning EfficientNetB0 on Food101 Data with learning rate 0.0001\" \\\n",
        "  --one_shot"
      ],
      "execution_count": 73,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "2021-05-20 15:03:57.721107: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcudart.so.11.0\n",
            "Data for the \"text\" plugin is now uploaded to TensorBoard.dev! Note that uploaded data is public. If you do not want to upload data for this plugin, use the \"--plugins\" command line argument.\n",
            "\n",
            "New experiment created. View your TensorBoard at: https://tensorboard.dev/experiment/qynNJdVnRBazSMZqTghMJw/\n",
            "\n",
            "\u001b[1m[2021-05-20T15:03:59]\u001b[0m Started scanning logdir.\n",
            "\u001b[1m[2021-05-20T15:04:01]\u001b[0m Total uploaded: 32 scalars, 0 tensors, 2 binary objects (2.0 MB)\n",
            "\u001b[1m[2021-05-20T15:04:01]\u001b[0m Done scanning logdir.\n",
            "\n",
            "\n",
            "Done. View your TensorBoard at https://tensorboard.dev/experiment/qynNJdVnRBazSMZqTghMJw/\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "AigqmlLRLQ1b",
        "outputId": "4c23b31b-0263-48cc-8a9f-32d28c39a4b6"
      },
      "source": [
        "# View past TensorBoard experiments\n",
        "!tensorboard dev list"
      ],
      "execution_count": 74,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "2021-05-20 15:04:08.121622: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcudart.so.11.0\n",
            "Data for the \"text\" plugin is now uploaded to TensorBoard.dev! Note that uploaded data is public. If you do not want to upload data for this plugin, use the \"--plugins\" command line argument.\n",
            "https://tensorboard.dev/experiment/qynNJdVnRBazSMZqTghMJw/\n",
            "\tName                 [No Name]\n",
            "\tDescription          [No Description]\n",
            "\tId                   qynNJdVnRBazSMZqTghMJw\n",
            "\tCreated              2021-05-20 15:03:59 (27 seconds ago)\n",
            "\tUpdated              2021-05-20 15:04:01 (25 seconds ago)\n",
            "\tRuns                 4\n",
            "\tTags                 3\n",
            "\tScalars              32\n",
            "\tTensor bytes         0\n",
            "\tBinary object bytes  2079943\n",
            "https://tensorboard.dev/experiment/hAfKvCgKSOWegwQYnU9X5Q/\n",
            "\tName                 [No Name]\n",
            "\tDescription          [No Description]\n",
            "\tId                   hAfKvCgKSOWegwQYnU9X5Q\n",
            "\tCreated              2021-05-15 09:19:49\n",
            "\tUpdated              2021-05-15 09:19:54\n",
            "\tRuns                 10\n",
            "\tTags                 3\n",
            "\tScalars              168\n",
            "\tTensor bytes         0\n",
            "\tBinary object bytes  4163295\n",
            "https://tensorboard.dev/experiment/mvdnVWTjQx2AxBJMvM6Otw/\n",
            "\tName                 [No Name]\n",
            "\tDescription          [No Description]\n",
            "\tId                   mvdnVWTjQx2AxBJMvM6Otw\n",
            "\tCreated              2021-05-13 17:43:05\n",
            "\tUpdated              2021-05-13 17:43:12\n",
            "\tRuns                 6\n",
            "\tTags                 3\n",
            "\tScalars              84\n",
            "\tTensor bytes         0\n",
            "\tBinary object bytes  10153263\n",
            "Total: 3 experiment(s)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "JpZ9oz-GLcxU"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}