03_testing_fine_tune.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "testing fine_tune",
      "provenance": [],
      "collapsed_sections": [],
      "machine_shape": "hm",
      "authorship_tag": "ABX9TyM2/VwnHFkHa+ZrebPWUiW5",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU",
    "widgets": {
      "application/vnd.jupyter.widget-state+json": {
        "58a723fab628473e982fb2475080c01b": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "model_module_version": "1.5.0",
          "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_c8e5b43e90a74dc3963f762252f0018a",
            "_model_module": "@jupyter-widgets/controls",
            "children": [
              "IPY_MODEL_a5109e1b236b4f16b727f43d5bab4791",
              "IPY_MODEL_c3699a1d42044c53810f583157d85f8a",
              "IPY_MODEL_b1efb601051542938c749774f974fb1a"
            ]
          }
        },
        "c8e5b43e90a74dc3963f762252f0018a": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "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
          }
        },
        "a5109e1b236b4f16b727f43d5bab4791": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_8193617f60174c00a1c5c69fce0ba01a",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": "",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_e1c0708f696d4c2cb19fa53ab2bc351c"
          }
        },
        "c3699a1d42044c53810f583157d85f8a": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "ProgressView",
            "style": "IPY_MODEL_3090fbccd17640efa085a83596894c6d",
            "_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_9b488f2ad4734107b6f37b99d2c9ca02"
          }
        },
        "b1efb601051542938c749774f974fb1a": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_6e870798e63c48a68e201474c62b6f71",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": " 1565769/? [00:20&lt;00:00, 4080102.02B/s]",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_b3c2accb0e624d10b5945db3da03a565"
          }
        },
        "8193617f60174c00a1c5c69fce0ba01a": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "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"
          }
        },
        "e1c0708f696d4c2cb19fa53ab2bc351c": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "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
          }
        },
        "3090fbccd17640efa085a83596894c6d": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "ProgressStyleModel",
            "description_width": "",
            "_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"
          }
        },
        "9b488f2ad4734107b6f37b99d2c9ca02": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "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": "20px",
            "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
          }
        },
        "6e870798e63c48a68e201474c62b6f71": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "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"
          }
        },
        "b3c2accb0e624d10b5945db3da03a565": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "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
          }
        },
        "ddcec391dc6b439482d2f1c77ea05df8": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "model_module_version": "1.5.0",
          "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_d8ee541abe824d5d839660f8736e5153",
            "_model_module": "@jupyter-widgets/controls",
            "children": [
              "IPY_MODEL_0c817235eb9643819855976fd5c081ad",
              "IPY_MODEL_bafb1c902379401b8739b947821d4f49",
              "IPY_MODEL_c5deb542c0fb4ec78c0b9cbbff2a1e69"
            ]
          }
        },
        "d8ee541abe824d5d839660f8736e5153": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "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
          }
        },
        "0c817235eb9643819855976fd5c081ad": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_64ca873f3bd24d1796f2b19649e285dd",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": "100%",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_6aa84fafb79d4b44a73d7753f1f43f8f"
          }
        },
        "bafb1c902379401b8739b947821d4f49": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "ProgressView",
            "style": "IPY_MODEL_d4390bbf4e694651b83df530d7249783",
            "_dom_classes": [],
            "description": "",
            "_model_name": "FloatProgressModel",
            "bar_style": "",
            "max": 791918971,
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": 791918971,
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "orientation": "horizontal",
            "min": 0,
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_87e509f7d08d4703933b726c89f4e832"
          }
        },
        "c5deb542c0fb4ec78c0b9cbbff2a1e69": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_a26f0df882e14069a65a632e39027361",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": " 791918971/791918971 [00:39&lt;00:00, 32034781.79B/s]",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_22d7e5f0222d41e2b5dca12d6845c3a6"
          }
        },
        "64ca873f3bd24d1796f2b19649e285dd": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "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"
          }
        },
        "6aa84fafb79d4b44a73d7753f1f43f8f": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "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
          }
        },
        "d4390bbf4e694651b83df530d7249783": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "ProgressStyleModel",
            "description_width": "",
            "_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"
          }
        },
        "87e509f7d08d4703933b726c89f4e832": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "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
          }
        },
        "a26f0df882e14069a65a632e39027361": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "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"
          }
        },
        "22d7e5f0222d41e2b5dca12d6845c3a6": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "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
          }
        },
        "d90cfdd4a4d246cc95d857c9af9ed87d": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "model_module_version": "1.5.0",
          "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_343c406cdcb947aa8d76ea842bfc72c8",
            "_model_module": "@jupyter-widgets/controls",
            "children": [
              "IPY_MODEL_a89e783b91bd4e3a8b7ed258de50e122",
              "IPY_MODEL_8548b88b5db64238904c8f005ecffff1",
              "IPY_MODEL_5be8eb634b764a9ab3331ab276f37bea"
            ]
          }
        },
        "343c406cdcb947aa8d76ea842bfc72c8": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "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
          }
        },
        "a89e783b91bd4e3a8b7ed258de50e122": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_17d8571b585743c5b7231ce1652b73dd",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": "100%",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_967ee2c0ffcf4b04860f321fe0e76545"
          }
        },
        "8548b88b5db64238904c8f005ecffff1": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "ProgressView",
            "style": "IPY_MODEL_248288fd0254471e808a44fe78cb97c3",
            "_dom_classes": [],
            "description": "",
            "_model_name": "FloatProgressModel",
            "bar_style": "",
            "max": 19173078,
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": 19173078,
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "orientation": "horizontal",
            "min": 0,
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_3f3dc922b3944d3ca0c03237b00e9801"
          }
        },
        "5be8eb634b764a9ab3331ab276f37bea": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_f10e9e8a2df7437aa3e4be2933a4a2b1",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": " 19173078/19173078 [00:17&lt;00:00, 25786585.88B/s]",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_68102b8b521c411ea9f1be19359a9360"
          }
        },
        "17d8571b585743c5b7231ce1652b73dd": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "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"
          }
        },
        "967ee2c0ffcf4b04860f321fe0e76545": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "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
          }
        },
        "248288fd0254471e808a44fe78cb97c3": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "ProgressStyleModel",
            "description_width": "",
            "_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"
          }
        },
        "3f3dc922b3944d3ca0c03237b00e9801": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "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
          }
        },
        "f10e9e8a2df7437aa3e4be2933a4a2b1": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "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"
          }
        },
        "68102b8b521c411ea9f1be19359a9360": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "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
          }
        },
        "543dcd15ca134f9fbf0d0db0dcf9501d": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "model_module_version": "1.5.0",
          "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_f9bbea2e692247968c8f3a3bf385c8e9",
            "_model_module": "@jupyter-widgets/controls",
            "children": [
              "IPY_MODEL_f182a1e17a7a41eca4a869ec99c059e1",
              "IPY_MODEL_976f9fae79e74812a05f692198c937f0",
              "IPY_MODEL_254a5d8881d147ccb3612174f8fe1dc0"
            ]
          }
        },
        "f9bbea2e692247968c8f3a3bf385c8e9": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "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
          }
        },
        "f182a1e17a7a41eca4a869ec99c059e1": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_70f11e12f9d240fba882320bf08a4828",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": "100%",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_01432e9d582847179861a4addde478d8"
          }
        },
        "976f9fae79e74812a05f692198c937f0": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "ProgressView",
            "style": "IPY_MODEL_b50d775707df4778a389d89eeb966948",
            "_dom_classes": [],
            "description": "",
            "_model_name": "FloatProgressModel",
            "bar_style": "success",
            "max": 3686,
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": 3686,
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "orientation": "horizontal",
            "min": 0,
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_5b6d44ea33744d159021f8304cee17c8"
          }
        },
        "254a5d8881d147ccb3612174f8fe1dc0": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_f667816300194b219e2e54beb55c3307",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": " 3686/3686 [00:01&lt;00:00, 2543.31it/s]",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_8d243b529b2f4171817d4b6fe1e38624"
          }
        },
        "70f11e12f9d240fba882320bf08a4828": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "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"
          }
        },
        "01432e9d582847179861a4addde478d8": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "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
          }
        },
        "b50d775707df4778a389d89eeb966948": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "ProgressStyleModel",
            "description_width": "",
            "_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"
          }
        },
        "5b6d44ea33744d159021f8304cee17c8": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "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
          }
        },
        "f667816300194b219e2e54beb55c3307": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "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"
          }
        },
        "8d243b529b2f4171817d4b6fe1e38624": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "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
          }
        },
        "9cbf1670a328415ab9b71119685c56de": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "model_module_version": "1.5.0",
          "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_93f09d816eb847aeb54646a1f30f4c7d",
            "_model_module": "@jupyter-widgets/controls",
            "children": [
              "IPY_MODEL_52564629839744aea76071a63eba213d",
              "IPY_MODEL_8b007989bcbb4d60b83b3b55488d2a3f",
              "IPY_MODEL_f212ac7b6cb14e51a89eadfe9d7d3440"
            ]
          }
        },
        "93f09d816eb847aeb54646a1f30f4c7d": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "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
          }
        },
        "52564629839744aea76071a63eba213d": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_e0b95518792641d9b046afa9d74b00cd",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": "100%",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_ffdab57e5b2c4fb5a200454456968bfe"
          }
        },
        "8b007989bcbb4d60b83b3b55488d2a3f": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "ProgressView",
            "style": "IPY_MODEL_cee9d7558c0e4f909527f89c5f6d10d1",
            "_dom_classes": [],
            "description": "",
            "_model_name": "FloatProgressModel",
            "bar_style": "success",
            "max": 2948,
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": 2948,
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "orientation": "horizontal",
            "min": 0,
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_f2528f3503384957b82527280c37f05a"
          }
        },
        "f212ac7b6cb14e51a89eadfe9d7d3440": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_7cca28502ed540349116a3fc42e30505",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": " 2948/2948 [00:00&lt;00:00, 13045.25it/s]",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_9e54b93488594f3981beaeabe76b6251"
          }
        },
        "e0b95518792641d9b046afa9d74b00cd": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "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"
          }
        },
        "ffdab57e5b2c4fb5a200454456968bfe": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "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
          }
        },
        "cee9d7558c0e4f909527f89c5f6d10d1": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "ProgressStyleModel",
            "description_width": "",
            "_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"
          }
        },
        "f2528f3503384957b82527280c37f05a": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "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
          }
        },
        "7cca28502ed540349116a3fc42e30505": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "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"
          }
        },
        "9e54b93488594f3981beaeabe76b6251": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "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
          }
        },
        "ec582091aa9b40edafe375a38a57a9aa": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "model_module_version": "1.5.0",
          "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_35a45d8ccbac41a5a959367e34d8ea3e",
            "_model_module": "@jupyter-widgets/controls",
            "children": [
              "IPY_MODEL_496bcf46fc4e44029030009460aa2bb6",
              "IPY_MODEL_473e858775204eeeb1302493b21ae229",
              "IPY_MODEL_89176c41899f4eea8a810d7d69d830e0"
            ]
          }
        },
        "35a45d8ccbac41a5a959367e34d8ea3e": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "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
          }
        },
        "496bcf46fc4e44029030009460aa2bb6": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_d9cb4de6ff3a4f15bf589338b69736ba",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": "100%",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_33e98f4cad46480e9f66005ca57afc5e"
          }
        },
        "473e858775204eeeb1302493b21ae229": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "ProgressView",
            "style": "IPY_MODEL_504f7f10681045579ed06eae706a836e",
            "_dom_classes": [],
            "description": "",
            "_model_name": "FloatProgressModel",
            "bar_style": "success",
            "max": 738,
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": 738,
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "orientation": "horizontal",
            "min": 0,
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_8a818fe059f04b03a0cccababe9bfa69"
          }
        },
        "89176c41899f4eea8a810d7d69d830e0": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_77e34b63756f482584fe7c2d7b958c3a",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": " 738/738 [00:00&lt;00:00, 10033.76it/s]",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_42a821b6a6184719ac3afa57560c16f5"
          }
        },
        "d9cb4de6ff3a4f15bf589338b69736ba": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "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"
          }
        },
        "33e98f4cad46480e9f66005ca57afc5e": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "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
          }
        },
        "504f7f10681045579ed06eae706a836e": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "ProgressStyleModel",
            "description_width": "",
            "_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"
          }
        },
        "8a818fe059f04b03a0cccababe9bfa69": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "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
          }
        },
        "77e34b63756f482584fe7c2d7b958c3a": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "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"
          }
        },
        "42a821b6a6184719ac3afa57560c16f5": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "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
          }
        },
        "9efdcf7920e24885ace0fc46ed4ebff9": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "model_module_version": "1.5.0",
          "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_b1c050d993e2482fa7762e391be2ac01",
            "_model_module": "@jupyter-widgets/controls",
            "children": [
              "IPY_MODEL_53cef9bbb68e4910a970d04db611204f",
              "IPY_MODEL_4bf6c0401605470aa5afc0dac382851e",
              "IPY_MODEL_8d585435c18645fd94089754cb41b90f"
            ]
          }
        },
        "b1c050d993e2482fa7762e391be2ac01": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "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
          }
        },
        "53cef9bbb68e4910a970d04db611204f": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_5c5d43ce2d3e4b81b61b72f210442914",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": "100%",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_712c885ef9904e35bc63a22880e7b2d3"
          }
        },
        "4bf6c0401605470aa5afc0dac382851e": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "ProgressView",
            "style": "IPY_MODEL_061682709a0b4a3ba2966798cb9ce428",
            "_dom_classes": [],
            "description": "",
            "_model_name": "FloatProgressModel",
            "bar_style": "",
            "max": 152149505,
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": 152149505,
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "orientation": "horizontal",
            "min": 0,
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_8e768c2d2b064a3e938cbdf6bb0a5a85"
          }
        },
        "8d585435c18645fd94089754cb41b90f": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_09af428d3e864a44acf1d940aa2c9b5e",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": " 152149505/152149505 [00:35&lt;00:00, 7658014.17B/s]",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_adf06273e5314ff6a7793f35b295c0d8"
          }
        },
        "5c5d43ce2d3e4b81b61b72f210442914": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "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"
          }
        },
        "712c885ef9904e35bc63a22880e7b2d3": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "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
          }
        },
        "061682709a0b4a3ba2966798cb9ce428": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "ProgressStyleModel",
            "description_width": "",
            "_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"
          }
        },
        "8e768c2d2b064a3e938cbdf6bb0a5a85": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "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
          }
        },
        "09af428d3e864a44acf1d940aa2c9b5e": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "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"
          }
        },
        "adf06273e5314ff6a7793f35b295c0d8": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "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/yrodriguezmd/IceVision_miniprojects/blob/main/testing_fine_tune.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "LWHd-RzUKQC9",
        "outputId": "482aca1c-307f-4f81-c613-63d7238f0232"
      },
      "source": [
        "!wget https://raw.githubusercontent.com/airctic/icevision/master/install_colab.sh\n",
        "!bash install_colab.sh\n",
        "\n",
        "print(torch.__version__)\n",
        "print(torchvision.__version__)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "--2021-08-25 22:33:45--  https://raw.githubusercontent.com/airctic/icevision/master/install_colab.sh\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: 1244 (1.2K) [text/plain]\n",
            "Saving to: ‘install_colab.sh’\n",
            "\n",
            "\rinstall_colab.sh      0%[                    ]       0  --.-KB/s               \rinstall_colab.sh    100%[===================>]   1.21K  --.-KB/s    in 0s      \n",
            "\n",
            "2021-08-25 22:33:45 (102 MB/s) - ‘install_colab.sh’ saved [1244/1244]\n",
            "\n",
            "Installing icevision + dependencices for CUDA 10\n",
            "Uninstalling some dependencies to prevent errors\n",
            "Found existing installation: torchvision 0.10.0+cu102\n",
            "Uninstalling torchvision-0.10.0+cu102:\n",
            "  Successfully uninstalled torchvision-0.10.0+cu102\n",
            "Found existing installation: fastai 1.0.61\n",
            "Uninstalling fastai-1.0.61:\n",
            "  Successfully uninstalled fastai-1.0.61\n",
            "Installing some dependencies to prevent errors\n",
            "\u001b[K     |████████████████████████████████| 721 kB 5.1 MB/s \n",
            "\u001b[K     |████████████████████████████████| 94 kB 5.0 MB/s \n",
            "\u001b[?25h  Building wheel for datascience (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "\u001b[K     |████████████████████████████████| 394.7 MB 17 kB/s \n",
            "\u001b[K     |████████████████████████████████| 462 kB 84.6 MB/s \n",
            "\u001b[K     |████████████████████████████████| 3.8 MB 71.6 MB/s \n",
            "\u001b[K     |████████████████████████████████| 2.9 MB 69.5 MB/s \n",
            "\u001b[?25h- Installing torch and its dependencies\n",
            "- Installing torch and its dependencies\n",
            "\u001b[K     |████████████████████████████████| 7.1 MB 4.2 MB/s \n",
            "\u001b[K     |████████████████████████████████| 735.5 MB 14 kB/s \n",
            "\u001b[K     |████████████████████████████████| 763.5 MB 11 kB/s \n",
            "\u001b[K     |████████████████████████████████| 17.3 MB 291 kB/s \n",
            "\u001b[?25h- Installing fastai\n",
            "\u001b[K     |████████████████████████████████| 194 kB 5.1 MB/s \n",
            "\u001b[K     |████████████████████████████████| 56 kB 5.2 MB/s \n",
            "\u001b[?25h- Installing icevision from master\n",
            "\u001b[K     |████████████████████████████████| 72 kB 737 kB/s \n",
            "\u001b[K     |████████████████████████████████| 57 kB 6.4 MB/s \n",
            "\u001b[K     |████████████████████████████████| 3.0 MB 29.7 MB/s \n",
            "\u001b[K     |████████████████████████████████| 918 kB 65.4 MB/s \n",
            "\u001b[K     |████████████████████████████████| 111 kB 102.3 MB/s \n",
            "\u001b[K     |████████████████████████████████| 74 kB 4.4 MB/s \n",
            "\u001b[K     |████████████████████████████████| 1.6 MB 41.1 MB/s \n",
            "\u001b[K     |████████████████████████████████| 49 kB 6.1 MB/s \n",
            "\u001b[K     |████████████████████████████████| 948 kB 101.6 MB/s \n",
            "\u001b[K     |████████████████████████████████| 37.1 MB 75 kB/s \n",
            "\u001b[K     |████████████████████████████████| 376 kB 76.3 MB/s \n",
            "\u001b[K     |████████████████████████████████| 112 kB 69.0 MB/s \n",
            "\u001b[K     |████████████████████████████████| 272 kB 71.6 MB/s \n",
            "\u001b[K     |████████████████████████████████| 118 kB 95.5 MB/s \n",
            "\u001b[K     |████████████████████████████████| 829 kB 81.4 MB/s \n",
            "\u001b[K     |████████████████████████████████| 1.3 MB 75.4 MB/s \n",
            "\u001b[K     |████████████████████████████████| 154 kB 75.5 MB/s \n",
            "\u001b[K     |████████████████████████████████| 49 kB 6.9 MB/s \n",
            "\u001b[K     |████████████████████████████████| 133 kB 67.6 MB/s \n",
            "\u001b[K     |████████████████████████████████| 97 kB 8.6 MB/s \n",
            "\u001b[K     |████████████████████████████████| 170 kB 69.2 MB/s \n",
            "\u001b[K     |████████████████████████████████| 63 kB 2.1 MB/s \n",
            "\u001b[K     |████████████████████████████████| 142 kB 71.5 MB/s \n",
            "\u001b[K     |████████████████████████████████| 294 kB 70.1 MB/s \n",
            "\u001b[?25h  Building wheel for icevision (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "  Building wheel for antlr4-python3-runtime (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "  Building wheel for future (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "  Building wheel for subprocess32 (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "  Building wheel for fvcore (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "  Building wheel for pathtools (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "- Installing icedata from master\n",
            "\u001b[K     |████████████████████████████████| 87 kB 3.9 MB/s \n",
            "\u001b[?25h  Building wheel for icedata (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "  Building wheel for fire (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "- Installing yolov5-icevision\n",
            "\u001b[K     |████████████████████████████████| 757 kB 5.1 MB/s \n",
            "\u001b[?25h- Installing mmcv\n",
            "\u001b[K     |████████████████████████████████| 31.4 MB 71 kB/s \n",
            "\u001b[K     |████████████████████████████████| 185 kB 5.0 MB/s \n",
            "\u001b[?25h- Installing mmdet\n",
            "\u001b[K     |████████████████████████████████| 610 kB 5.0 MB/s \n",
            "\u001b[?25h  Building wheel for terminaltables (setup.py) ... \u001b[?25l\u001b[?25hdone\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 280,
          "referenced_widgets": [
            "58a723fab628473e982fb2475080c01b",
            "c8e5b43e90a74dc3963f762252f0018a",
            "a5109e1b236b4f16b727f43d5bab4791",
            "c3699a1d42044c53810f583157d85f8a",
            "b1efb601051542938c749774f974fb1a",
            "8193617f60174c00a1c5c69fce0ba01a",
            "e1c0708f696d4c2cb19fa53ab2bc351c",
            "3090fbccd17640efa085a83596894c6d",
            "9b488f2ad4734107b6f37b99d2c9ca02",
            "6e870798e63c48a68e201474c62b6f71",
            "b3c2accb0e624d10b5945db3da03a565",
            "ddcec391dc6b439482d2f1c77ea05df8",
            "d8ee541abe824d5d839660f8736e5153",
            "0c817235eb9643819855976fd5c081ad",
            "bafb1c902379401b8739b947821d4f49",
            "c5deb542c0fb4ec78c0b9cbbff2a1e69",
            "64ca873f3bd24d1796f2b19649e285dd",
            "6aa84fafb79d4b44a73d7753f1f43f8f",
            "d4390bbf4e694651b83df530d7249783",
            "87e509f7d08d4703933b726c89f4e832",
            "a26f0df882e14069a65a632e39027361",
            "22d7e5f0222d41e2b5dca12d6845c3a6",
            "d90cfdd4a4d246cc95d857c9af9ed87d",
            "343c406cdcb947aa8d76ea842bfc72c8",
            "a89e783b91bd4e3a8b7ed258de50e122",
            "8548b88b5db64238904c8f005ecffff1",
            "5be8eb634b764a9ab3331ab276f37bea",
            "17d8571b585743c5b7231ce1652b73dd",
            "967ee2c0ffcf4b04860f321fe0e76545",
            "248288fd0254471e808a44fe78cb97c3",
            "3f3dc922b3944d3ca0c03237b00e9801",
            "f10e9e8a2df7437aa3e4be2933a4a2b1",
            "68102b8b521c411ea9f1be19359a9360",
            "543dcd15ca134f9fbf0d0db0dcf9501d",
            "f9bbea2e692247968c8f3a3bf385c8e9",
            "f182a1e17a7a41eca4a869ec99c059e1",
            "976f9fae79e74812a05f692198c937f0",
            "254a5d8881d147ccb3612174f8fe1dc0",
            "70f11e12f9d240fba882320bf08a4828",
            "01432e9d582847179861a4addde478d8",
            "b50d775707df4778a389d89eeb966948",
            "5b6d44ea33744d159021f8304cee17c8",
            "f667816300194b219e2e54beb55c3307",
            "8d243b529b2f4171817d4b6fe1e38624",
            "9cbf1670a328415ab9b71119685c56de",
            "93f09d816eb847aeb54646a1f30f4c7d",
            "52564629839744aea76071a63eba213d",
            "8b007989bcbb4d60b83b3b55488d2a3f",
            "f212ac7b6cb14e51a89eadfe9d7d3440",
            "e0b95518792641d9b046afa9d74b00cd",
            "ffdab57e5b2c4fb5a200454456968bfe",
            "cee9d7558c0e4f909527f89c5f6d10d1",
            "f2528f3503384957b82527280c37f05a",
            "7cca28502ed540349116a3fc42e30505",
            "9e54b93488594f3981beaeabe76b6251",
            "ec582091aa9b40edafe375a38a57a9aa",
            "35a45d8ccbac41a5a959367e34d8ea3e",
            "496bcf46fc4e44029030009460aa2bb6",
            "473e858775204eeeb1302493b21ae229",
            "89176c41899f4eea8a810d7d69d830e0",
            "d9cb4de6ff3a4f15bf589338b69736ba",
            "33e98f4cad46480e9f66005ca57afc5e",
            "504f7f10681045579ed06eae706a836e",
            "8a818fe059f04b03a0cccababe9bfa69",
            "77e34b63756f482584fe7c2d7b958c3a",
            "42a821b6a6184719ac3afa57560c16f5"
          ]
        },
        "id": "xZRESAQ0KV0J",
        "outputId": "6d52d0d2-a8fe-411e-a8b5-51fd421a95df"
      },
      "source": [
        "from icevision.all import *\n",
        "import icedata\n",
        "#import icevision\n",
        "#print(icevision.__version__)\n",
        "\n",
        "path = icedata.pets.load_data()\n",
        "\n",
        "class_map = icedata.pets.class_map()\n",
        "\n",
        "data_splitter = RandomSplitter([0.8, 0.2])\n",
        "\n",
        "parser = icedata.pets.parser(data_dir=path)# removed mask\n",
        "\n",
        "train_records, valid_records = parser.parse(data_splitter)"
      ],
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\u001b[1m\u001b[1mINFO    \u001b[0m\u001b[1m\u001b[0m - \u001b[1mDownloading default `.ttf` font file - SpaceGrotesk-Medium.ttf from https://raw.githubusercontent.com/airctic/storage/master/SpaceGrotesk-Medium.ttf to /root/.icevision/fonts/SpaceGrotesk-Medium.ttf\u001b[0m | \u001b[36micevision.visualize.utils\u001b[0m:\u001b[36mget_default_font\u001b[0m:\u001b[36m69\u001b[0m\n",
            "\u001b[1m\u001b[1mINFO    \u001b[0m\u001b[1m\u001b[0m - \u001b[1mDownloading mmdet configs\u001b[0m | \u001b[36micevision.models.mmdet.download_configs\u001b[0m:\u001b[36mdownload_mmdet_configs\u001b[0m:\u001b[36m31\u001b[0m\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "58a723fab628473e982fb2475080c01b",
              "version_minor": 0,
              "version_major": 2
            },
            "text/plain": [
              "0B [00:00, ?B/s]"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "ddcec391dc6b439482d2f1c77ea05df8",
              "version_minor": 0,
              "version_major": 2
            },
            "text/plain": [
              "  0%|          | 0/791918971 [00:00<?, ?B/s]"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "d90cfdd4a4d246cc95d857c9af9ed87d",
              "version_minor": 0,
              "version_major": 2
            },
            "text/plain": [
              "  0%|          | 0/19173078 [00:00<?, ?B/s]"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "543dcd15ca134f9fbf0d0db0dcf9501d",
              "version_minor": 0,
              "version_major": 2
            },
            "text/plain": [
              "  0%|          | 0/3686 [00:00<?, ?it/s]"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "stream",
          "text": [
            "\u001b[1m\u001b[1mINFO    \u001b[0m\u001b[1m\u001b[0m - \u001b[1m\u001b[34m\u001b[1mAutofixing records\u001b[0m\u001b[1m\u001b[34m\u001b[0m\u001b[1m\u001b[0m | \u001b[36micevision.parsers.parser\u001b[0m:\u001b[36mparse\u001b[0m:\u001b[36m136\u001b[0m\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "9cbf1670a328415ab9b71119685c56de",
              "version_minor": 0,
              "version_major": 2
            },
            "text/plain": [
              "  0%|          | 0/2948 [00:00<?, ?it/s]"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "ec582091aa9b40edafe375a38a57a9aa",
              "version_minor": 0,
              "version_major": 2
            },
            "text/plain": [
              "  0%|          | 0/738 [00:00<?, ?it/s]"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BtG_vjH_LD_3"
      },
      "source": [
        "image_size = 384\n",
        "train_tfms = tfms.A.Adapter([*tfms.A.aug_tfms(size=image_size, presize=512), tfms.A.Normalize()])\n",
        "valid_tfms = tfms.A.Adapter([*tfms.A.resize_and_pad(image_size), tfms.A.Normalize()])\n",
        "\n",
        "train_ds = Dataset(train_records, train_tfms)\n",
        "valid_ds = Dataset(valid_records, valid_tfms)"
      ],
      "execution_count": 2,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3WTVnEqRMDdT"
      },
      "source": [
        "metrics = [COCOMetric(metric_type=COCOMetricType.bbox)]"
      ],
      "execution_count": 4,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "c3y0v1dkSvqn"
      },
      "source": [
        "#### side project fine_tune cumulative?"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "avle-SEdLcqT"
      },
      "source": [
        "import matplotlib.pyplot as plt\n",
        "def plot_metrics(learn, title, x, y):\n",
        "  plt.plot(L(learn.recorder.values).itemgot())#\n",
        "  plt.xlabel('epoch')\n",
        "  plt.ylabel('mAP (green), Loss (blue, orange)')\n",
        "  plt.title(title)\n",
        "  plt.text(x, y, 'Legend: mAP(green), train_loss(blue), valid_loss(orange');"
      ],
      "execution_count": 9,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "SDrMqXYvM_gK"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 188,
          "referenced_widgets": [
            "9efdcf7920e24885ace0fc46ed4ebff9",
            "b1c050d993e2482fa7762e391be2ac01",
            "53cef9bbb68e4910a970d04db611204f",
            "4bf6c0401605470aa5afc0dac382851e",
            "8d585435c18645fd94089754cb41b90f",
            "5c5d43ce2d3e4b81b61b72f210442914",
            "712c885ef9904e35bc63a22880e7b2d3",
            "061682709a0b4a3ba2966798cb9ce428",
            "8e768c2d2b064a3e938cbdf6bb0a5a85",
            "09af428d3e864a44acf1d940aa2c9b5e",
            "adf06273e5314ff6a7793f35b295c0d8"
          ]
        },
        "id": "cEKHXyWALJ2p",
        "outputId": "49969bc4-0596-4208-9095-2c5c37c96538"
      },
      "source": [
        "# using retinanet/ resnet50: focal loss\n",
        "model_type_ret = models.mmdet.retinanet #\n",
        "backbone_r50 = model_type_ret.backbones.resnet50_fpn_1x(pretrained=True)#\n",
        "model_ret = model_type_ret.model(backbone=backbone_r50(pretrained=True), #\n",
        "                      num_classes=len(parser.class_map))\n",
        "\n",
        "train_dl_ret = model_type_ret.train_dl(train_ds, batch_size=16, num_workers=4, shuffle=True)#\n",
        "valid_dl_ret = model_type_ret.valid_dl(valid_ds, batch_size=16, num_workers=4, shuffle=False)#\n",
        "\n",
        "learn_ret = model_type_ret.fastai.learner(dls=[train_dl_ret, valid_dl_ret], model=model_ret, metrics=metrics)#\n",
        "#learn_ret.lr_find()#\n",
        "# use ***"
      ],
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "9efdcf7920e24885ace0fc46ed4ebff9",
              "version_minor": 0,
              "version_major": 2
            },
            "text/plain": [
              "  0%|          | 0/152149505 [00:00<?, ?B/s]"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/mmcv/cnn/utils/weight_init.py:119: UserWarning: init_cfg without layer key, if you do not define override key either, this init_cfg will do nothing\n",
            "  'init_cfg without layer key, if you do not define override'\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "Use load_from_local loader\n",
            "The model and loaded state dict do not match exactly\n",
            "\n",
            "size mismatch for bbox_head.retina_cls.weight: copying a param with shape torch.Size([720, 256, 3, 3]) from checkpoint, the shape in current model is torch.Size([333, 256, 3, 3]).\n",
            "size mismatch for bbox_head.retina_cls.bias: copying a param with shape torch.Size([720]) from checkpoint, the shape in current model is torch.Size([333]).\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 205
        },
        "id": "Brbe0M0rLajv",
        "outputId": "9b4e3ca2-5cb2-4253-c3c4-9dc8bf8002b5"
      },
      "source": [
        "learn_ret.fine_tune(3, 0.00012, freeze_epochs=1) #"
      ],
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: left;\">\n",
              "      <th>epoch</th>\n",
              "      <th>train_loss</th>\n",
              "      <th>valid_loss</th>\n",
              "      <th>COCOMetric</th>\n",
              "      <th>time</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <td>0</td>\n",
              "      <td>0.885305</td>\n",
              "      <td>0.738110</td>\n",
              "      <td>0.071786</td>\n",
              "      <td>01:38</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: left;\">\n",
              "      <th>epoch</th>\n",
              "      <th>train_loss</th>\n",
              "      <th>valid_loss</th>\n",
              "      <th>COCOMetric</th>\n",
              "      <th>time</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <td>0</td>\n",
              "      <td>0.631863</td>\n",
              "      <td>0.579138</td>\n",
              "      <td>0.227640</td>\n",
              "      <td>01:43</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>1</td>\n",
              "      <td>0.508577</td>\n",
              "      <td>0.465558</td>\n",
              "      <td>0.376020</td>\n",
              "      <td>01:37</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>2</td>\n",
              "      <td>0.450621</td>\n",
              "      <td>0.433971</td>\n",
              "      <td>0.414720</td>\n",
              "      <td>01:37</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 364
        },
        "id": "Xq7tIua9Lfq8",
        "outputId": "3c72a77a-b535-4cac-ae3f-708cd340d6b1"
      },
      "source": [
        " plot_metrics(learn_ret, 'Mean Average Precision and Losses for Retinanet/Resnet50', 0, 0.01)"
      ],
      "execution_count": 10,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "sMlzgmDMOack",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 205
        },
        "outputId": "ba737c3d-958c-45c5-9087-e7c399a05406"
      },
      "source": [
        "# 2nd fine tune\n",
        "learn_ret.fine_tune(3, 0.00012, freeze_epochs=1) #\n",
        "## continuation from prior?"
      ],
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: left;\">\n",
              "      <th>epoch</th>\n",
              "      <th>train_loss</th>\n",
              "      <th>valid_loss</th>\n",
              "      <th>COCOMetric</th>\n",
              "      <th>time</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <td>0</td>\n",
              "      <td>0.475929</td>\n",
              "      <td>0.449610</td>\n",
              "      <td>0.432463</td>\n",
              "      <td>01:27</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: left;\">\n",
              "      <th>epoch</th>\n",
              "      <th>train_loss</th>\n",
              "      <th>valid_loss</th>\n",
              "      <th>COCOMetric</th>\n",
              "      <th>time</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <td>0</td>\n",
              "      <td>0.439045</td>\n",
              "      <td>0.404363</td>\n",
              "      <td>0.500039</td>\n",
              "      <td>01:36</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>1</td>\n",
              "      <td>0.388750</td>\n",
              "      <td>0.372058</td>\n",
              "      <td>0.568475</td>\n",
              "      <td>01:36</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>2</td>\n",
              "      <td>0.352186</td>\n",
              "      <td>0.346571</td>\n",
              "      <td>0.606487</td>\n",
              "      <td>01:35</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 513
        },
        "id": "dIhfb53xfYkF",
        "outputId": "4e43dbc0-1486-4e05-f4a7-78b1d53f46b2"
      },
      "source": [
        " plot_metrics(learn_ret, '2nd run - Mean Average Precision and Losses for Retinanet/Resnet50', 0, 0.01)"
      ],
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 205
        },
        "id": "7LAsfEiBfYha",
        "outputId": "4dbfc521-67f7-402b-f69a-3f82df99bd60"
      },
      "source": [
        "# 3rd run\n",
        "learn_ret.fine_tune(3, 0.00012, freeze_epochs=1) #"
      ],
      "execution_count": 13,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: left;\">\n",
              "      <th>epoch</th>\n",
              "      <th>train_loss</th>\n",
              "      <th>valid_loss</th>\n",
              "      <th>COCOMetric</th>\n",
              "      <th>time</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <td>0</td>\n",
              "      <td>0.378452</td>\n",
              "      <td>0.419991</td>\n",
              "      <td>0.539947</td>\n",
              "      <td>01:26</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: left;\">\n",
              "      <th>epoch</th>\n",
              "      <th>train_loss</th>\n",
              "      <th>valid_loss</th>\n",
              "      <th>COCOMetric</th>\n",
              "      <th>time</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <td>0</td>\n",
              "      <td>0.369953</td>\n",
              "      <td>0.379993</td>\n",
              "      <td>0.579699</td>\n",
              "      <td>01:35</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>1</td>\n",
              "      <td>0.343431</td>\n",
              "      <td>0.320099</td>\n",
              "      <td>0.662186</td>\n",
              "      <td>01:34</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>2</td>\n",
              "      <td>0.300995</td>\n",
              "      <td>0.301165</td>\n",
              "      <td>0.682112</td>\n",
              "      <td>01:34</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 408
        },
        "id": "EOH7Ht59fYeo",
        "outputId": "9bb253ee-b6d9-4d0b-cc5f-af95145ce86e"
      },
      "source": [
        " plot_metrics(learn_ret, '2nd run - Mean Average Precision and Losses for Retinanet/Resnet50', 0, 0.01)"
      ],
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "SS8hk5ZDfrQG"
      },
      "source": [
        "# Finding:  cumulative"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4oULzUI3frGN"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 344
        },
        "id": "zlGfk8mqkbMB",
        "outputId": "c99cc1bc-542d-4654-f62f-8331b6f13d53"
      },
      "source": [
        "# 1st run\n",
        "model_type_ret = models.mmdet.retinanet #\n",
        "backbone_r50 = model_type_ret.backbones.resnet50_fpn_1x(pretrained=True)#\n",
        "model_ret = model_type_ret.model(backbone=backbone_r50(pretrained=True), #\n",
        "                      num_classes=len(parser.class_map))\n",
        "\n",
        "train_dl_ret = model_type_ret.train_dl(train_ds, batch_size=16, num_workers=4, shuffle=True)#\n",
        "valid_dl_ret = model_type_ret.valid_dl(valid_ds, batch_size=16, num_workers=4, shuffle=False)#\n",
        "\n",
        "learn_ret = model_type_ret.fastai.learner(dls=[train_dl_ret, valid_dl_ret], model=model_ret, metrics=metrics)#\n",
        "\n",
        "learn_ret.fine_tune(3, 0.00012, freeze_epochs=1) #"
      ],
      "execution_count": 15,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/mmcv/cnn/utils/weight_init.py:119: UserWarning: init_cfg without layer key, if you do not define override key either, this init_cfg will do nothing\n",
            "  'init_cfg without layer key, if you do not define override'\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "Use load_from_local loader\n",
            "The model and loaded state dict do not match exactly\n",
            "\n",
            "size mismatch for bbox_head.retina_cls.weight: copying a param with shape torch.Size([720, 256, 3, 3]) from checkpoint, the shape in current model is torch.Size([333, 256, 3, 3]).\n",
            "size mismatch for bbox_head.retina_cls.bias: copying a param with shape torch.Size([720]) from checkpoint, the shape in current model is torch.Size([333]).\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: left;\">\n",
              "      <th>epoch</th>\n",
              "      <th>train_loss</th>\n",
              "      <th>valid_loss</th>\n",
              "      <th>COCOMetric</th>\n",
              "      <th>time</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <td>0</td>\n",
              "      <td>0.866617</td>\n",
              "      <td>0.729996</td>\n",
              "      <td>0.077778</td>\n",
              "      <td>01:32</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: left;\">\n",
              "      <th>epoch</th>\n",
              "      <th>train_loss</th>\n",
              "      <th>valid_loss</th>\n",
              "      <th>COCOMetric</th>\n",
              "      <th>time</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <td>0</td>\n",
              "      <td>0.624622</td>\n",
              "      <td>0.546720</td>\n",
              "      <td>0.229606</td>\n",
              "      <td>01:40</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>1</td>\n",
              "      <td>0.507447</td>\n",
              "      <td>0.452356</td>\n",
              "      <td>0.396525</td>\n",
              "      <td>01:38</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>2</td>\n",
              "      <td>0.445417</td>\n",
              "      <td>0.420490</td>\n",
              "      <td>0.458363</td>\n",
              "      <td>01:37</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 367
        },
        "id": "QANNUQg1gGGQ",
        "outputId": "4b09ce2e-9350-4c2b-adda-f0a8b93dd135"
      },
      "source": [
        " plot_metrics(learn_ret, 'Mean Average Precision and Losses for Retinanet/Resnet50', 0, 0.01)"
      ],
      "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": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 344
        },
        "id": "tBEiA4drf5xz",
        "outputId": "a5f755e6-b8eb-432d-8aa5-aef15f61611b"
      },
      "source": [
        "# 2nd run\n",
        "model_type_ret = models.mmdet.retinanet #\n",
        "backbone_r50 = model_type_ret.backbones.resnet50_fpn_1x(pretrained=True)#\n",
        "model_ret = model_type_ret.model(backbone=backbone_r50(pretrained=True), #\n",
        "                      num_classes=len(parser.class_map))\n",
        "\n",
        "train_dl_ret = model_type_ret.train_dl(train_ds, batch_size=16, num_workers=4, shuffle=True)#\n",
        "valid_dl_ret = model_type_ret.valid_dl(valid_ds, batch_size=16, num_workers=4, shuffle=False)#\n",
        "\n",
        "learn_ret = model_type_ret.fastai.learner(dls=[train_dl_ret, valid_dl_ret], model=model_ret, metrics=metrics)#\n",
        "\n",
        "learn_ret.fine_tune(3, 0.00012, freeze_epochs=1) #"
      ],
      "execution_count": 17,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/mmcv/cnn/utils/weight_init.py:119: UserWarning: init_cfg without layer key, if you do not define override key either, this init_cfg will do nothing\n",
            "  'init_cfg without layer key, if you do not define override'\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "Use load_from_local loader\n",
            "The model and loaded state dict do not match exactly\n",
            "\n",
            "size mismatch for bbox_head.retina_cls.weight: copying a param with shape torch.Size([720, 256, 3, 3]) from checkpoint, the shape in current model is torch.Size([333, 256, 3, 3]).\n",
            "size mismatch for bbox_head.retina_cls.bias: copying a param with shape torch.Size([720]) from checkpoint, the shape in current model is torch.Size([333]).\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: left;\">\n",
              "      <th>epoch</th>\n",
              "      <th>train_loss</th>\n",
              "      <th>valid_loss</th>\n",
              "      <th>COCOMetric</th>\n",
              "      <th>time</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <td>0</td>\n",
              "      <td>0.875697</td>\n",
              "      <td>0.764025</td>\n",
              "      <td>0.070503</td>\n",
              "      <td>01:33</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: left;\">\n",
              "      <th>epoch</th>\n",
              "      <th>train_loss</th>\n",
              "      <th>valid_loss</th>\n",
              "      <th>COCOMetric</th>\n",
              "      <th>time</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <td>0</td>\n",
              "      <td>0.631249</td>\n",
              "      <td>0.568328</td>\n",
              "      <td>0.220269</td>\n",
              "      <td>01:40</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>1</td>\n",
              "      <td>0.514557</td>\n",
              "      <td>0.476069</td>\n",
              "      <td>0.349182</td>\n",
              "      <td>01:38</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>2</td>\n",
              "      <td>0.459442</td>\n",
              "      <td>0.439911</td>\n",
              "      <td>0.395777</td>\n",
              "      <td>01:37</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 358
        },
        "id": "ykhaknsogH4n",
        "outputId": "494ed9ff-459c-45e7-adba-17324457cdda"
      },
      "source": [
        " plot_metrics(learn_ret, '2nd-run full Mean Average Precision and Losses for Retinanet/Resnet50', 0, 0.01)"
      ],
      "execution_count": 18,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 344
        },
        "id": "zTzGyVgDf_AS",
        "outputId": "f2d8cd81-e10f-4dc0-9a18-1eb02f790882"
      },
      "source": [
        "# 3rd run\n",
        "model_type_ret = models.mmdet.retinanet #\n",
        "backbone_r50 = model_type_ret.backbones.resnet50_fpn_1x(pretrained=True)#\n",
        "model_ret = model_type_ret.model(backbone=backbone_r50(pretrained=True), #\n",
        "                      num_classes=len(parser.class_map))\n",
        "\n",
        "train_dl_ret = model_type_ret.train_dl(train_ds, batch_size=16, num_workers=4, shuffle=True)#\n",
        "valid_dl_ret = model_type_ret.valid_dl(valid_ds, batch_size=16, num_workers=4, shuffle=False)#\n",
        "\n",
        "learn_ret = model_type_ret.fastai.learner(dls=[train_dl_ret, valid_dl_ret], model=model_ret, metrics=metrics)#\n",
        "\n",
        "learn_ret.fine_tune(3, 0.00012, freeze_epochs=1) #\n",
        "\n",
        "#learn_ret.lr_find()#\n",
        "## if don't restart, starts at high map --> accumulated?"
      ],
      "execution_count": 19,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/mmcv/cnn/utils/weight_init.py:119: UserWarning: init_cfg without layer key, if you do not define override key either, this init_cfg will do nothing\n",
            "  'init_cfg without layer key, if you do not define override'\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "Use load_from_local loader\n",
            "The model and loaded state dict do not match exactly\n",
            "\n",
            "size mismatch for bbox_head.retina_cls.weight: copying a param with shape torch.Size([720, 256, 3, 3]) from checkpoint, the shape in current model is torch.Size([333, 256, 3, 3]).\n",
            "size mismatch for bbox_head.retina_cls.bias: copying a param with shape torch.Size([720]) from checkpoint, the shape in current model is torch.Size([333]).\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: left;\">\n",
              "      <th>epoch</th>\n",
              "      <th>train_loss</th>\n",
              "      <th>valid_loss</th>\n",
              "      <th>COCOMetric</th>\n",
              "      <th>time</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <td>0</td>\n",
              "      <td>0.867833</td>\n",
              "      <td>0.737112</td>\n",
              "      <td>0.083764</td>\n",
              "      <td>01:33</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: left;\">\n",
              "      <th>epoch</th>\n",
              "      <th>train_loss</th>\n",
              "      <th>valid_loss</th>\n",
              "      <th>COCOMetric</th>\n",
              "      <th>time</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <td>0</td>\n",
              "      <td>0.621873</td>\n",
              "      <td>0.556343</td>\n",
              "      <td>0.256905</td>\n",
              "      <td>01:41</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>1</td>\n",
              "      <td>0.507060</td>\n",
              "      <td>0.470288</td>\n",
              "      <td>0.378525</td>\n",
              "      <td>01:38</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>2</td>\n",
              "      <td>0.444856</td>\n",
              "      <td>0.428171</td>\n",
              "      <td>0.445341</td>\n",
              "      <td>01:37</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 391
        },
        "id": "M2wYC2cYgMpt",
        "outputId": "e4ed5446-4021-4abd-c625-4899fcceb39b"
      },
      "source": [
        " plot_metrics(learn_ret, '3rd run - full Mean Average Precision and Losses for Retinanet/Resnet50', 0, 0.01)"
      ],
      "execution_count": 20,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAcEAAAF2CAYAAAAWdCMlAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nOy9d3hdV5X3/1nq1SqWJduSrtx7XCTHcgohJIQ4ToeQYhlIAgQY8pJ5KTO0GUKoMzD8yG+GniEMYzkmMaSQQkiAFCCukuPELa4qlm3ZliXLktXX+8c+kq5lSVfWvfZVWZ/nuY909jl7n3Xq9+y1195bVBXDMAzDGI1EhNsAwzAMwwgXJoKGYRjGqMVE0DAMwxi1mAgahmEYoxYTQcMwDGPUYiJoGIZhjFrOuwiKyJUiUnm+9zNYRCReRH4vInUi8sQAtn9FRD7m/X+3iPz1/FtphBsR+amI/MsAttsmIldeAJMGzUi7b0XkVhGpEJFTIrIo3Pb0ZDjcE6OZgCIoIqtE5JCInBSRdzoFYARxG5AFjFXVD4aqUBGZJCIqIqU90jNEpEVEDoRqX+eK9xJUEbkjXDaECr/zfMr7HRCRL4Z6P6r6SVX9xgC2m6uqr4R6/xcKv/MZFW5bzoHvA/erapKqlgbcOgDeh26Tdz8dE5HficiEAeb9lYh80z9tKNwT/X34iMguEZnh2d7iHXeNiLwkIrMutK1+dp11L3qVqg6/5/2UiHzEb326iDwpIg0iUiYiKwLtZyA1we8Ak1R1DHAT8E0RKejD6KAenDA9eHnAO6radp7KTxCReX7LK4D952lfA+UjQA3w4fNReJiuY6qqJgF3Af8qIst6bjDMXuzGwMkDtg0mo4hE9rHqfu9+mgYk4YR2xCEiU4FIVX3HS/p377izgYPAf4fNuL6p8j54On//47fuR0ALrmJTBPxEROb2V1hAEVTVbara3Lno/aZCt6tTRP5ZRA4Dj3ruxV+JyAkR2Q5c3F/5ntJ/WkR2A7v7UP+zXJAi8n1vH/tF5LpAx9HHvr8O/Ctwh/dF8VEReVBEVvltE+yX8f/iRKeTDwO/7mHHRBH5rYgc9Y7nM37rlojIGyJS69XI/0tEYvzWq4h8UkR2e9v8SESkn2POA94N3AdcKyLjvfSfiMj3e2z7tIh8dgA2Pigiaz2vwUng7gHY/T7vC7RORH4sIq+Kn5dBRO4VkR3eNX7RszsgqvoG7oU4r4/7M0JEvigie0XkuIg8LiLpfvu9XET+7tldISJ3e+ldX/jiavPPetvUiMjrIhLhrTsgIu/1/o8VkR+KSJX3+6GIxHrrOm37nIhUe+fonn6u2z3e+agXkX0i8gm/df2WJSJjReQZcd6cDXjP77ni3QPPeMe8R0Q+7rduiYhs8vZxRER+4KXHeffFce98bRSRLG9dioj8t2fvQRH5pniiJCLTvHuiTlxt7De92BMrIqeASOBNEdnrpc8W986oFeeKvMkvz6+8e/15EWkA3tPfMatqLfAUsNCvjFniakk13j18u5d+H+7F+0/i3ie/99L974kHvXvu19613CYii/3K7rw360Vku4jc6reu33dfX+dTRGYDPwUu8eyq9TvE64Hneznu08DjPY470Huqt+vf+f78iIiUe9fyK375+nseX/P+1np2X9LftRKRROADwL+o6ilV/SvwDPCh/vKhqgF/wI+BRpwAlgBJXvqVQBvwb0AsEA98F3gdSAdygbeByn7KVuAlb/t4YJKXFuW3zSvAx7z/7wZagY/jbv5PAVWADORYetn/g8CqfpbPsKcXW/7aR7md+SYBFZ6tc4CdwHuBA952EcBmnBjHAFOAfcC13voCYCkQ5ZW1A/jHHufvWSAV8AFHgWX9HO+/ABu8/98CPuf9f4Vnp3jLacBpYOIAbHzQuya3eNvG92c3kAGcBN7vrX/Ay995Xm8G9gCzvfVfBf4e4DxHAQJchrtXr6b3+/MBYB2Q46X9DHjMKysPqMfVJqOBscBCb92vgG96/38H91KJ9n7v8jtvB4D3ev8/5O0rExgH/B34Ro9n5yGvjOWe3Wl9HOf1OPES3EdMI5A/kLKANbgXWiIwD/eFH+i+jepl3Wu4d0Ec7uV4FLjKW/cG8CHv/yRgqff/J4DfAwm4Z6AAGOOte9I7/4neOdoAfMJb9xjwFdz9FAdcHuAdMs37Pxp373wZd69e5V3TmX7XsQ53n0QAcb2U9wrd9+JY4GXgaW85Efec3IO75xYBx4A5Pe8Tv/L874kHgSbvGkXi7qV1ftt+kO5n7g6gAZgwkHdfgPN5d2/XHPgD3c9xl+1eGf8LvDnA91Rf13+Sd31+gXv+FgDNwGxvfX/PY2defy24ElfTO4LzqP1/QKK3bhHQ2OP4Pg/8vl8NOAexiAQux72QonsYFOe33T78XsK4GkcgEbyqv4eQs4Vnj9+6BG/78QM9lh77f5DzK4JRuIfoWtwHwlc4UwQLgfIeeb8EPNpHuf8IPNnj/F3ut/w48MV+jnc33WL0JbpvcgHKgSu85Y8Dfx6Ijd45ey3Aee6yG1cbfsNvneBeLJ3n9QXgo37rI3Av9bx+znMtcAIntp/p5/7cAVzttzwB92KJ8o7pyT7s/xXdL4iHgKfxXrw9tjtA9wtvL7Dcb921ftf9StxHhv99Xo338hjAffsU8ECgsnDPbSswy2/dtzlHEcR90LYDyX5p3wF+5f3/GvB1IKNHvntx4j+/R3oW7mUY75d2F/AX7/9fAz8HcgZwLvxF8F3AYSDCb/1jwIN+1/HXAcp7xbvf6ryytwA+b90dwOs9tv8Z8LWe90kf98SDwMt+6+YAp/uxZQtws/f/3fTx7hvA+by75zX38h8HYv1sb8I9Sx04kZk/wHdAX9e/837K8UvbANw5gOexM6//fT3eO2cRwGRvvz/zv/Y99v9x4JX+rveAo0NVtV1d9TIH9wXSyVFVbfJbnoh7oXVSNoDiKwJvcgaH/exq9P5N6rmRiBRJd+PpC+e4j1Dya9xNeBfu68qfPGCi57qp9VwVX8bd1IhrsH5WRA6LczV+G1eT8uew3/+N9HIuvLIuw904a7yk1cBFIrJQ3R2zxrMRXNtl8UBs9DjjGgaw+4x7xNu3fwRxHvCw375qcEKZ3dtxeWSoapqqzlbV/98vvef9mQc86Vf2DtzLPQv3ot/bzz46+R6utvFHca7JvgJxJnLm/V/mpXVyXM9si+7v2l0nIus8F1wtribhfx/0VdY43AvlXJ/JnkwEalS1vkc5ndfko8AMYKfn8rzBS/9f4EVgjTiX8L+LSDTuOkQDh/yuxc9wNRiAf8Jd8w2ey/Dec7CzQlU7+rATBva++YyqpgDzcV6RHC89Dyjs8SwU4V7OA6Xn8xonXnOLiHxYRLb4lT2PM69zX+++QOezN67GeVia/dK+r6qpOAE6Dcz0O+7+3gF9Xf++jrnzPu/veTwLVT2sqttVtUNV9+Pukw94q08BY3pkGYPzBPTJYLpIRHFmm4L2WH8I9zLpxDeAMv3LaPD+JvilncsN1l2oarF2N54OtN2wIRT77sFvce6sfapa3mNdBbBfVVP9fsmqutxb/xOcC3W6uuCkL+NeDoPhI17eLeLayNb7pYP7Yr5NXPtboWf3QGyEs++D/uw+RPdLBRER/2Vvf5/osb94Vf37II65p10VwHU9yo5T1YPeuoDtZapar6qfU9UpuGCxz4rI1b1sWoV7yDvxeWnnhLh2xN/igjOyvJfU8wzsPjiKc5We6zPZkyogXUSSe5RzEEBVd6vqXbiX7r8Ba0UkUVVbVfXrqjoHuBS4AecJqMDVXDL8rsMYVZ3rlXdYVT+uqhNxLtUfi8i0AdqZK14bbU87PXreE32iqm8B3wQ629orgFd73D9JqtpZMRhw2T3xnrtfAPfjotVTcc1JA7nO/Z7PPuxaTi/tgQDee+oB3MdoPAHeAX1d/wHa3dfzOJBzqXTr2DtAlIhM91u/gABBU/2KoIhkisidIpLkNbBei6sp/KmfbI8DXxKRNBHJAf5PoKPwR1WP4m7Yld4+72WQDfmDZAtwhYj4RCQFV+UPClVtwLVN9Na9ZANQLy54I9475nki0hlQlIxrPzslLlz5U72UERARiQNux7mnF/r9/g+wQkSi1IWXHwMeAV5UFxQwEBt7oz+7n8PVQG/xvoA/zZkfGz/F3UNzPdtTRCRU3Vd+CnzLe+EgIuNE5GZvXTHwXhG5XUSixAWULOxZgIjcIC5wQ3Aus3ac+6gnjwFf9faRgWtPWdXLdoGIwbWXHAXaxAVDvG8gGVW1Hfgd8KCIJIjIHM4M1OqLWHFBLXHevXMQ59b8jpc2H/f1vwpARFaKyDivBtZ533SIyHtE5CJxAS8nca6uDlU9BPwR+A8RGSMuQGKqiLzbK++D3vsDnJtb6f0c92Q9rqbxTyISLa5/3o10ez8Gw//gaiY34drfZ4jIh7zyo0XkYnHBJ+DaqqYMcj+JuOM8Ci4YClcTDEig8+nZlSN+wWnAdbhnsa8yX8J9VNxHgHdAX9d/AKb39zwe9croOp/e/ZQnjlxcE9PTnr0NuHv9IRFJFOf5upmzvW9nEKgmqLiXVyXuRvw+rj3pmX7yfB3nftiPuyj9GtAHHwe+gPNXz8U9fBcE78L/BtiKawh+NkTlblLVs1xt3kvqBpwg7adbhFK8TT6Pc03W474Sz4qSGyC34Nwbv/a+sg+r6mHgl7jafWeXgtW4NsvV52Bjb/Rpt6oewwUA/DvuGs8BNuG+ZFHVJ3Ffk2vEuVLfxj2woeBhXMTYH0WkHtcoX+jttxz3dfw5nAt2C+5LsifTce28p3ABAT9W1b/0st03vePaigtCKvHSzgnPBfkZ3AfmCdx57e8Z7Mn9OPfTYVy7z6MDyHMKd790/q7CfQBPwr0Yn8S1g73sbb8M2CYuWvNhXJvPadzHzVqcAO4AXqX7nfBhnMBv945rLa5NCFxU+XqvvGdw7Z/7Ahmtqi040bsOd5/+GPiwqu4cwDH3V+bDuKjDetwHyJ3eeThMd+AVuC4Fczz33lPnuJ/twH/g7qkjwEXA386hiP7O559xNaLD4iI05wGnevFM9eR7OJdjFP2/A/q6/oHo73lsBL4F/M07n0txwS9/x3ns/o57rj7jV94/4AJwqnEfoZ9S1X5rgp1RRYYRNjzXVSVQ1IeYGIYRQkTkn3Cu038Kty3hxsYONcKCiFwrIqlee1dne+G6MJtlGKOFAwzMIzDisRE0jHBxCc7l2um+uWWA7hPDMIJEVR8Ptw1DBXOHGoZhGKMWc4cahmEYoxYTQcMwDGPUMmraBDMyMnTSpEnhNsMwDGPYsHnz5mOqOi7cdpxPRo0ITpo0iU2bNoXbDMMwjGGDiAxmiL1hhblDDcMwjFGLiaBhGIYxajERNAzDMEYtJoKGYRjGqMVE0DAMwxi1mAgahmEYoxYTwQC8+s5R6k63htsMwzAM4zwwavoJDoa6xlY+/j+biIiAmxZMZOXSPObnpIbbLMMwDCNEmAj2Q0pCNL/7h0spXl/GU6VVPL6pkvk5KRQV+rhxwUQSYuz0GYZhDGdGzSwSixcv1mBGjDnZ1MpTpQdZta6Md46cIjkuig/k51BU6GN6VnIILTUMwxgaiMhmVV0cbjvOJyaC54iqsqnsBKvWlfHCW4dpae+gcHI6RUvzWDZ3PDFR1sxqGMbIwERwBBEqEfTn+Klmnthcyer15ZTXNJKRFMMHF+eyYomP3PSEkO7LMAzjQmMiOII4HyLYSUeH8vqeY6xaV8afdhxBgStnjKOoMI/3zMokMkLOy34NwzDOJyaCI4jzKYL+VNWeZs3GCtZsKKe6vpmJKXHctcTHHUtyyUyOO+/7NwzDCBUmgiOICyWCnbS2d/CnHUdYta6cv+45RlSEcO3c8RQV+rhk6lhErHZoGMbQZjSIoMX4nyeiIyNYNm8Cy+ZNYN/RUzy2oZwnNlfy3FuHmDIukaLCPG7LzyElITrcphqGYYxawhbKKCLLRGSXiOwRkS/2sc3tIrJdRLaJyGq/9I+IyG7v95ELZ/XgmDIuia9cP4d1X7qaH9y+gNT4aL7x7HaWfPtlPv/Em5SWn2C01MgNwzCGEmFxh4pIJPAOcA1QCWwE7lLV7X7bTAceB65S1RMikqmq1SKSDmwCFgMKbAYKVPVEf/u80O7QQGyvOul1wj9IQ0s7cyeOYeXSPG5aMJHEWKugG4YRfkaDOzRcNcElwB5V3aeqLcAa4OYe23wc+FGnuKlqtZd+LfCSqtZ4614Cll0gu0PGnIlj+NatF7H+K+/lG7fMo71D+dLv3mLpt//Evz79NrsO14fbRMMwjBFPuKoc2UCF33IlUNhjmxkAIvI3IBJ4UFX/0Efe7N52IiL3AfcB+Hy+kBgeapJio/jQ0jxWFvooKT9B8bpy1mys4NdvlHHxpDRWLs1j2bzxxEZFhttUwzCMEcdQ9rtFAdOBK4Ec4DURuehcClDVnwM/B+cODbWBoUREKMhLpyAvna/eMIe1mysoXl/OA2u2kJ4YwwcX57BiiY+8sYnhNtUwDGPEEC4RPAjk+i3neGn+VALrVbUV2C8i7+BE8SBOGP3zvnLeLA0D6Ykx3HfFVD52+RT+tvcYxevKeeT1/fzs1X1cMWMcKwt9XDUrk6hIG6LNMAwjGMIVGBOFC4y5GidqG4EVqrrNb5tluGCZj4hIBlAKLKQ7GCbf27QEFxhT098+h1pgzLlyuK6JNRvLWbOhgsMnm5iQEsedF/u4c0kuWWOsE75hGKFnNATGhK2zvIgsB36Ia+/7pap+S0QeAjap6jPiepP/By7opR34lqqu8fLeC3zZK+pbqvpooP0NdxHspK29gz/trKZ4fTmvvXOUyAjhmtlZrFyax6VTxxJhQ7QZhhEiTARHECNFBP0pO97A6vXlPL6pghONrUzOSGTFEh+3FeSQlhgTbvMMwxjmmAiOIEaiCHbS1NrOH94+TPH6MjYeOEFMVAQ3XDSBoqV55PtSbYg2wzAGhYlgoMwimcBlwETgNPA2zp3ZERrzQsdIFkF/dh4+SfG6cp4sPcip5jZmTxhDUaGPWxZlk2Sd8A3DOAdMBPvKJPIe4ItAOi5gpRqIw/XtmwqsBf5DVU+GztTgGC0i2ElDcxtPb6li1boyth86SWJMJLcsymbl0jxmTxgTbvMMwxgGmAj2lUnke8B/qmp5L+uigBuASFX9bfAmhobRJoKdqCpbKmpZta6cZ7dW0dzWQb4vlZVL81h+0QTioq0TvmEYvWMiOIIYrSLoT21jC2s3V7J6fTn7jjWQmhDNBwtyWFGYx+QM64RvGMaZmAgGyiySBXwbyFbVZSIyB7hEVf87VAaGChPBblSVN/YeZ9X6Mv647QhtHcrl0zJYudTH1bOziLZO+IZhYCIYOLPIC8CjwFdUdYHnCi1V1XMa3uxCYCLYO9Unm/jNxgoe21BOVV0TWWNiueNiH3ctyWVCSny4zTMMI4yYCAbKLLJRVS8WkVJVXeSlbVHVhSGzMESYCPZPe4fyl53VrFpfxqvvHEWAq71O+O+almGd8A1jFDIaRDDYmPkGERmLG8oMEVkK1AVtlXHBiYwQ3jsni/fOyaKippHVG8p5fGMFL20/gi89gRWFPj5YkMPYpNhwm2oYhhEygq0J5gP/CczD9REcB9ymqltDY17osJrgudPc1s6L246wal0ZG/bXEBMZwfKLxlO0NI/FeWnWCd8wRjijoSYYdHSo1w44ExBglzfrw5DDRDA4dh+pp3h9Ob/dXEl9cxszs5IpWurj1kXZJMdFh9s8wzDOAyaCgTKLvL+X5DrgLb+Z4IcEJoKhobGljd+/WcWqdeW8dbCOhJhIbl6YTVGhj3nZKeE2zzCMEGIiGCizyHPAJcBfvKQrcdMcTQYeUtX/DdbAUGEiGHq2Vtayal0Zz7xZRVNrBwtzXSf8G+ZbJ3zDGAmYCAbKLPIi8GFVPeItZwG/Bu4CXlPVeSGxMgSYCJ4/6hpb+V1pJavWlbH3aAMp8dHcVpDDikIfU8clhds8wzAGiYlgoMwi21V1jt+yANtUdY5/t4mhgIng+UdVWb+/hlXrynhx22Fa25VLp45l5dI8rpljnfANY7gxGkQw2C4Sr4jIs8AT3vIHvLREoDbIso1hhoiwdMpYlk4Zy9H6Zh7fVMHq9eX8Q3EJ45JjufPiXO5c4iM71TrhG4YxNAi2Jig44bvMS/ob8FsdggOSWk0wPLR3KK+9c5RV68r4865qBLhqViZFS/O4Yvo4Iq0TvmEMWUZDTdAG0O4PVfjV9ZA5B2ZeB5PeBVE2Y/tgqTzRyJoNFazZWMGxU83kpMWzotDH7YtzybBO+IYx5DARDJTZdZH4NyAT109QAFXVgBPWicgy4GEgEnhEVb/bY/3dwPeAg17Sf6nqI966duAtL71cVW8KtL9BiWBzPfzuE7D3z9B2GmKSYfp7Yeb17m982rmVZwDQ0tbBH7cfpnhdOW/sO050pLBs3gRWFvpYMjndOuEbxhDBRDBQZpE9wI2quuMc80UC7wDXAJXARuAuVd3ut83dwGJVvb+X/KdU9ZzCDoNyh7Y0wv5XYedz8M4foOEoSCTkXQqzrne1xLRJgyt7lLOn+hSr15ezdnMFJ5vamJ6ZRFGhj1vzc0iJt074hhFOTAQDZRb5m6peFnjLs/JdAjyoqtd6y18CUNXv+G1zN0NFBP3p6ICDm2HXc7DrBTi606VnznViOHM5TFwEERYJeS6cbmnn91urKF5fzpsVtcRHR3LTgomsXJrHRTnWCd8wwoGJYKDMIg8D44GngObOdFX9XYB8twHLVPVj3vKHgEJ/wfNE8DvAUVyt8f+qaoW3rg3YArQB31XVp/rYz33AfQA+n6+grKxscAfaH8f3OjHc9QKU/x20A5LGw8xlzm06+QqIjgv9fkcwbx+so3h9GU+VVnG6tZ35OSmsLMzjxgUTiY+xTviGcaEwEQyUWeTRXpJVVe8NkG8gIjgWOKWqzSLyCeAOVb3KW5etqgdFZArwZ+BqVd3b3z4vSHRoYw3s/iPseh72/AlaTkF0Ikx9j3ObTr8WEseeXxtGECebWnmy5CDF68t458gpkuOi+EB+DiuX+piWmRxu8wxjxGMieL52OgB3aI/tI4EaVT3LLyYivwKeVdW1/e3zgneRaG2CA3/tdpvWHwKJgNxC5zKduRwypl04e4YxqsrGAycoXl/GC28dpqW9g8LJ6axcmse1c8cTE2WuZ8M4H5gIBsosEgd8FJgLdPn8BlATjMK5OK/GRX9uBFao6ja/bSao6iHv/1uBf1bVpSKSBjR6NcQM4A3gZv+gmt4Iaz9BVTi0BXY+7wTxiBfYmjHDa0e8HnIWQ4S5+gJx7FQzT2yqZPWGMipqTpORFMPti3O5a4mP3PSEcJtnGCMKE8FAmUWeAHYCK4CHgCJgh6o+MIC8y4Ef4rpI/FJVvyUiDwGbVPUZEfkOcBOu3a8G+JSq7hSRS4GfAR1ABPBDVf3vQPsbUp3lT5S5KNNdz7vaYkcbJGR47YjLYcp7IMZe6P3R0aG8tvsoq9aV8+edR1DgyhnjWLk0jytnZlonfMMIASaCgTJ744OKyFZVnS8i0cDrqro0dCaGhiElgv6croU9LztB3P0SNJ+EqDgnhDOvgxnLIDkr3FYOaapqT7NmQzlrNlZQXd9Mdmo8dy3J5faLc8lMtqAkwxgsJoKBMotsUNUlIvIa8A/AYWCDqk4JlYGhYsiKoD9tLVD2Ny/a9HmoqwDEuUo73abjZoJ1Ju+V1vYOXt5+hFXry/jbnuNERQjXzh1P0VIfl0wZa53wDeMcMREMlFnkY8BvgYuAXwFJwL+o6s9CYl0IGRYi6I8qHHnbCeLO51ybIkDaZOcynbUccpdCZLBjoI9M9h11nfCf2FxJ3elWpoxLpKgwj9vyc0hJsE74hjEQTAT7yygSAdymqo+H1qTzw7ATwZ7UHexuR9z/GrS3uGHbpl/raonTroZY6zbQk6bWdp7beohV68soLa8lNiqCG71O+AtyUqx2aBj9YCIYKLPIpuFygoa9CPrTXO/6Ie56AXa/CKdPQGSM65jfOWrNmInhtnLIsa2qjuL15TxVepDGlnbmZY+hqDCPmxdOJCHGatSG0RMTwUCZRb4LHAN+AzR0pqtqTfCmhZYRJYL+tLdBxbput+mJ/S59wsLucU2z5lk7oh/1Ta08taWK4nVl7DxcT3JsFLfmZ7NyaR4zsqw2bRidmAgGyiyyv5dktcCYMKEKR3c5l+mu56FyE6CQ4vNqiNfBpMsh0trEwHXCLyk/wap15Tz31iFa2jpYMimdoqU+ls0bT2yU9ds0RjcmgiOIUSGCPak/4rUjvgD7/gJtTRCb4k0HtRymvRfiU8Nt5ZCgpqGFtZsrKF5fTtnxRtITY/jg4hyKluThG2t9No3RiYlgoMyuX+CngCu8pFeAn6lqa/CmhZZRKYL+tDTAvle8WuIfoPEYRERB3mXObTpjGaTlhdvKsNPRofxt7zFWrSvj5R3VdKhyxfRxFBX6uGpWJlGRNkSbMXowEQyUWeQRIBr4Hy/pQ0B758DYQ4lRL4L+dLQ7V2nnuKbH3nHpWfO8cU2vc9NBjfJ2xMN1TazZWM6aDRUcPtnEhJQ47rzYx51LcskaY53wjZGPiWCgzCJvquqCQGlDARPBfji2x6shvuCCbLQDkid0R5pOvgKiYsNtZdhoa+/gTzurWbWujNd3HyMyQnjfnCyKCvO4dOpYImyINmOEYiIYKLNICfDBzmmMvKmN1qpqfojsCxkmggOk4ZjfdFB/htYGiEmCqVd500G9DxLSw21l2Cg73sDq9eU8vqmCE42tTM5IZMUSH7cV5JCWGBNu8wwjpJgIBsoscjXwKLAPECAPuEdV/xIa80KHieAgaG1yHfN3PefaEU8ddtNB+S7pdpuOnRpuK8NCU2s7f3j7MKvWlbGp7AQxURHcMH8CRYV55PtSrRO+MSIwERxIASKxwExvcZeqNve3fbgwEQySjg44VNo9HVS1N+vVuFnd45pmF0DE6Asc2Xn4JMXrynmy9CCnmtuYPWEMK5f6uHlhNkmx1gnfGL6YCI4gTARDzIwXbpkAACAASURBVIkD3QN9H/gbaDskZsKMa53bdPK7R910UKea23hmSxWr1pWx/dBJkmKjuGXRRIoK85g9YUy4zTOMc8ZEcARhIngeOX0Cdr/s3Ka7X4aWeoiKd+2IndNBJY0Lt5UXDFVlS0Utq9aV8+zWKprbOijIS2PlUh/XzZtAXLR1wjeGByaCIwgTwQtEWwsceN2rJb4AJysBgdwl3dGmGTNGTfeL2sYW1m6uZPX6cvYdayAtIZoPLs5lxRIfkzISw22eYfSLieC5FiYyAagZiu2CJoJhQBUOb+12mx5606WnT3WCOOt6yFkyKqaDUlXe2HucVevL+OO2I7R1KO+ankFRYR5Xz84k2jrhG0MQE8FzLUzkZWAq8FtV/XzICg4BJoJDgLrKbkHc/zp0tEJ8umtHnLncuU9jk8Jt5Xmn+mQTv9lYwWMbyqmqayI+OpL5OSnk56WR70tjkS+VjKTR2y/TGDqYCA6mQBcbPkdVtwXYbhnwMBAJPKKq3+2x/m7ge8BBL+m/VPURb91HgK966d9U1f8hACaCQ4ymk7D3Ty7adPeL0FQHkbEw5d1eO+J1MGZCuK08r7S1d/DqO0d5ffcxSstPsK3qJG0d7nn0pSeQ70tlkc8J46wJyVZbNC44JoIDKUDkcmC6qj4qIhlAsqr2NruEf55I4B3gGqAS2Ajcparb/ba5G1isqvf3yJsObAIWAwpsBgpU9UR/+zQRHMK0t0L5G93TQdWWufSJ+TBruaslZs4Z8e2ITa3tvH2wjpLyE5SU1VJSfoLqeteyEBcdwfzsVBblpXbVFjOTbeg24/xiIhgos8jXcGI0U1VniMhE4AlVvSxAvkuAB1X1Wm/5SwCq+h2/be6mdxG8C7hSVT/hLf8MeEVVH+tvnyaCwwRVqN7RPR3Uwc0uPTWvu4N+3qWjYjooVaWqromSshOUljtR3FZVR2u7e2Zz0uK7BDHfl8bsCWOIibLaohE6RoMIBhuRcCuwCCgBUNUqERnIrKTZQIXfciVQ2Mt2HxCRK3C1xv+rqhV95M0ehO3GUEQEsua43xWfh/rD3ZGmm34J638CcSlu+LaZ17npoOJSwm31eUFEyE6NJzs1nhsXTARcbXFb1UlKy09QUn6CjQdqeObNKgBioyK4KLuzbdG5Um2gb8Pon2BFsEVVVUQUQERCGfP9e+AxVW0WkU/gZqq46lwKEJH7gPsAfD5fCE0zLhjJ42HxPe7X0gB7/+wE8Z0/wFtPQES0myi4s5aYmhtui88rcdGRFOSlUZCX1pV2qO40JWW1XcL4q78d4OevdQCQnRrPoq62xVTmTkyx2qJh+BGsO/TzwHRc2953gHuB1ar6nwHyBXSH9tg+Etf1IsXcoQbgpoOq2NDtNj2+x6WPv8gN4TbzOpiwYMS3I/ZGc1s726tOUuK5ULeU13Kw9jQAMVERzJs4hnxfGvl5zpU6ISU+zBYbQ5XR4A4NRWDMNcD7cANov6iqLw0gTxTOxXk1LvpzI7DCP6JURCao6iHv/1uBf1bVpV5gzGagc6aKElxgTE1/+zQRHOEc2+2Cana9ABXrAYUx2V4H/etg0rtG9XRQR056bYsVtZSUnWDrwTpa2lxtcUJKXFfb4iJfGvOyxxAbZaPaGCaC53fHIsuBH+K6SPxSVb8lIg8Bm1T1GRH5DnAT0AbUAJ9S1Z1e3nuBL3tFfUtVHw20PxPBUcSpo67bxa4XnPu0tRFikmHa1c5tOv2aUT0dFEBLWwc7Dp10kajlThi7aouREczpqi06YZyYEmczY4xCTAQDZRapx3VTAIjBzTLfoKpDbrRgE8FRSutp2Pdq96TBDdUgkS7CtLMdMX1yuK0cElSfbKKkvJbSihOUltWy9WAtTa2utpg1JvaMSNR52Sk2BuoowETwXApyn4k3A0tV9YshKTSEmAgadHRAVUm32/ToDpeeOad7XNOJ+aNyOqjeaG3vYOeheq+26H4VNa62GB0pzJkwxgXc5KWxKDeVnLR4qy2OMEwEB1OgSKmqLgppoSHARNA4i5p93d0vyv7upoNKynKzXsy6HiZfAdEWNOLP0fpmSsv92hYr6zjd2g7AuOTYM0a5mZ9jtcXhjolgoMwi7/dbjMB1nH+3ql4SrGGhxkTQ6JfGGtj9kpsOas+foOUURCd400Etd+ObJmaE28ohR1t7BzsP13vdM1w0atnxRgCiIoQ5E8ewKDfVqy2mkZtutcXhhIlgoMwi/gEpbcAB4BeqWh2kXSHHRNAYMG3NbjqonV47Yn0VSATkFvpNBzU93FYOWY6fau4a4aa0vJY3K2tpbHG1xYykGBb5tS3Oz0khIWbkzyIyXDERHEGYCBqDQhUObeme/eLwWy597HS/6aAuhghz+/VFW3sHu47UnyGM+481ABAZIcyekMyiXC8SNTeNvLEJVlscIpgI9pVJ5D/pjgo9C1X9TDBGnQ9MBI2QUFsOu/7gBPHA69DRBgljXTvizOUw9T0QY5PlBuJEQwulFW6g8NIK16G/wastjk2M6eqzuMiXyoKcVBJjrbYYDkwE+8rkpjLqk4FMbXShMRE0Qk5THex52ZsO6iVoroOoOJhypTcd1DI37JsRkPYO5R2/2mJJ+Qn2HXW1xQiBWePHdLlQF/lSmZyRaLXFC4CJ4EALERkDqKrWB2/S+cFE0DivtLe6CNNdzztRrCt36dmLu92m42aNymHcBkttYwulFbWUeiPdbCmvpb65DYC0hGhXU/SCbhbkppJktcWQYyIYKLPIYuBRIBk3bFotcK+qbg6NeaHDRNC4YKjCkW3d7YhVJS49bVL3uKa+SyDSXtrnQnuHsqf6VNdA4SXlteypPgW42uKMrOSugcLz89KYYrXFoDERDJRZZCvwaVV93Vu+HPixqs4PkX0hw0TQCBsnq9ysFzufh/2vQnsLxKW6bhczr4O8yyBxnNUSB0Hd6Va2eH0WSyvcTBr1Ta62mBIffYYLdWFuKslxI38eylBiIhgocy8d40WkRFXz+8oTLkwEjSFBc/2Z00GdPuHS41Jg7DTvNx3GTvX+n2qBNudAR4ey9+ipM9oWd1efQtV9Y8zITO4Sxvy8VKZkJBERYR8ffWEi2FcmkU6R+zAQDzyGixa9A2hS1c+GzMIQYSJoDDna26ByAxx6000FdWw3HN8LJyvP3G5MtieK050wZngimeIzl+oAONnUypsVtV2RqKXltdSdbgVgTFwUC33dkxAvzE0lJd5qi52YCPaVSeQv/axWVT2nyW8vBCaCxrChpcEN6XZ8Dxzb4/4e3wPHd7uI1E4ioiF9SneNMWN6d00yMcPcq33Q0aHsO9bQNcpNafkJdh2pp/NVOD0zya+2mMa0caO3tmgiOIIwETSGParQeNyrMe4581ezz7U1dhKb0kMYp3a7Wc29ehb1Ta1srayjpMy5UEsraqltdLXF5NgoFvr1W8zPTSMlYXTUFk0E+8okshIo1j4yi8hUYIKq/jVI+0KGiaAxoulodx35j+91NcZOcTy2px/36rRuF+vYqZCaZ+5VD1Vl/7GGrppiSXktuw6fpMN7400dl9g1UHh+XirTM5OJHIG1RRPBvjKJPADci5vhfTNwFIgDpgHvBo4BX1TV3aEzNThMBI1RS0uj517d3cPF2pt7dfKZgTmdNUmLXqWhuY03K2td0I0XjVrT4GrfSbFRLMhN6YpEXZSbRlpiTJgtDh4Twf4yikQCVwGXAROA08AO4AVVLQ+ZhSHCRNAwetDpXu2qNe4O7F7tjGDNmNb9/yh1r6oqZccbu8ZDLSk/wc7D9bR71cUpGYks7Gxb9KUxIyuJqMjhNVelieAIwkTQMM6Bjnaoqzg7MOf4XpfuT/LEHu2P3m8UulcbW9pc22K5Ny5q+QmOe7XFhJhIFuSkdg0UvsiXytik2DBb3D8mgudzxyLLgIeBSOARVf1uH9t9AFgLXKyqm0RkEq7GucvbZJ2qfjLQ/kwEDSNEdLlX/YSxsybZVNu9XZd7ddrZXTxGiXtVVamoOe3VFl3b4vZDJ7tqi5PGJnSNcrPIl8as8clDqrZoIni+dupcqe8A1wCVwEbgLlXd3mO7ZOA5IAa4308En1XVeeeyTxNBwzjPqLrJibvE0a/vY80+aG/u3jZ2zJnC2FmTTJ8KsUnhO4YLwOmWdt462FlbdMJ47JQ7N/HRkczPSfEmIXbDv2WEsbY4GkQwXL6KJcAeVd0HICJrgJuB7T22+wbwb8AXLqx5hmGcMyKQONb9fIVnrut0rx7f40Sxs/2x/A146/Ezt02e4Nf2OH3EuVfjYyJZMjmdJZPTAVdbrDxxuqttsbT8BL94bR9tXm3Rl57Q3W/Rl8asCclED6Ha4nAnqDvKixJ9FKgHHgEW4aJC/xggazbg37BQCZzx1Hij0uSq6nMi0lMEJ4tIKXAS+Grn2KWGYQxRIiLdAOJpk2Dae89c13ra1RS7AnO8bh7bnuzhXo2CtMk9AnM8kUzKHLbuVREhNz2B3PQEbl6YDUBTaztvH+xuW3xj73Ge3lIFQFx0BPOzU1nktS3m56WSmRwXzkMY1gT7WXWvqj4sItcCacCHgP8FAolgv4hIBPAD4O5eVh8CfKp6XEQKgKdEZK6qnuylnPuA+wB8Pl8wJhmGcb6Ijoesue7Xk8YaP3H0E8m9f+7DvTrt7N8wdK/GRUeyeFI6iyd11xar6ppc1wwvEvWXf91Pa/s+AHLS4rtn0PClMXvCGGKirLY4EIKeRUJV54vIw8Arqvpkb4Nq95LvEuBBVb3WW/4SgKp+x1tOAfYCp7ws44Ea4CZV3dSjrFeAz/dM74m1CRrGCKKjHeoqzw7M6Ype9Xuv+btX/d2sqT6IHL4jvzS1trOt6mT31FJltRw+2QRAbEwLU7KPMia9krTEDn5y3TcGNa2UtQkGZrOI/BGYDHzJC2TpGEC+jcB0EZkMHATuBFZ0rlTVOiCjc9lf6ERkHFCjqu0iMgWYDuwL8jgMwxhORERCWp779eVe9RfG43tg+1Pds3aA516d1D04gH/7Y1LWkHevxkVHUpCXRkFeGsdOH6PkyHFeryhlw6HNHDq9l0oUbYggqnZyuE0d0gQrgh8FFgL7VLVRRNKBewJlUtU2EbkfeBHXReKXqrpNRB4CNqnqM/1kvwJ4SERacYL7SVWtCfI4DMMYKQRyr54xMMDu3t2rMclnj5rTGcUam3zhjqUXVJWK+go2H9lMSXUJJUdKKK9345PER8UzP2M+N8+4hvysfGalzaOuQWxy4X4I1h16GbBFVRu88UTzgYdVtSxUBoYKc4cahtEnHR1nRq/6j79a28O9mjS+ezqrsX4CmZZ3Xtyr7R3t7K7d7UTvSAkl1SUcO30MgNTYVBZlLiI/M5/8rHxmj51NdETobDB3aGB+AiwQkQXA53ARor/GjR9qGIYxPIiI8HOvXn3mutbTULP/zMCcY7th+zNw2s8J1eVenXZ2F49zcK82tzfz9rG3KTlSwubqzbxZ/SanWl14xITECRROKCQ/M5+CrAImp0wmQiwAJhiCFcE2VVURuRn4L1X9bxH5aCgMMwzDGBJEx0PWHPfrSdfgAHvO7OKx7xVoa+rezt+96j8x8thp1Atsqd7S5dp869hbtHa4aZympU7jusnXkZ+VT0FmAROSJlyYYx5FBCuC9V5k54eAd3ldG4ZvuJVhGMa5kJAOCUsgd8mZ6R0dbgqrnhMjV27g6I4n2RwXS0lsLCVxsbwTE42KEIUwJ24cRVmXkJ99GYsmX0NqwrjwHNcoIlgRvAMX1Xmvqh4WER/wveDNMgzDGMZERECqD03JpSxjCqXV49g8JoaSxCYq6t0m8RExLIjN4FMdUeQ3nuKiYxUkNJYBm4BVIN4AAz0HJh87DZLHD/no1eFCUCLoCV8xcLGI3ABsUNVfh8Y0wzCM4UV7Rzu7TuzqCmApOVLC8abjAKTFprEocxF3zLyDgqwCZqbPPDuIpbGm94mRz3KvJp09MXLGNDf2atyYC3fAI4Bgh027HVfzewUQ4D9F5AuqujYEthmGYQxpmtqaXBCLJ3hbjm6hobUBgOykbC6deCn5WfnkZ+YzOWVy4K4KCenul3vxmen+7lX/sVcrN8Hbv+PM6NWsMydGHjcTZlwb2gMfQQTrDv0KboqjagCvI/vLuKmPDMMwRhQnW06ypXpLV3eFbce3nRHEcsOUG7q6K4xPHB+6HXvuVVJ9MPWqM9e1NsGJ/WePvbrzWTdpcorPRLAfghXBiE4B9DgOWLyuYRgjgiMNRyitLu3qmL77xG4UJSoiirlj57JyzkoKMgtYmLmQlNiU8BgZHQeZs92vJ4010HD0wts0jAhWBP8gIi8Cj3nLdwAvBFmmYRjGBUdVOXDyQFd73uYjmzl46iAACVEJLBi3gGsWXkNBVgHzMuYRHxUfZosHQKd71eiTYANjviAi7wcu95J+rqpPBm+WYRjG+aWto41dNbu62vNKqkuoaXKd39Pj0snPzKdodhH5mfnMTJ9JVMTwn8vQOJugr6qq/g74XeeyiJSrqs1bZBjGkKKprYm3jr3V1Z735tE3aWxrBFwQy+XZl3e1500aM8nG2xwlnI9PG7tzDMMIO3XNdZRWl3bV9LYd30ZbRxuCMD1tOjdNvYmCrAIWZS4iKzEr3OYaYeJ8iODgR+Q2DMMYJIcbDp/Rnrendg8AURFRzBs7jw/P+TAFWQUsGLcgfEEsxpBjUCIoIp/taxUw/KZxNgxjWKGq7K/bf0Z7XmcQS2J0IgvHLXRjbmbmMy9jHnFRcWG22BiqDLYm2N+EWg8PskzDMIxeaetoY2fNzq72vNLqUk40uwly0+PSKcgqYOXsleRn5TMjbYYFsRgDZlB3iqp+PdSGGIZhdHK67TRbj27tqum9efRNTredBiA3OZcrcq6gIKuA/Kx8fMk+C2IxBo19LhmGEXbqmuvOGG9z+/HttKkLYpmRNoNbpt3SNfxYZkJmuM01RhAmgoZhXHAOnTrE5urNlB5x0ZudQSzREdFclHERd8+7m/zMfBZkLmBMjA0IbZw/wiaCIrIM134YCTyiqt/tY7sP4MYivVhVN3lpXwI+CrQDn1HVFy+M1YZhnCuqyr66fV1Dj5UcKeFQwyEAkqKTWJC5gOWTl5Of5YJYYiNjw2yxMZoIqQh6M8wfVtX1AbaLBH4EXANUAhtF5BlV3d5ju2TgAWC9X9oc4E5gLjAReFlEZqhqeyiPxTCMwdHa0cqO4zu6xtwsrS6ltrkWgIz4DPIz8/nI3I9QkFXA9NTpREZEhtliYzQT6ppgIXCRiESp6nX9bLcE2KOq+wBEZA1wM7C9x3bfAP4N+IJf2s3AGlVtBvaLyB6vvDdCdAyGYZwDja2NbD221bXpHSlh67GtXUEsvmQfV+ZeSX5mPgVZBeQm51oQizGkCKkIquqXB7hpNlDht1yJE9AuRCQfyFXV50TkCz3yruuRN3sQ5hqGMQhONJ2gpLqkqz1vx/EdtGkbERLBzLSZvH/6+7uGH8uIzwi3uYbRL8FOqpsJXIZzS54G3gY2qWpHkOVGAD8A7g6ynPuA+wB8PhvO1DAGQ9WpqjPa8/bV7QMgJiKGeRnzuGfePeRn5bNw3EKSYmysDGN4MdgRY94DfBFIB0qBaiAOuAWYKiJrgf9Q1ZN9FHEQyPVbzvHSOkkG5gGveK6T8cAzInLTAPJ2oao/B34OsHjxYhvOzTAC0KEd7K3d291dobqEww2HAUiOTmZh5kJunHojBVkFzB07l5jImDBbbBjBMdia4HLg46pa3nOFiEQBN+CCXn7bR/6NwHQRmYwTsDuBFZ0rVbUO6PKjiMgrwOdVdZOInAZWi8gPcDXQ6cCGQR6HYYxqWttb2V6zvas9r/RoKXXNdQCMix9HflY+98y9h4KsAqalTrMgFmPEMdgRY77Qz7o24KkA+dtE5H7gRVwXiV+q6jYReQjnTn2mn7zbRORxXBBNG/Bpiww1jIHR2NrIlqNbuoYe23p0K03tTQBMGjOJq31Xd7Xn5STlWBCLMeIR1cF7CUUkFvgAMAk/QVXVh4K2LMQsXrxYN23aFG4zDOOCUtNUQ+mRUjZXuzE3d9bspF3bu4JYOoceW5S5yIJYjLMQkc2qujjcdpxPgo0OfRqoAzYDzcGbYxjGYFFVDp46eMbMCvvr9gMQGxnLRRkX8dGLPkpBZgELMheQGJ0YZosNI/wEK4I5qrosJJYYhnFOdGgHe2r3dLXnba7eTHVjNQDJMcksylzEzVNvpiCrgDlj51gQi2H0QrAi+HcRuUhV3wqJNYZh9Elreyvbjm/rGoWltLqUky0uADszIZOCTOfazM/KZ1rqNCIkIswWG8bQJ1gRvBy4W0T249yhAqiqzg/aMsMY5TS0NvBm9Ztd7XlvHXuL5nbX6jA5ZTLX5F3TNbNCdlK2BbEYxiAIVgT7GxrNMIxz4Pjp42e05+2s2UmHdhApkcxKn8XtM2+nILOARVmLSI9LD7e5hjEiCEoEVbVMRC4HpqvqoyIyDrAhIwxjALR1tPFKxSu8fvB1So6UcODkAQDiIuOYP24+982/j0WZi1g4biEJ0QnhNdYwRijBDpv2NWAxMBN4FIgGVuGGUjMMoxdqm2pZu3sta3au4UjjEcbEjCE/M9+NuZmVz5z0OURHRofbTMMYFQTrDr0VWASUAKhqlTf9kWEYPdh9YjfFO4p5bt9zNLU3UTi+kK8UfoUrcq6wkVgMI0wEK4ItqqoiogAiYh2PDMOP9o52Xqt8jeIdxaw/vJ7YyFhumHIDK2avYEbajHCbZxijnmBF8HER+RmQKiIfB+4FfhG8WYYxvKlvqeepPU+xesdqKk9VkpWQxQP5D3Db9NtIjUsNt3mGYXgEGxjzfRG5BjiJaxf8V1V9KSSWGcYwpOxkGat3rOapPU/R2NbIosxF/GPBP3KV7yqiI6ydzzCGGqGYVPcdXN/Al0UkQUSSVbU+BOUaxrBAVXmj6g1W7VjF6wdfJyoiiusmXUfRnCLmjp0bbvMMw+iHYKNDP46btDYdmIqb4f2nwNXBm2YYQ5vG1kae3fcsxTuK2Ve3j/S4dD614FPcPvN2G4zaMIYJwdYEPw0sAdYDqOpub7Z5wxixVJ2q4rGdj/Hb3b+lvqWeOWPn8O3Lv821k6618TkNY5gRrAg2q2pL53BN3oS6NoO7MeJQVTYf2UzxjmL+XPFnBOFq39WsnLOSheMW2pBlhjFMCVYEXxWRLwPxXoDMPwC/D94swxgaNLc38/y+51m9czU7a3aSEpvCPXPv4c5ZdzI+cXy4zTMMI0iCFcF/Bj4GvAV8AngeeCRYowwj3FQ3VvObXb9h7TtrqWmqYVrqNL52yde4fsr1xEfFh9s8wzBCxKBFUEQigW2qOgvrG2iMELYe3cqqHat46cBLtGs77855N0VziigcX2guT8MYgQxaBFW1XUR2iYhPVcvPNb+ILAMeBiKBR1T1uz3WfxIXeNMOnALuU9XtIjIJ2AHs8jZdp6qfHOxxGEZrRysvHXiJ4h3FbD22laToJO6cdScrZq0gd0xuuM0zDOM8Eqw7NA3YJiIbgIbORFW9qb9MXi3yR8A1QCWwUUSeUdXtfputVtWfetvfBPwA6JzFfq+qLgzSdmOUU9NUwxO7nuDxXY9TfbqavDF5fHHJF7ll2i0kRtsIgIYxGghWBP9lkPmWAHtUdR+AiKwBbga6RFBVT/ptn4hFnRohYlfNLlbtWMXz+56npaOFSydeytcu/RqXZ19us7Ebxigj2GHTXh1k1mygwm+5EijsuZGIfBr4LBADXOW3arKIlOKGa/uqqr4+SDuMUUJ7Rzt/qfgLxTuK2XRkE/FR8dwy7RaKZhcxJXVKuM0zDCNMBDtiTD1n19DqgE3A5zpreoNFVX8E/EhEVgBfBT4CHAJ8qnpcRAqAp0Rkbo+aY6d99+FGtMHn8wVjijFMqWuu48ndT/LYzseoaqhiYuJEPlfwOW6dfispsSnhNs8wjDATrDv0h7ha3GpAgDtxw6eVAL8Eruwj30HAP+Igx0vrizXATwBUtRlo9v7fLCJ7gRk44T0DVf058HOAxYsXmzt1FLGvbh+rd6zmmb3PcLrtNAVZBXzh4i9wZe6VREWEYshcwzBGAsG+DW5S1QV+yz8XkS2q+s9eJ/q+2AhMF5HJOPG7E1jhv4GITFfV3d7i9cBuL30cUONFp04BpgNB1TiNkUGHdvDXg3+leEcxf6/6OzERMSyfspyi2UXMSp8VbvMMwxiCBCuCjSJyO7DWW74NaPL+77PmpaptInI/8CKui8QvVXWbiDwEbFLVZ4D7ReS9QCtwAucKBbgCeEhEWoEO4JOqWhPkcRjDmIbWBp7e8zSP7XyMAycPMC5+HPcvvJ/bZtzG2Pix4TbPMIwhjKgO3kvo1cQeBi7Bid464P/iancFqvrXUBgZChYvXqybNp3lMTWGMRX1FV1z951qPcVFGRdRNLuI9+W9j+hIm7vPMIJFRDar6uJw23E+CTY6dB9wYx+rh4wAGiMHVWXD4Q2s2rGKVyteJVIiuWbSNaycvZL54+aH2zzDMIYZgxJBEfkq8OO+3JAichWQoKrPBmOcYXTS1NbUNXffnto9pMWm8bGLPsYdM+8gKzEr3OYZhjFMGWxN8C3g9yLShIsEPQrE4YJUFgIvA98OiYXGqOZww2HW7FzD2t1rqWuuY2baTB669CGWT1lObGRsuM0zDGOYMygRVNWngadFZDpwGTAB13F9FW6Mz9OhM9EYbagqW45uoXhHMS+XvYyivCf3PRTNLmJx1mIbyNowjJARbJvgbryuC4YRLC3tLbx44EVW7VjF9uPbSY5J5kNzPsSds+4kOyk73OYZhjECsV7DRtg5dvoYT+x6gt/s+g3Hm44zOWUyXy38KjdOvZGE6IRwm2cYxgjGRNAIG9uOb6N4ezF/OPAHWjtaeVf2uyiaXcQlEy+xxvkFJwAAHLFJREFUgawNw7ggBDt2aIaqHguVMcbIp62jjT+V/4niHcWUVpeSEJXAbTNuY8WsFUxKmRRu8wzDGGUMtovEjbixQdtEpB24XVX/HlLLjBFFbVMta3evZc3ONRxpPEJ2UjZfWPwFbp1+K8kxyeE2zzCMUcpga4LfAt6lqjtFpBD4d+DdoTPLGCnsPrGb4h3FPLfvOZramygcX8hXCr/CFTlXEBkRGW7zDMMY5QxWBNtUdSeAqq4XEfuUN7po72jntcrXKN5RzPrD64mNjOWGKTewYvYKZqTNCLd5hmEYXQxWBDNF5LN9LavqD4IzyxiO1LfU89Sep1i9YzWVpyrJSsjigfwHuG36baTGpYbbPMMwjLMYrAj+AkjuY9nm7RtllJ0s6xrIurGtkYXjFvJAwQNc7bua6AgbyNowjKHLYEeM+Xpf60Tk4sGbYwwXVJU3qt5g1Y5VvH7wdaIiorhu0nUUzS5ibsbccJtnGIYxIELST1BE5gB3eb9aYERPvTGaaWxt7BrIel/dPtLj0vnUgk9x+8zbyYjPCLd5hmEY58SgRVBEJtEtfK1AHrBYVQ+EwjBjaFF1qorHdj7G/2vv3qOsqM68j39/NHcwiIoEUQQUIhojKoNRUTEmETUqKhkRyJKok8toxowzzpuMJia+ycSMM5lcNKPG5asOrSCoCVESxLuOF0RFEAEhiBGj4hUFBbk87x+1G4pOd3P6cvrQ1O+z1lldtev21D7V5zl7V52q25fczgcff8CQXYbw4xE/ZlT/UXSs6ljp8MzMmqSpvxN8HPgEMBk4IyKWSHrJCXDHEhE8/cbTVC+s5v5X7keI4/odx4T9JzC011DfyNrM2rymtgTfAPoCvYFeZDfR9gUxO4h1G9cxY9kMbll0C4veWUSPTj2YeMBExn5qLH2696l0eGZmLaapF8aMltQDOB34QXqk0s6ShkfE7FLWIWkU8AugCrg+Iq6oNf0bwPnARmA12SOaXkjTvgucm6b9Q0TMbMp+2NZWfriSKYunMO3Fabyz9h323XlfLjv8Mk4aeBJd2nepdHhmZi1OEc1vwEnqDfwtMBboFxF7bWP+KuBF4AvACuAp4KyaJJfm+UREvJ+GTwH+PiJGpYtwbgWGA3uQPcB3cERsbGibw4YNizlz5jR1F3do896cx6SFk5i1fBYbYyPH7HkM4/cfz2GfPMxdnmYFJunpiNihL3RskatDI+IN4FfAryTtXcIiw4GlEbEMQNJk4FRgcxKsSYBJN7Z0t54KTI6IdcBLkpam9T3e7B0pkPWb1jNr+SyqF1Yz7615dOvQjbH7jWXcfuPY6xMNfocxM9thNPcpEsOAS8iuDM2v6zPbWLQv8EpufAVwWB3rPx+4COgIfC637BO1lvUTV0v0ztp3mLp4Krctvo2VH62k3079+M7w7zB639F069Ct0uGZmbWq5rYEq4GLgfnApuaHs7WIuBq4WtI44FLg7MYsL+lrwNcA+vXr19LhtSmL31nMpIWTmLFsBh9v+pgj9jiCy464jBF9R/jZfWZWWM1Ngm9GxPQmLPcqkO9z2zOV1Wcy8N+NXTYirgOug+ycYBPibNM2btrIA688QPXCaua8MYcu7bswet/RjBsyjn123qfS4ZmZVVxzk+Blkq4H7gPW1RRGxB3bWO4pYJCkAWQJbCwwLj+DpEERsSSNnkT2MwyA6cAtkn5GdmHMIKCkK1KLYtW6Vdy55E5uXXQrf1nzF/p068NFh17E6YNOp0enHpUOz8xsu9HcJPhVYD+gA1u6QwNoMAlGxAZJFwAzyX4icUNELJB0OTAntS4vkPR5srvRvEvqCk3z3UZ2Ec0G4PxtXRlaFMtWLeOWhbcw/U/T+WjDRxza+1Au/puLGbnXSNq3a5FroMzMdijN+omEpMUR8akWjKdsdtSfSGyKTTz66qNUL6zmsb88Rod2HThxwImMHzKeIbsOqXR4ZtaG+ScS2/aYpP3zv++z1rFm/Rp+t/R33LroVpa/v5xeXXpxwdALGDN4DLt22bXS4ZmZtQnNTYKfBeZKeonsnKCAiIht/UTCmuiVD17Z/Oy+1etXc+BuB3LFUVfwxb2/SIcqP7vPzKwxmpsER7VIFNagiGD267OZtHASD73yEFWq4gv9v8D4IeM5qNdBlQ7PzKzNalYSjIiXWyoQ+2trN6zd/Oy+pe8tpWennpx34Hmc+akz6d2td6XDMzNr83zJ4Hbo9TWvM3nRZKYtmcaqdasY3HMwlx9xOScMOIHO7TtXOjwzsx2Gk+B2IiKY++ZcqhdWc+/L9xIEx+51LOOHjGdY72G+kbWZWRk4CVbYxxs/ZubymUxaOIkX3n6BnTrsxIQhEzhryFn07e5bopqZlZOTYIW89dFbTF08lSmLp/D22rcZ0GMAlx52KSfvczJdO3StdHhmZoXgJNjKFry9gOoXqvnj8j+yftN6RvQdwYQhEzh8j8N9I2szs1bmJNgKNmzawH1/vo/qhdU8u/JZurTvwpjBYzhrv7MY0GNApcMzMyssJ8Eyem/te0xbMo3Jiybzxodv0Ld7Xy4edjGnDTqNnTruVOnwzMwKz0mwDJa8u4TqhdXcvexu1m5cy/BPDudfD/tXjtnzGKraVVU6PDMzS5wEW8jGTRt5eMXDVC+s5snXn6RTVSe+NPBLjBsyjsE9B1c6PDMzq4OTYDN98PEH/Hbpb7ll4S2sWL2C3bvuzoWHXMgZg86gZ+eelQ7PzMwa4CTYRC+///LmG1l/uOFDhvYayoWHXshx/Y6jQzvfyNrMrC1wEmyEiODxvzzOpIWTeOTVR2jfrj2j+o9iwpAJHLDbAZUOz8zMGslJsAQfrv9w842sl61axi6dd+GbB32TLw/+Mr269qp0eGZm1kROgg1Ys34N1zx3DbcvuZ0PPv6AIbsM4ccjfsyo/qPoWNWx0uGZmVkzVSwJShoF/AKoAq6PiCtqTb8IOA/YALwJnFPz6CZJG4H5adY/R8Qp5YixY1VH7ll+D5/t81kmDJnAwbsf7BtZm5ntQCqSBCVVAVcDXwBWAE9Jmh4RL+RmexYYFhEfSvom8O/AmWnaRxExtNxxdmjXgd+N/p0fX2RmtoOq1M0qhwNLI2JZRHwMTAZOzc8QEQ9ExIdp9Algz1aOEcAJ0MxsB1apJNgXeCU3viKV1edc4A+58c6S5kh6QtLocgRoZmY7vu3+whhJE4BhwDG54r0j4lVJA4H7Jc2PiD/VsezXgK8B9OvXr1XiNTOztqNSLcFXgb1y43umsq1I+jxwCXBKRKyrKY+IV9PfZcCDwMF1bSQirouIYRExrFcv/5TBzMy2Vqkk+BQwSNIASR2BscD0/AySDgauJUuAK3PlPSV1SsO7AUcC+QtqzMzMSlKR7tCI2CDpAmAm2U8kboiIBZIuB+ZExHTgSqA7MDX9LKHmpxBDgGslbSJL4lfUuqrUzMysJIqISsfQKoYNGxZz5sypdBhmZm2GpKcjYlil4yinSnWHmpmZVVxhWoKS3gRebuLiuwFvtWA4LcVxNY7jahzH1Tg7Ylx7R8QOfVVhYZJgc0iasz12CTiuxnFcjeO4GsdxtU3uDjUzs8JyEjQzs8JyEizNdZUOoB6Oq3EcV+M4rsZxXG2QzwmamVlhuSVoZmaF5SRoZmaFVegkKGmUpMWSlkr6Th3TO0makqY/Kal/btp3U/liSce3clwXSXpB0jxJ90naOzdto6S56TW99rJljmuipDdz2z8vN+1sSUvS6+xWjuu/cjG9KOm93LRy1tcNklZKer6e6ZL0yxT3PEmH5KaVs762Fdf4FM98SY9JOig3bXkqnyupRW/BVEJcIyWtyr1f389Na/AYKHNcF+diej4dU7ukaeWsr70kPZA+CxZIurCOeSpyjLUpEVHIF9k9S/8EDAQ6As8B+9ea5++Ba9LwWGBKGt4/zd8JGJDWU9WKcR0LdE3D36yJK42vrmB9TQSuqmPZXYBl6W/PNNyzteKqNf+3yO5VW9b6Sus+GjgEeL6e6SeSPSdTwGeBJ8tdXyXGdUTN9oATauJK48uB3SpUXyOBu5p7DLR0XLXmPRm4v5Xqqw9wSBreCXixjv/JihxjbelV5JbgNp9un8ZvSsPTgOMkKZVPjoh1EfESsDStr1XiiogHIuLDNPoE2aOoyq2U+qrP8cCsiHgnIt4FZgGjKhTXWcCtLbTtBkXEw8A7DcxyKnBzZJ4AdpbUh/LW1zbjiojH0nah9Y6vUuqrPs05Nls6rtY8vl6LiGfS8AfAQv764eQVOcbakiInwVKebr95nojYAKwCdi1x2XLGlXcu2Te9Gp0lzZH0hKTRLRRTY+I6I3W7TJNU88zI7aK+UrfxAOD+XHG56qsU9cVezvpqrNrHVwD3SHpa2UOrW9vhkp6T9AdJB6Sy7aK+JHUlSyS354pbpb6Unao5GHiy1qS2cIxV1Hb/ZHmrn6QJwDDgmFzx3hHxqqSBwP2S5kfEn1oppN8Dt0bEOklfJ2tFf66Vtl2KscC0iNiYK6tkfW3XJB1LlgRH5IpHpPraHZglaVFqKbWGZ8jer9WSTgR+CwxqpW2X4mTgfyMi32ose31J6k6WeL8dEe+35LqLoMgtwVKebr95HkntgR7A2yUuW864kPR54BKyhw6vqymPiFfT32XAg2TfDlslroh4OxfL9cChpS5bzrhyxlKrq6qM9VWK+mIvZ32VRNJnyN7DUyPi7ZryXH2tBO6k5U4DbFNEvB8Rq9PwDKCDsgdrV7y+koaOr7LUl6QOZAmwOiLuqGOW7fYY225U+qRkpV5kreBlZN1jNSfTD6g1z/lsfWHMbWn4ALa+MGYZLXdhTClxHUx2IcCgWuU9gU5peDdgCS10gUCJcfXJDZ8GPJGGdwFeSvH1TMO7tFZcab79yC5SUGvUV24b/an/Qo+T2Pqihdnlrq8S4+pHdp77iFrl3YCdcsOPAaNaMa5P1rx/ZMnkz6nuSjoGyhVXmt6D7Lxht9aqr7TvNwM/b2Ceih1jbeVV8QAquvPZlVMvkiWUS1LZ5WStK4DOwNT0gTAbGJhb9pK03GLghFaO617gDWBuek1P5UcA89OHwHzg3FaO6yfAgrT9B4D9csuek+pxKfDV1owrjf8AuKLWcuWur1uB14D1ZOdczgW+AXwjTRdwdYp7PjCsleprW3FdD7ybO77mpPKBqa6eS+/zJa0c1wW54+sJckm6rmOgteJK80wku1guv1y562sE2TnHebn36sTt4RhrSy/fNs3MzAqryOcEzcys4JwEzcyssJwEzcyssJwEzcyssJwEzcyssJwEzbZz6ekJd1U6DrMdkZOgmZkVlpOgWQuRNEHS7PTsuGslVUlanZ5nuEDZsx97pXmHppt2z5N0p6SeqXxfSfemm0Q/I2mftPru6abkiyRVp6eZmFkzOQmatQBJQ4AzgSMjYiiwERhPdrusORFxAPAQcFla5Gbg/0TEZ8ju5FFTXg1cHREHkd3R5rVUfjDwbbJnWQ4Ejiz7TpkVgJ8iYdYyjiO7YfhTqZHWBVgJbAKmpHkmAXdI6gHsHBEPpfKbgKmSdgL6RsSdABGxFiCtb3ZErEjjc8nuZflo+XfLbMfmJGjWMgTcFBHf3apQ+l6t+Zp6n8J1ueGN+H/XrEW4O9SsZdwHjEnPjUPSLukhvu2AMWmeccCjEbEKeFfSUan8K8BDkT0dfEXNw30ldUoPajWzMvG3SbMWEBEvSLqU7Cni7cieOHA+sAYYnqatJDtvCHA2cE1KcsuAr6byrwDXSro8rePLrbgbZoXjp0iYlZGk1RHRvdJxmFnd3B1qZmaF5ZagmZkVlluCZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CZmZWWE6CJZK0ejuIob+k51t4nT+X9KqkdrmyiZLelDRX0guS/i43bbSk77dkDI0h6T8kfa6E+UZL2r8J6z9F0neasFw53ptpkgam4TqPP0k3ShrTxPUfKOnGZoRY33o314WkYZJ+Wc98yyXt1sB6WvR/Lh3rR7fkOq3tcxIssJT4TgNeAY6pNXlKRAwFRgL/Jql3Kv8X4NeN2Eb7Fgg171dAKUlqNFBnEmwopoiYHhFXNDG2FiPpAKAqIpaVaxsRMR/YU1K/Mm5jTkT8Q7nWXypJuwKfjYiHG7FMVRlDsu2Ek2AzSNpH0h8lPS3pEUn75cqfkDRf0o/y32glXSzpKUnzJP0wlfWXtFDSbyQtkHSPpC5p2qGSnpP0HHB+iXGtlnRlWte9koZLelDSMkmn5GYdCSwA/hs4q651RcRK4E/A3pIGA+si4q2G9lPSyFQf04EXJFWleGr2++tNrY+IeBnYVdInG9j/I4BTgCtTa3aftP8/lzQHuFDSyZKelPRsqqPeadmJkq5KwzdK+qWkx1LdldTiktRZ0v9L9fKspGNT+QGSZqeY5kkaJKmbpLvTe/y8pDPTasYDv6u13v9K9XGfpF51bHdzyyq1wB5Mw90k3ZC2/aykU3OL/R4Yu439mSzppNz4jZLGpPfpEUnPpNcRdSw7UtJdaXjX9F4ukHQ9oG3XJihzZaqf+TV1JKmPpIdTfT4v6ah0rN2Ym/cf02rOAP6YW+dxqS7mp7rplKvDn0p6BviypL9Lx+dzkm6X1DVXB391bEhqJ+nXkhZJmiVpRm7aoZIeUvZ5MVNSn1L238osIvwq4QWsrqPsPmBQGj4MuD8N3wWclYa/UbMs8EXgOrJ//nZpvqOB/sAGYGia7zZgQhqeBxydhq8Enk/DewAz6ok1gBPS8J3APUAH4CBgbm6+3wBfAT4BvAp0SOUTgavS8EBgJbAL8FXgP3PL17efI4E1wIA0/jXg0jTcCZgDDGhKfeTiPmMb79eNwJjc+IPAr3PjPQGl4fNq9qvWvt8ITE2x7Q8sbWB7/XPvzT8BN6Th/YA/A53JWrHjU3lHoAvZh/Nvcuvpkf4+BBxY6z2tWfb7tWIck4aXA7ul4WHAg2n439hyPO0MvAh0S+NHAr/fRl2eBtyUi/uVFHtXoHMqHwTMqaMuRgJ3peFfAt9PwyelfdptW/9zqY5mAVVA71SffVI9X5LmqQJ2Ag4FZuXWsXP6exNwchrunPZhcBq/Gfh2rg7/Jbf8rrnhHwHfaujYAMYAM1L5J4F3U1kH4DGgV5rvzJpjxK/Kvlq6q6owJHUHjgCmSpu/0HZKfw8n644DuAX4jzT8xfR6No13J/vw+DPwUkTMTeVPA/0l7Uz2T1zThfM/wAkAEfEX4MR6wvuYLd9655O13tZLmk/2AYWkjmn5iyLiA0lPAseTJSKAMyWNANYBX4+Id9I31zdz26lvPwFmR8RLuf3+TK4l1SPtd6PqI7fulWRfAhprSm54T2BK2qeOwEt1L8JvI2ITWYu2dz3z1DaCLOEREYskvQwMBh4HLpG0J3BHRCxJ78l/SvopWbJ4JK2jdl1vysU/CbijxFggq+NTJP1zGu8M9AMWUlpd/gH4RWotjQIejoiPJPUArpI0FNiY9rEhRwOnA0TE3ZLeLTH+EcCtEbEReEPSQ8DfAE8BN0jqQPY+zZW0DBgo6VfA3WRfAGHr+vwU2fH1Yhq/iayX5edpPH+cfFrSj8i+PHQHZuam1XVsjACmpvLXJT2Q2+angVnp86IKeK3E/bcychJsunbAe5GdNyuVgJ9ExLVbFUr9yZJNjY1k37Sban2kr5tkH57rACJik7acDzue7B97fvqn7Ap8xJYkOCUiLqi13o/IElgp1uSGRfYNOv8BgqTjaVp9dE6xNFY+pl8BP4uI6ZJGAj+oZ5l8HCV139UnIm5JXzZOAmZI+npE3C/pELIvJD+SdF9EXE62f50bWl0dZRvYcoojv6zIWs6L61hmm3UZEWtT1+rxZC2YyWnSPwJvkPUwtAPWNrSelhYRDyu70OUk4EZJP4uImyUdlGL9BvC3wDlsuz7z8sfJjcDoiHhO0kSylm2NxhwbAhZExOElxmCtxOcEmygi3gdekvRl2Hze4qA0+QmyLhzY+nzLTOCc1IpEUl9JuzewjfeA91KLDLLzRC3lLOC8iOgfEf3Juie/UHPOox4LgX1z4/XtZ20zgW+mb+xIGiypG42sj5zBQM3Vhz+RdFod83xA1j1Wnx5kXcAAZ5ewzcZ4hPReKTuP2g9YrOxKz2UR8Uuy832fkbQH8GFETCLr7j4kraN2Xbcj61YDGAc8Wsd2l5N1B8KW9wWyev6W0rcdSQfnpuXrcrikm+vZpylk3eFHsaWXoQfwWmr1fIWsddOQh1PsSDqBrEu6FI+Q9UxUKTsXejQwW9LewBsR8RvgeuAQZedE20XE7cCl1F2fi8l6WmrGv0LW/VyXnYDX0rFbyv/f/wJnpHODvdmSNBcDvSQdDiCpg7KLn6zCnARL11XSitzrIrJ/inOVXbSyAKi54ODbwEWS5pH9460CiIh7yLoNH0/dYNNo+IMasg+eqyXNJfdtU9IekmY0ZUdSohtF1l1Eim0N2QfryQ0s+jBwcM2HKfXsZx2uB14AnlF26fy1QPum1Ef6MNqX7LwiwIHA63XMOhm4OF38sE8d039A1pX9NPBWQ9tsgl8D7dI+TQEmRsQ6slbJ8+m9/DTZuagDyT7Q5wKXkZ13guy9GZlb5xpgeKq/zwGX17HdH5J1W84haz3X+L9k56TmSVqQxmscy5bjoB/1twrvIbuC+N6I+Di3n2en438/tm5B1eWHwNEphtPJur1LcSfZufHngPvJztm9TlY/z0l6lqyF+gugL/Bgqs9JwHfTOjbXZ0SsJfu/mpreo03ANfVs+3vAk2TJbVEJsd4OrCA73icBzwCrUp2NAX6a6msu2ekUqzBt6TWzlpKSzEcREZLGkl08cuq2lmsLJP2C7EKKeyuxn6nVd0hEfC+Nz4yI48u5zUpQdjXsA8CR6VxYObbRiawFNCIiNki6EvifiJhXju1VmqRHgS+lHpZybqd7RKxW9rOM2WTvYV1f1Gw74CRYBpKOAq4ia7m9B5wTEUsrG1XLSF08h6Vzaa2+n6n7eVa5P8i2B+mc6cKIKLXF1Nj1DwL6RsSD5Vj/9kbSYWRf2sqa5NP5053JLrj694i4sZzbs+ZxEjRrBEkHkl2lm7cuIg6rRDxtXWot3VfHpOMi4u3WjseKx0nQzMwKyxfGmJlZYTkJmplZYTkJmplZYTkJmplZYTkJmplZYf1/8LArb/k2psIAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jSoCSJZg5PCr"
      },
      "source": [
        "# Finding: baseline, learn reset at each run"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zt2g0SbbO6SP"
      },
      "source": [
        "#  next: see if lr_find changes with each iteration"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 439
        },
        "id": "4nspUbpEnc8s",
        "outputId": "45783755-2659-454b-e4b1-437f9b779f20"
      },
      "source": [
        "# using retinanet/ resnet50: focal loss\n",
        "model_type_ret = models.mmdet.retinanet #\n",
        "backbone_r50 = model_type_ret.backbones.resnet50_fpn_1x(pretrained=True)#\n",
        "model_ret = model_type_ret.model(backbone=backbone_r50(pretrained=True), #\n",
        "                      num_classes=len(parser.class_map))\n",
        "\n",
        "train_dl_ret = model_type_ret.train_dl(train_ds, batch_size=16, num_workers=4, shuffle=True)#\n",
        "valid_dl_ret = model_type_ret.valid_dl(valid_ds, batch_size=16, num_workers=4, shuffle=False)#\n",
        "\n",
        "learn_ret = model_type_ret.fastai.learner(dls=[train_dl_ret, valid_dl_ret], model=model_ret, metrics=metrics)#\n",
        "learn_ret.lr_find()#\n",
        "# use ***"
      ],
      "execution_count": 24,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/mmcv/cnn/utils/weight_init.py:119: UserWarning: init_cfg without layer key, if you do not define override key either, this init_cfg will do nothing\n",
            "  'init_cfg without layer key, if you do not define override'\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "Use load_from_local loader\n",
            "The model and loaded state dict do not match exactly\n",
            "\n",
            "size mismatch for bbox_head.retina_cls.weight: copying a param with shape torch.Size([720, 256, 3, 3]) from checkpoint, the shape in current model is torch.Size([333, 256, 3, 3]).\n",
            "size mismatch for bbox_head.retina_cls.bias: copying a param with shape torch.Size([720]) from checkpoint, the shape in current model is torch.Size([333]).\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              ""
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "SuggestedLRs(lr_min=3.981071640737355e-05, lr_steep=0.0003311311302240938)"
            ]
          },
          "metadata": {},
          "execution_count": 24
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 205
        },
        "id": "JI-gYpJOnjq5",
        "outputId": "c83196c6-3f23-42f2-c603-22a3296c1f36"
      },
      "source": [
        "# 1st run\n",
        "learn_ret.fine_tune(3, 3.981071640737355e-05, freeze_epochs=1) #\n"
      ],
      "execution_count": 25,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: left;\">\n",
              "      <th>epoch</th>\n",
              "      <th>train_loss</th>\n",
              "      <th>valid_loss</th>\n",
              "      <th>COCOMetric</th>\n",
              "      <th>time</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <td>0</td>\n",
              "      <td>0.991052</td>\n",
              "      <td>0.803205</td>\n",
              "      <td>0.050723</td>\n",
              "      <td>01:34</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: left;\">\n",
              "      <th>epoch</th>\n",
              "      <th>train_loss</th>\n",
              "      <th>valid_loss</th>\n",
              "      <th>COCOMetric</th>\n",
              "      <th>time</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <td>0</td>\n",
              "      <td>0.723789</td>\n",
              "      <td>0.669747</td>\n",
              "      <td>0.093637</td>\n",
              "      <td>01:43</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>1</td>\n",
              "      <td>0.629509</td>\n",
              "      <td>0.584949</td>\n",
              "      <td>0.181908</td>\n",
              "      <td>01:42</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>2</td>\n",
              "      <td>0.586359</td>\n",
              "      <td>0.569237</td>\n",
              "      <td>0.210489</td>\n",
              "      <td>01:41</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "id": "ThMy1VdKnpdc",
        "outputId": "d03b9f9d-e36a-4c74-c606-1deb15f403be"
      },
      "source": [
        " plot_metrics(learn_ret, '1st run- Mean Average Precision and Losses for Retinanet/Resnet50', 0, 0.01)"
      ],
      "execution_count": 26,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 300
        },
        "id": "7dgdFtsfnucC",
        "outputId": "8baf2a9d-f70d-4820-8645-aa4da3a8fdc9"
      },
      "source": [
        "#2nd run\n",
        "learn_ret.lr_find()#\n"
      ],
      "execution_count": 27,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              ""
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "SuggestedLRs(lr_min=5.754399353463669e-07, lr_steep=1.5848931980144698e-06)"
            ]
          },
          "metadata": {},
          "execution_count": 27
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 205
        },
        "id": "hul97W87nxc0",
        "outputId": "73f0c958-1135-47ef-98bd-f6d70129f540"
      },
      "source": [
        "# 2nd run\n",
        "learn_ret.fine_tune(3, 1e-5, freeze_epochs=1) #\n"
      ],
      "execution_count": 28,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: left;\">\n",
              "      <th>epoch</th>\n",
              "      <th>train_loss</th>\n",
              "      <th>valid_loss</th>\n",
              "      <th>COCOMetric</th>\n",
              "      <th>time</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <td>0</td>\n",
              "      <td>0.583649</td>\n",
              "      <td>0.555875</td>\n",
              "      <td>0.229365</td>\n",
              "      <td>01:32</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: left;\">\n",
              "      <th>epoch</th>\n",
              "      <th>train_loss</th>\n",
              "      <th>valid_loss</th>\n",
              "      <th>COCOMetric</th>\n",
              "      <th>time</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <td>0</td>\n",
              "      <td>0.565111</td>\n",
              "      <td>0.540557</td>\n",
              "      <td>0.245409</td>\n",
              "      <td>01:41</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>1</td>\n",
              "      <td>0.554450</td>\n",
              "      <td>0.535192</td>\n",
              "      <td>0.261858</td>\n",
              "      <td>01:41</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>2</td>\n",
              "      <td>0.542291</td>\n",
              "      <td>0.523560</td>\n",
              "      <td>0.272261</td>\n",
              "      <td>01:40</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 403
        },
        "id": "uFRnK8Bxn5M0",
        "outputId": "48bf3920-0d36-49b0-828e-57b2f250a671"
      },
      "source": [
        " # 2nd run\n",
        " plot_metrics(learn_ret, '2nd run- Mean Average Precision and Losses for Retinanet/Resnet50', 0, 0.01)"
      ],
      "execution_count": 29,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 300
        },
        "id": "_Zp8JnXan_Jb",
        "outputId": "5053d42a-b7a5-4a08-b689-91c4dc2bb445"
      },
      "source": [
        "# 3rd run\n",
        "learn_ret.lr_find()#\n"
      ],
      "execution_count": 30,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              ""
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "SuggestedLRs(lr_min=7.585775847473997e-08, lr_steep=4.786300905834651e-06)"
            ]
          },
          "metadata": {},
          "execution_count": 30
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 205
        },
        "id": "gWhHMuP9oEg2",
        "outputId": "e372bbf4-f54a-47f1-fdc2-210474d4bae0"
      },
      "source": [
        "# 3rd run\n",
        "learn_ret.fine_tune(3, 1e-6, freeze_epochs=1) #\n"
      ],
      "execution_count": 31,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: left;\">\n",
              "      <th>epoch</th>\n",
              "      <th>train_loss</th>\n",
              "      <th>valid_loss</th>\n",
              "      <th>COCOMetric</th>\n",
              "      <th>time</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <td>0</td>\n",
              "      <td>0.533509</td>\n",
              "      <td>0.521496</td>\n",
              "      <td>0.274555</td>\n",
              "      <td>01:31</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: left;\">\n",
              "      <th>epoch</th>\n",
              "      <th>train_loss</th>\n",
              "      <th>valid_loss</th>\n",
              "      <th>COCOMetric</th>\n",
              "      <th>time</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <td>0</td>\n",
              "      <td>0.536507</td>\n",
              "      <td>0.520570</td>\n",
              "      <td>0.276359</td>\n",
              "      <td>01:40</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>1</td>\n",
              "      <td>0.535959</td>\n",
              "      <td>0.517687</td>\n",
              "      <td>0.279339</td>\n",
              "      <td>01:41</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>2</td>\n",
              "      <td>0.537849</td>\n",
              "      <td>0.517783</td>\n",
              "      <td>0.280043</td>\n",
              "      <td>01:41</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 459
        },
        "id": "eMml3wGXoHid",
        "outputId": "1199f84a-ec6e-4098-a9a0-22037829e5a7"
      },
      "source": [
        "# 3rd run\n",
        "plot_metrics(learn_ret, \n",
        "              '3rd run- Mean Average Precision and Losses for Retinanet/Resnet50', 0, 0.01)"
      ],
      "execution_count": 34,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "oRXrZJyetWhw"
      },
      "source": [
        "# Finding:  LR goes smaller and smaller\n",
        "#    if adjust lr everytime to smaller values, learning decreases"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ANwxKoRmpdnf"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 439
        },
        "id": "G_MIqi6epeeC",
        "outputId": "8d376777-6a1b-4b7c-cb20-17a487243384"
      },
      "source": [
        "# 1st run\n",
        "model_type_ret = models.mmdet.retinanet #\n",
        "backbone_r50 = model_type_ret.backbones.resnet50_fpn_1x(pretrained=True)#\n",
        "model_ret = model_type_ret.model(backbone=backbone_r50(pretrained=True), #\n",
        "                      num_classes=len(parser.class_map))\n",
        "\n",
        "train_dl_ret = model_type_ret.train_dl(train_ds, batch_size=16, num_workers=4, shuffle=True)#\n",
        "valid_dl_ret = model_type_ret.valid_dl(valid_ds, batch_size=16, num_workers=4, shuffle=False)#\n",
        "\n",
        "learn_ret = model_type_ret.fastai.learner(dls=[train_dl_ret, valid_dl_ret], model=model_ret, metrics=metrics)#\n",
        "\n",
        "learn_ret.lr_find()"
      ],
      "execution_count": 36,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/mmcv/cnn/utils/weight_init.py:119: UserWarning: init_cfg without layer key, if you do not define override key either, this init_cfg will do nothing\n",
            "  'init_cfg without layer key, if you do not define override'\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "Use load_from_local loader\n",
            "The model and loaded state dict do not match exactly\n",
            "\n",
            "size mismatch for bbox_head.retina_cls.weight: copying a param with shape torch.Size([720, 256, 3, 3]) from checkpoint, the shape in current model is torch.Size([333, 256, 3, 3]).\n",
            "size mismatch for bbox_head.retina_cls.bias: copying a param with shape torch.Size([720]) from checkpoint, the shape in current model is torch.Size([333]).\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              ""
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "SuggestedLRs(lr_min=8.317637839354575e-05, lr_steep=0.00013182566908653826)"
            ]
          },
          "metadata": {},
          "execution_count": 36
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYsAAAEKCAYAAADjDHn2AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO3dd3xW5f3/8dfnvrNICAmQMAOywt4EkClWrYi4cKG1oFIRR622ta3+2q9+7bTDbVWUodiKo2hx4WhFBFFIhCB7b4SwVyAkuX5/5PbbaEnuBHLn3OP9fDzygPuc65z7k8NN3jnnOue6zDmHiIhIZXxeFyAiIuFPYSEiIkEpLEREJCiFhYiIBKWwEBGRoBQWIiISVJzXBdSkjIwM16pVK6/LEBGJGHl5ebudc5nB2kVVWLRq1Yrc3FyvyxARiRhmtqkq7XQZSkREglJYiIhIUAoLEREJSmEhIiJBKSxERCQohYWIiASlsKjEjgOF7DhQ6HUZIiKeU1hUYN+RIi578lMmvPiF16WIiHguqh7KqynOOX4xYwlfHTzGzkPH2HekiPopCV6XJSLiGZ1ZnMT0hVt4b9lOLuzWFOdg/vo9XpckIuIphcW3rN11mP99cxlDsjN4+Oqe1E2MY+7a3V6XJSLiKYVFOceLS7jjpUUkJ8Txlyt7kBDn48w2DZinsBCRGKewKOfP761i+Y6DPHh5dxrVSwJgULsMNu05ypa9Rz2uTkTEOwqLgDmrC3j2kw18/8wzOK9z4/9bPrhdBoDOLkQkpiksgD2Hj/OTV/PJblSX/3dhp2+sa9eoLo1SE9VvISIxLeZvnXXO8bPXlnCg8AQv3NiPpHj/N9abGYPbZTB7dQGlpQ6fzzyqtGKFRSU8+8l6XlqwmXpJ8TRNT6JZeh2apSXRNK0OLRokk3NG/bCsXUQiQ8yHxcHCYnYdOs49F3SkU9N6J20zODuDGYu2seKrg3RpllbLFVbMOcfM/O08+O5Kth84xtD2mSTG+dhxoJAlWw+w90jR/7W97syW/ObSbh5WKyKRLObDIi05nn/cMpB4f8W/dQ8K9FvMXbM7bMJi8Zb9PPDmMr7YvJ8uzerx0NU9ObNNw2+0KSwqYceBQqbM28i0zzYxrH0jzi3XHyMiUlUxHxYACXGVd900rpdEdqO6zF27m5vPaltLVZVxznHwWDF7Dh9nz5Eidh86zgfLdzJj0TYy6ibyx8u7c3mfLPwnucRUJ8FPm8y6/HJkJ/I27ePn/1jCrBZDyUxNrNXvQUQin8Kiiga1y2D6ws0cO1HyX/0aofD0x+t4/tON7DlcRFFJ6TfWJfh93DKsLbed3Y66icH/CRPj/Dw6uicjH5/Lz17LZ/L1fTFT/4WIVJ3CoooGt8tg6qcb+WLzPga2zQjpex06doLH/7WG1pkpXNyzGRkpiWSkJtAwJZGMuok0T69DWnJ8tfaZ3TiVey7oyP1vLufFzzbx/QGtQlO8iEQlhUUV9W/TAL/PmLd2d8jDYsYX2zhSVMJvLu1GzxbpNbbfsQNb8dGqAn7z9goGtG1Iu0apNbZvEYlues6iilKT4unZIp25a0M7qKBzjmmfbaJ7VlqNBgWU3Qb8pyu7k5IYx50vL6aouDT4RiIiKCyqZVC7DL7cup8DR0+E7D3mr9vD2l2HGROiy0SNUpP4w6huLN12kIc+WP2NdUeLisnfsp9Xc7cwa+lXOOdCUoOIRB5dhqqGIdkZPPavNcxfv4fhXZuE5D1emL+J+snxjOzeNCT7B/hulyZc068Fz8xZR2FRMdv2F7Jq5yG27P3mrIA5Z9TngUu60rnZyZ8/EZHYobCohp4t0klJ8DNv7e6QhMWOA4V8sGInPxjSOuR3XP3yws4s2LCXv32+mTaZKfTISufKPi1o3ziV9o3rkrtxH3+YtZKRj3/CmAGtuOu89qTVqV6nuohED4VFNcT7ffRv0zBkgwr+/fPNlDrHdf3PCMn+y0tJjOPdHw0FTv6cSZvMuny3S2P+8v5qXpi/kbeWbOcXF3RiVK/mGjZEJAapz6KaBrXLYP3uI2zbXxi8cTUUFZfy0oItfKdDI1o0SK7RfVckIc5X6QOJ6ckJ/PrSrsy8fTAtGyTz01fzueyv83h2znrW7DykPg2RGKIzi2r6vyHL1+zmqr4tgLJRaz9YvpN3l37Fqq8OMfXGvnRsUr3r/O8u3cHuw8f5/oDQn1VUV9fmabw2YSCvfbGV5z5Zz2/fWcFv31lB8/Q6DG2fybAOmQxs25DUJF2mEolWIQsLM5sMjAR2Oee6nmT93cD3ytXRCch0zu01s+HAo4AfeM4594dQ1Vld7RvXJaNuIrOWfcXx4hLe+fIrPt+wh1IHLRskc7y4hLtezueftw0KOoxIedPmb6JVw2SGZmeGsPpT5/MZV+W04KqcFmzbX8ic1QXMXrWLN/O389KCzfgMMlMTaZSaRON6iTSql0Tj1CSapCXynY6NNcSISISzUF1KMLOhwGHghZOFxbfaXgTc5Zz7jpn5gdXAecBWYCFwjXNuebD3zMnJcbm5uadffBB3Tl/EG4u3A9A2M4UR3ZoyvGsTOjetx4crdnHTC7ncdnZb7j6/Y5X2t3z7QUY89gm/vLATPxjSJpSl17gTJaXkbdrH/HV72HGgkJ0Hj7Pz4DEKDpWNZQVQPzme317WjRHdQneHl4icGjPLc87lBGsXsjML59wcM2tVxebXAC8F/t4PWOucWw9gZtOBS4CgYVFb7jgnmy7N0hjWIZPsxt98Cvq8zo25sk8WT81exzmdGtO7Zf2g+5v22UaS4n1c2adFqEoOmXi/jzPbNPyvEW+hrB9m9c5D3Pv6l9z6ty+4tGcz/vfirtUeqkREvOd5B7eZJQPDgX8EFjUHtpRrsjWwrKLtx5tZrpnlFhQUhK7Qctpk1uWmoW3+Kyi+9j8XdaZpWh1+8ko+R4uKK93XgcITvLFoO5f0aB51P0QT4nx0bZ7GP24ZyJ3nZvPWkh2c/8gc5qyunX8nEak5nocFcBEwzzm391Q2ds5NdM7lOOdyMjPD43p/alI8f7qyOxt2H+HBd1dW2va1vK0UnigJy47tmhLv93Hnue15/dZBpCbFMWbyAn75xpdBg1REwkc4hMVo/nMJCmAbUP56TFZgWUQZ2DaDGwa14vn5m5i75r+fyzheXMKLn23iyY/W0rtlOl2bh8ekSqHULSuNN384mJuGtOZvn29myIMf8ef3VvHVgWNelyYiQYSsgxsg0GfxVkUd3GaWBmwAWjjnjgSWxVHWwX0OZSGxELjWObcs2PvVVgd3VR07UcKIxz6hsKiEWXcOJa1OPMdOlPBK7haemr2OHQeO0atlOr+7rFuFU7pGq7xN+3j643V8uGInfjOGd23CDYNa07tluubaEKlFVe3gDuXdUC8Bw4AMYCdwHxAP4Jx7OtDmemC4c270t7YdATxC2a2zk51zv63Ke4ZbWEDZ9KeXP/UpI7s3pVeLdJ76eB07Dx4n54z6/OjcbAa3y4jpH46b9xxl2mcbmb5wC4eOFdM9K41bh7UL2dhbIvJNnoeFF8IxLAAeen8Vj/17LQD9WjfgznOyGdC2YUyHxLcdOV7M64u2MWXeBtYVHOHhq3twWa8sr8sSiXqe3zor//HDc7JJiPPR54wGDGj737eYStlYVdedeQZX5mRx/eSF3P3qEhqmJDK0fXjctCAS68Khgzvqxft93P6dbAVFFSTG+XlmTB+yG6cy4cU8lmzd73VJIoLCQsJQvaR4nr+hLw1SErhhykI27j7idUkiMU9hIWGpUb0kXrixHw4YM3kBuw7p9loRLyksJGy1yazLpLE5FBw6zg1TFnL4uB7iE/GKwkLCWq+W9fnrdb1Z+dUhbp6Wy4HC0M1/LiIVU1hI2Du7QyP+eHl35q/bw/kPz+HfK3d6XZJIzFFYSES4vE8Wr986iLQ68dw4NZe7Xl7MvsAQ6CISegoLiRg9WqQz84eDuOOcbN7M3855D89h1tIdXpclEhP0BLdEpOXbD3L3a/ks236Q4V2a0Ld1A+L9ht9nxPmMOJ+POL/RqmEKnZvVI96v34tETkbDfUjUO1FSysQ563n0X2soKi6tsF1SvI8eWenktKpPnzPq07tlfdKTE2qxUpHwpbCQmHG8uIRjRaUUl5ZSXOooLnWUlDiKSkpY+dUh8jbtI2/TPpZtP0hJadnnfUh2Br8f1Y2s+skeVy/iLYWFyLccLSomf8sBPlu/h0lzN2DA/Rd3YVTv5hrUUWJWVcNCF3IlZiQnxDGgbUPuOq897/5oCJ2a1uMnr+Zzy4tfsFd3VolUSmEhMalFg2ReGn8m91zQkX+v3MV39fyGSKUUFhKz/D7j5rPa8s/bB5FRN4Ebp+byk1fyyd+yn2i6PCtSE9RnIUJZJ/nDH6xh8rwNFBWX0jYzhVG9s7i0V3Oap9fxujyRkFEHt8gpOFB4gne+3MHrX2xjwca9mMGZrRtyeZ8sLu3ZjDg9ryFRRmEhcpo27znK64u2MWPRVjbtOUqHxqncf3EXTWIlUUVhIVJDnHO8v3wnv35rOVv3FXJRj2bcO6IjTdN0eUoin26dFakhZsb5XZrw4Y/P4kfnZPPesq845y8f89TsdZU+OS4STRQWIlWUFO/nrvPa8+FdZzGoXQYPzlrJ8EfmsPKrg16XJhJyCguRamrZMJlnx+Qw5Ya+HD5ezI1TFmraV4l6CguRU3R2h0ZMvr4v+46e4KYX8jh2osTrkkRCRmEhchq6Nk/jkdE9WbJ1P3e/tkQP80nUUliInKbzuzThZ+d35M387Tz6rzVelyMSEiELCzObbGa7zGxpJW2GmdliM1tmZh+XW77RzL4MrNO9sBL2JpzVhst7Z/HIh2t4M3+71+WI1Li4EO57KvAE8MLJVppZOvBXYLhzbrOZNfpWk7Odc7tDWJ9IjTEzfjeqK1v2HuWnr+bTokEyPVuke12WSI0J2ZmFc24OsLeSJtcCM5xzmwPtd4WqFpHakBjn5+nv96FRvUR+8Hwu2/YXel2SSI3xss+iPVDfzGabWZ6ZjSm3zgHvB5aP96g+kWprkJLA5LF9OX6ihIsfn8uspV95XZJIjfAyLOKAPsCFwPnAr8ysfWDdYOdcb+AC4DYzG1rRTsxsvJnlmlluQUFByIsWCSa7cSqv3TKQJmlJTHgxjzunL2L/UU2uJJHNy7DYCrznnDsS6JuYA/QAcM5tC/y5C3gd6FfRTpxzE51zOc65nMzMzFooWyS4Dk1SeeO2Qdx5bjZvLdmhyZUk4nkZFv8EBptZnJklA/2BFWaWYmapAGaWAnwXqPCOKpFwFe/3cee57XnjtkHUTy6bXOnuV/M5eOyE16WJVFvI7oYys5eAYUCGmW0F7gPiAZxzTzvnVpjZLGAJUAo855xbamZtgNfN7Ov6/u6cmxWqOkVCrWvzNGb+cBCPfriGpz9eR96mfUwffyaN6iV5XZpIlWmIcpFa9Nn6Pdw4dSHN0+swffyZNKyb6HVJEuM0RLlIGDqzTUMmje3L5r1H+f6kBRw4qktSEhkUFiK1bEDbhkwck8PaXYcZM2UBh9SHIRFAYSHigbPaZ/Lk93qzbNsBbpy6kKNFxV6XJFIphYWIR87r3JhHR/cib9M+bnohV0OcS1hTWIh46MLuTfnzlT34dN0ebnohV5MoSdhSWIh4bFTvLP4wqhufrd/Dd/78Mc/OWa+5vSXsKCxEwsDVfVvy/l1n0b91A377zgqGPzqH2as0tqaED4WFSJhonZHCpOv7MuX6vjgH109ZyA+ez2XTniNelyaisBAJN2d3bMR7dw7lngs6Mn/dbs57aA7vfrnD67IkxiksRMJQQpyPm89qy0c/HUa3rDRuf2kRby9RYIh3FBYiYaxRvSSev7EfvVumc8f0RZqyVTyjsBAJc3UT45h6Qz/6tKzPj6Yv4p+Lt3ldksQghYVIBEhJjGPKDX3p26oBd728WIEhtU5hIRIhvg6M/q0bctfLi3l90VavS5IYorAQiSDJCXFMvr4vZ7ZpyI9fyWfa/I1E0zQDEr4UFiIRpk6Cn0lj+/KdDo341T+Xce/rX+qJbwk5hYVIBKqT4GfimBxuO7stLy3YwjXPfqZxpSSkFBYiEcrvM+4+vyNPXNuL5dsPcvHj81iydb/XZUmUUliIRLiR3Zvx2i0D8PuMK5+er45vCQmFhUgU6NIsjZm3D6JXy3TuejmfP7+3Sh3fUqMUFiJRomHdRKaN68/ovi144qO1PPTBaq9LkigS53UBIlJz4v0+fndZNwAe//da4v0+7jgn2+OqJBooLESijM9n/O6ybhSXOh76YDVxfuPWYe28LksinMJCJAr5fMaDl3enuKSUP85aRbzPx01D23hdlkQwhYVIlPL7jD9f2YPiUsdv31mB32fcOLi112VJhKpSWJhZClDonCs1s/ZAR+Bd59yJkFYnIqclzu/j4at7UlLqeOCt5fh9xtiBrbwuSyJQVe+GmgMkmVlz4H3g+8DUUBUlIjUn3u/j0dG9OK9zY+6buYzfvbOCklLdVivVU9WwMOfcUWAU8Ffn3JVAl0o3MJtsZrvMbGklbYaZ2WIzW2ZmH5dbPtzMVpnZWjP7RRVrFJEKJMT5eOp7vRkz4AwmzlnPzdNyOXK82OuyJIJUOSzMbADwPeDtwDJ/kG2mAsMr2WE68FfgYudcF+DKwHI/8CRwAdAZuMbMOlexThGpQJzfxwOXdOWBS7rw0aoCLn/qU7btL/S6LIkQVQ2LO4F7gNedc8vMrA3wUWUbOOfmAHsraXItMMM5tznQfldgeT9grXNuvXOuCJgOXFLFOkUkiDEDWjH5+r5s21fIJU/MY9HmfV6XJBGgSmHhnPvYOXexc+5BM/MBu51zd5zme7cH6pvZbDPLM7MxgeXNgS3l2m0NLBORGnJW+0xm3DqQ5AQ/V0/8TDPvSVBVCgsz+7uZ1QvcFbUUWG5md5/me8cBfYALgfOBXwXutKoWMxtvZrlmlltQUHCaJYnEjuzGqbxx2yB6ZqXzo+mLuWHKAlbvPOR1WRKmqnoZqrNz7iBwKfAu0JqyO6JOx1bgPefcEefcbsruuOoBbANalGuXFVh2Us65ic65HOdcTmZm5mmWJBJbGqQk8OIP+nPviI7kbtrH8EfmcM+MLzU3hvyXqoZFvJnFUxYWMwPPV5zuvXf/BAabWZyZJQP9gRXAQiDbzFqbWQIwGph5mu8lIhVIiPMxfmhb5tx9NmMHtuLV3C0M+9NsHv1wDUeLdMeUlKlqWDwDbARSgDlmdgZwsLINzOwlYD7Qwcy2mtk4M5tgZhMAnHMrgFnAEmAB8Jxzbqlzrhi4HXiPsvB4xTm3rPrfmohUR/2UBO67qAsf/PgszmqfycMfrubsP89m8RZNqCRlz0+c2oZmcYEf7GEjJyfH5ebmel2GSFTI3biXu15ZzKFjxbx68wCyG6d6XZKEgJnlOedygrWragd3mpk99HVHspn9hbKzDBGJUjmtGvC3cWeS4Pdx3aTP2bL3qNcliYeqehlqMnAIuCrwdRCYEqqiRCQ8tGyYzLRx/Tl2opTvT/pcHd8xrKph0dY5d1/gQbn1zrn/BTTesUgM6NAklSk39GXnweOMmbSAA4UaPzQWVTUsCs1s8NcvzGwQoHECRGJE75b1mTimD+sKDnPj1IW6SyoGVTUsJgBPmtlGM9sIPAHcHLKqRCTsDMnO5NHRvVi0eR+3vPgFRcWlXpcktaiqw33kO+d6AN2B7s65XsB3QlqZiISdEd2a8vtR3fh4dQE3vZDLoWO6JBUrqnpmAYBz7mDgSW6AH4egHhEJc1f3bckfRnVj7trdXPn0fLZr5NqYUK2w+BarsSpEJKKM7teSqTeUjVx76ZPz+HLrAa9LkhA7nbDQVFsiMWxIdiav3TKQeL+Pq56ZzwfLd3pdkoRQpWFhZofM7OBJvg4BzWqpRhEJUx2apPL6bQPJblyX8dNymTR3A6c6KoSEt0rDwjmX6pyrd5KvVOdcXG0VKSLhq1FqEi+PH8B3Ozfm128t54G3liswotDpXIYSEQGgToKfp77XhxsGtWLKvI3cP3OZAiPK6OxARGqEz2f8z8jO+MyYNHcDZsZ9F3XGTPfCRAOFhYjUGDPjlxd2wjmYPG8DgAIjSigsRKRGmRm/GtkJh2PKvI2Ywf+MVGBEOoWFiNQ4s7JLUkBZYFAWIAqMyKWwEJGQ+Dowvr4kZQa/vFCBEakUFiISMl93cgNMmlvWh6HAiEwKCxEJKQVGdFBYiEjIKTAin8JCRGqFAiOyKSxEpNYoMCKXwkJEapUCIzIpLESk1v1nKJCywCgpdWVDhfgUGOFKYSEinvj6OYw4n/HsJxs4XlzCby/tpsAIUwoLEfGMmXHviE4kxvl54qO1HC8u5U9X9MCvwAg7CgsR8ZSZ8dPzO5AY5+MvH6zmRInjoat6EO/XDArhJGRhYWaTgZHALudc15OsHwb8E9gQWDTDOfdAYN1G4BBQAhQ753JCVaeIhIcfnpNNQpyP37+7kqLiEh67pheJcX6vy5KAUEb3VGB4kDafOOd6Br4e+Na6swPLFRQiMeLms9py/0WdeW/ZTiZMy6OwqMTrkiQgZGHhnJsD7A3V/kUkOl0/qDW/H9WN2asLGPHYJyzYoB8j4cDri4IDzCzfzN41sy7lljvgfTPLM7PxXhUnIt64pl9L/jauPydKSrl64nzun7mMI8eLvS4rpnkZFl8AZzjnegCPA2+UWzfYOdcbuAC4zcyGVrQTMxtvZrlmlltQUBDaikWk1gxsl8F7dw5l7IBWTP10I8MfncOna3d7XVbM8iwsnHMHnXOHA39/B4g3s4zA622BP3cBrwP9KtnPROdcjnMuJzMzsxYqF5HakpIYx/0Xd+GVmwcQ5/Nx7XOfc8+MLzl07ITXpcUcz8LCzJpY4Pl+M+sXqGWPmaWYWWpgeQrwXWCpV3WKiPf6tW7AO3cMYfzQNry8cDPXPfc5h3VZqlaFLCzM7CVgPtDBzLaa2Tgzm2BmEwJNrgCWmlk+8Bgw2jnngMbA3MDyBcDbzrlZoapTRCJDnQQ/947oxNPX9WHp9oNMmJbH8WLdLVVbrOznc3TIyclxubm5XpchIiH2Wt5WfvpqPiO6NeHxa3rrie/TYGZ5VXlEQU9wi0jEuaJPFvuPFvGbt1eQVmcpv7usq0atDTGFhYhEpB8MacPeI0X8dfY6GqTEc/f5Hb0uKaopLEQkYt19fgf2HS3iyY/WUT85gR8MaeN1SVFLYSEiEcvM+M2l3dh/9AS/eXsF9ZMTuLxPltdlRSWvn+AWETktfp/xyOieDGrXkF/MWELuRg0PEgoKCxGJeIlxfv56bR+ap9dhwotfsONAodclRR2FhYhEhbTkeJ4dk0NhUTE3T8vj2Ak9g1GTFBYiEjWyG6fy8NU9WbL1APfO+JJoeo7MawoLEYkq3+3ShLvObc+MRduYNHdD8A2kShQWIhJ1fviddpzfpTG/e2cFc9dopNqaoLAQkajj8xl/uaon7RrV5ba/f8GmPUe8LiniKSxEJCrVTYzj2TFlQx6NfyFPw5qfJoWFiEStMxqm8OS1vVlbcJg7XlpEcUmp1yVFLIWFiES1wdkZPHBJFz5aVcBv3l7hdTkRS8N9iEjU+17/M1hfcIRJczfQJjOFMQNaeV1SxFFYiEhMuHdEJzbtOcL9M5fRskEywzo08rqkiKLLUCISE/w+49HRvejQpB63/30Rq7465HVJEUVhISIxIyUxjkljc0hO8HPj1IUUHDrudUkRQ2EhIjGlWXodnhubw54jxxk/LVdjSFWRwkJEYk73rHQeubonizbv5yev5FNaqjGkglFYiEhMGt61Kfdc0JG3v9zBn95f5XU5YU93Q4lIzBo/tA2b9h7lqdnraNkgmWv6tfS6pLClsBCRmGVmPHBxF7btK+SXbywlq34dhmRnel1WWNJlKBGJaXF+H09c24vsRnW59cUvdEttBRQWIhLzUpPimXx9X5IT/dwwZQG7Dh7zuqSwo7AQEaHsltpJY/uyv/AE457P5WhRsdclhRWFhYhIQNfmaTx+TS+WbT/ALS9+QVGxRqn9WsjCwswmm9kuM1tawfphZnbAzBYHvv6n3LrhZrbKzNaa2S9CVaOIyLed06kxv7usGx+vLuCnr+oZjK+F8m6oqcATwAuVtPnEOTey/AIz8wNPAucBW4GFZjbTObc8VIWKiJQ3ul9L9h09wYOzVlI/OZ77L+6CmXldlqdCFhbOuTlm1uoUNu0HrHXOrQcws+nAJYDCQkRqzYSz2rDvaBET56ynfkoCd57b3uuSPOV1n8UAM8s3s3fNrEtgWXNgS7k2WwPLTsrMxptZrpnlFhQUhLJWEYkhZsY9F3Tkyj5ZPPLhGp7/dKPXJXnKy4fyvgDOcM4dNrMRwBtAdnV34pybCEwEyMnJ0cVFEakxZsbvR3Vjf+EJ7pu5jPTkeC7pWeHvrlHNszML59xB59zhwN/fAeLNLAPYBrQo1zQrsExEpNbF+X08fk0v+rduwE9eyeejVbu8LskTnoWFmTWxQI+RmfUL1LIHWAhkm1lrM0sARgMzvapTRCQp3s9zY3Po0CSV2/8Wm095h/LW2ZeA+UAHM9tqZuPMbIKZTQg0uQJYamb5wGPAaFemGLgdeA9YAbzinFsWqjpFRKoiNSmeSWP7kpIYx41TF7L7cGxNnGTORc9l/pycHJebm+t1GSISxZZs3c+VT8+na/M0/n5TfxLj/F6XdFrMLM85lxOsndd3Q4mIRJTuWen85aoe5G3axz0zviSafuGujIYoFxGpppHdm7Fu1xEe/nA17RrV5dZh7bwuKeQUFiIip+COc9qxtuAwf5y1ijYZdRnetYnXJYWULkOJiJwCM+NPV3SnR4t07np5MUu3HfC6pJBSWIiInKKkeD/Pfr8P6cnxXD9lAXmb9nldUsgoLERETkOjeklMG9ef5IQ4rpn4Ga8v2up1SSGhsBAROU3tGtXln7cNolfLdO56OZ8/vbcy6oY2V1iIiNSA+ikJTBvXn9F9W/DkR+u45W95UTXbnsJCRKSGJMT5+P2obvxqZGc+WL6TK56az/b9hV6XVSMUFiIiNcjMGDe4NUWloU0AAAmdSURBVJPG9mXz3qNc8uQ8lmzd73VZp01hISISAmd3bMSMWweS4Pdx9TOf8f6yr7wu6bQoLEREQqR941Rev20g7RvX5eYX85g8d4PXJZ0yhYWISAg1Sk1i+vgBnNepMQ+8tZz7Zy6jJALvlFJYiIiEWJ0EP09d14dxg1sz9dON3Dwt8u6UUliIiNQCv8/41cjOPHBJF/69cidXP/MZm/cc9bqsKlNYiIjUojEDWvHsmBw27D7C+Y/MYdLcDRFxWUphISJSy87p1JgPfjyUAW0b8uu3lnPF05+yZmd4T9WqsBAR8UDTtDpMGpvDo6N7snH3ES58bC6P/WsNJ0pKvS7tpBQWIiIeMTMu6dmcD358Fud3bcJDH6zmosfnMnfN7rCbgU9hISLisYy6iTx+TS+eHZPD/qMnuG7S54x8fC4z87dTHCZnGhZu6XU6cnJyXG5urtdliIicsmMnSnhj0TYmfrKe9QVHyKpfh3GDW3N13xYkJ9T85KZmluecywnaTmEhIhJ+Sksd/1q5i2c+Xkfupn2kJ8dzRe8sLuzelJ4t0jGzGnkfhYWISJTI27SPZ+es518rd3KixNEsLYkLujVlRLem9GqRjs936sGhsBARiTIHCk/wrxU7eefLHcxZvZuiklKapiUxvGsT7h3RiXh/9buhqxoWNX8BTEREQiKtTjyjemcxqncWB4+d4N8rdvH2lzv4YtO+UwqK6lBYiIhEoHpJ8VzaqzmX9mpeK1O46tZZEZEIdzp9FlV+j1Dt2Mwmm9kuM1sapF1fMys2syvKLSsxs8WBr5mhqlFERKomlJehpgJPAC9U1MDM/MCDwPvfWlXonOsZutJERKQ6QnZm4ZybA+wN0uyHwD+AXaGqQ0RETp9nfRZm1hy4DHjqJKuTzCzXzD4zs0uD7Gd8oG1uQUFBSGoVEYl1XnZwPwL83Dl3soFPzgjc93st8IiZta1oJ865ic65HOdcTmZmZqhqFRGJaV7eOpsDTA88sp4BjDCzYufcG865bQDOufVmNhvoBazzrFIRkRjn2ZmFc661c66Vc64V8Bpwq3PuDTOrb2aJAGaWAQwClntVp4iIhPDMwsxeAoYBGWa2FbgPiAdwzj1dyaadgGfMrJSyMPuDc65KYZGXl7fbzDYFXqYBB8qtDvY6A9hdlfc5Bd9+r5rerrJ2Fa072fKqLCv/OpTHrKJ6amqbYO0i9bhFymftZMuj9f9ouH/WzgiyvoxzLiq/gInVfJ1bW7XU9HaVtato3cmWV2VZ+dehPGanetxq4phF8nGLlM9asGNUweuIPG7R8lmL5ie436zm61A61feq6naVtato3cmWV2VZuB+3mjhmla0P9+MWKZ+1ky2PtM9aVbeLis9aVI06ezrMLNdVYeRF+Q8ds1Oj43ZqdNyqryaPWTSfWVTXRK8LiEA6ZqdGx+3U6LhVX40dM51ZiIhIUDqzEBGRoBQWIiISlMJCRESCUlhUgZkNMbOnzew5M/vU63oigZn5zOy3Zva4mY31up5IYWbDzOyTwOdtmNf1RAozSwkMKDrS61oihZl1CnzOXjOzW4K1j/qwqGgSJjMbbmarzGytmf2isn045z5xzk0A3gKeD2W94aAmjhlwCZAFnAC2hqrWcFJDx80Bh4EkYuC41dAxA/g58Epoqgw/NfRzbUXg59pVlA2rVPl7RvvdUGY2lLL/fC8457oGlvmB1cB5lP2HXAhcA/iB339rFzc653YFtnsFGOecO1RL5XuiJo5Z4Gufc+4ZM3vNOXcFUa6Gjttu51ypmTUGHnLOfa+26vdCDR2zHkBDygJ2t3Purdqp3js19XPNzC4GbgGmOef+Xtl7ejnqbK1wzs0xs1bfWtwPWOucWw9gZtOBS5xzvwdOehprZi2BA9EeFFAzxywwHlhR4GVJ6KoNHzX1WQvYBySGos5wUkOftWFACtAZKDSzd9zJpz6IGjX1WXPOzQRmmtnbQGyHRQWaA1vKvd4K9A+yzThgSsgqCn/VPWYzgMfNbAgwJ5SFhblqHTczGwWcD6RTNi1xLKrWMXPO/T8AM7uewJlZSKsLX9X9rA0DRlH2S8k7wXYeq2FRbc65+7yuIZI4545SFrBSDc65GZQFrVSTc26q1zVEEufcbGB2VdtHfQd3BbYBLcq9zgosk4rpmJ0aHbfq0zE7NSE9brEaFguBbDNrbWYJwGhgpsc1hTsds1Oj41Z9OmanJqTHLerDIjAJ03ygg5ltNbNxzrli4HbgPWAF8IpzbpmXdYYTHbNTo+NWfTpmp8aL4xb1t86KiMjpi/ozCxEROX0KCxERCUphISIiQSksREQkKIWFiIgEpbAQEZGgFBYS1czscC2/X43MdxKY1+KAmS02s5Vm9ucqbHOpmXWuifcX+TaFhUg1mFml46k55wbW4Nt94pzrCfQCRppZsDkHLqVs5FWRGqewkJhjZm3NbJaZ5QVmpesYWH6RmX1uZovM7MPAnBKY2f1mNs3M5gHTAq8nm9lsM1tvZneU2/fhwJ/DAutfC5wZ/M3MLLBuRGBZnpk9ZmaVzr/gnCsEFlM2qihmdpOZLTSzfDP7h5klm9lA4GLgT4GzkbYVfZ8ip0JhIbFoIvBD51wf4KfAXwPL5wJnOud6AdOBn5XbpjNwrnPumsDrjpQNJd4PuM/M4k/yPr2AOwPbtgEGmVkS8AxwQeD9M4MVa2b1gWz+M9T7DOdcX+dcD8qGdRjnnPuUsnGA7nbO9XTOravk+xSpNg1RLjHFzOoCA4FXA7/ow38mGcoCXjazpkACsKHcpjMDv+F/7W3n3HHguJntAhrz39OgLnDObQ2872KgFWWzm613zn2975eA8RWUO8TM8ikLikecc18Flnc1s99QNudFXcrGAqrO9ylSbQoLiTU+YH+gL+DbHqdsKtOZgYlh7i+37si32h4v9/cSTv5/qSptKvOJc26kmbUGPjOzV5xzi4GpwKXOufzAhD/DTrJtZd+nSLXpMpTEFOfcQWCDmV0JYGV6BFan8Z/x/8eGqIRVQJtyU2JeHWyDwFnIH4CfBxalAjsCl77Kz9F9KLAu2PcpUm0KC4l2yYEhnL/++jFlP2DHBS7xLAMuCbS9n7LLNnnA7lAUE7iUdSswK/A+h4ADVdj0aWBoIGR+BXwOzANWlmszHbg70EHfloq/T5Fq0xDlIrXMzOo65w4H7o56EljjnHvY67pEKqMzC5Had1Ogw3sZZZe+nvG4HpGgdGYhIiJB6cxCRESCUliIiEhQCgsREQlKYSEiIkEpLEREJCiFhYiIBPX/AQoZ1oYTY2WtAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 205
        },
        "id": "LT5oFgUGpime",
        "outputId": "c9f5e066-d655-4499-c991-ab0181541e2b"
      },
      "source": [
        "learn_ret.fine_tune(3, 1e-4, freeze_epochs=1) #"
      ],
      "execution_count": 37,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: left;\">\n",
              "      <th>epoch</th>\n",
              "      <th>train_loss</th>\n",
              "      <th>valid_loss</th>\n",
              "      <th>COCOMetric</th>\n",
              "      <th>time</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <td>0</td>\n",
              "      <td>0.878591</td>\n",
              "      <td>0.724653</td>\n",
              "      <td>0.075082</td>\n",
              "      <td>01:36</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: left;\">\n",
              "      <th>epoch</th>\n",
              "      <th>train_loss</th>\n",
              "      <th>valid_loss</th>\n",
              "      <th>COCOMetric</th>\n",
              "      <th>time</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <td>0</td>\n",
              "      <td>0.637661</td>\n",
              "      <td>0.599846</td>\n",
              "      <td>0.192262</td>\n",
              "      <td>01:41</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>1</td>\n",
              "      <td>0.532042</td>\n",
              "      <td>0.485529</td>\n",
              "      <td>0.353599</td>\n",
              "      <td>01:39</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>2</td>\n",
              "      <td>0.460732</td>\n",
              "      <td>0.447929</td>\n",
              "      <td>0.412514</td>\n",
              "      <td>01:38</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 338
        },
        "id": "xp6GIV5npnIs",
        "outputId": "58fbdd2d-9156-4015-8b89-b6093c97065d"
      },
      "source": [
        " plot_metrics(learn_ret, '1st run- Mean Average Precision and Losses for Retinanet/Resnet50', 0, 0.01)"
      ],
      "execution_count": 38,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 439
        },
        "id": "19fAjrjjptg3",
        "outputId": "89099a29-1aa5-4dee-b110-548e1ce1c7e1"
      },
      "source": [
        "# 2nd run\n",
        "model_type_ret = models.mmdet.retinanet #\n",
        "backbone_r50 = model_type_ret.backbones.resnet50_fpn_1x(pretrained=True)#\n",
        "model_ret = model_type_ret.model(backbone=backbone_r50(pretrained=True), #\n",
        "                      num_classes=len(parser.class_map))\n",
        "\n",
        "train_dl_ret = model_type_ret.train_dl(train_ds, batch_size=16, num_workers=4, shuffle=True)#\n",
        "valid_dl_ret = model_type_ret.valid_dl(valid_ds, batch_size=16, num_workers=4, shuffle=False)#\n",
        "\n",
        "learn_ret = model_type_ret.fastai.learner(dls=[train_dl_ret, valid_dl_ret], model=model_ret, metrics=metrics)#\n",
        "\n",
        "learn_ret.lr_find()"
      ],
      "execution_count": 39,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/mmcv/cnn/utils/weight_init.py:119: UserWarning: init_cfg without layer key, if you do not define override key either, this init_cfg will do nothing\n",
            "  'init_cfg without layer key, if you do not define override'\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "Use load_from_local loader\n",
            "The model and loaded state dict do not match exactly\n",
            "\n",
            "size mismatch for bbox_head.retina_cls.weight: copying a param with shape torch.Size([720, 256, 3, 3]) from checkpoint, the shape in current model is torch.Size([333, 256, 3, 3]).\n",
            "size mismatch for bbox_head.retina_cls.bias: copying a param with shape torch.Size([720]) from checkpoint, the shape in current model is torch.Size([333]).\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              ""
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "SuggestedLRs(lr_min=0.00012022644514217973, lr_steep=0.00013182566908653826)"
            ]
          },
          "metadata": {},
          "execution_count": 39
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 205
        },
        "id": "Xy48pK-apw2G",
        "outputId": "1984dac1-a568-4f28-a9b0-ccac916896c6"
      },
      "source": [
        "learn_ret.fine_tune(3,0.000120 , freeze_epochs=1) #"
      ],
      "execution_count": 40,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: left;\">\n",
              "      <th>epoch</th>\n",
              "      <th>train_loss</th>\n",
              "      <th>valid_loss</th>\n",
              "      <th>COCOMetric</th>\n",
              "      <th>time</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <td>0</td>\n",
              "      <td>0.879835</td>\n",
              "      <td>0.755599</td>\n",
              "      <td>0.071730</td>\n",
              "      <td>01:34</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: left;\">\n",
              "      <th>epoch</th>\n",
              "      <th>train_loss</th>\n",
              "      <th>valid_loss</th>\n",
              "      <th>COCOMetric</th>\n",
              "      <th>time</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <td>0</td>\n",
              "      <td>0.629463</td>\n",
              "      <td>0.605008</td>\n",
              "      <td>0.219114</td>\n",
              "      <td>01:41</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>1</td>\n",
              "      <td>0.506781</td>\n",
              "      <td>0.450517</td>\n",
              "      <td>0.402248</td>\n",
              "      <td>01:38</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>2</td>\n",
              "      <td>0.446536</td>\n",
              "      <td>0.420366</td>\n",
              "      <td>0.455418</td>\n",
              "      <td>01:38</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 358
        },
        "id": "ZwwShmx3pzA1",
        "outputId": "bf3b0335-700f-4c42-e406-9961d3ad7c24"
      },
      "source": [
        " plot_metrics(learn_ret, '2nd run- Mean Average Precision and Losses for Retinanet/Resnet50', 0, 0.01)"
      ],
      "execution_count": 41,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAaoAAAFVCAYAAAC+S6s6AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nOzdd3xb9b3/8dfHe9vZy3b2Dpk2CRDCLpSyykhoKQVKgTJabtn93d7eXu6lzWCEsiFlFpqETRmlpWWvOJvsnTiOkzhxvLf1+f1xjh3Z8VJiWbL8eT4efiSSjo4+ko701vd8v+d7RFUxxhhjglVYoAswxhhjWmJBZYwxJqhZUBljjAlqFlTGGGOCmgWVMcaYoGZBZYwxJqh1aFCJyNUi8kVHPqbpGkQkXURKRCS8leWuEJF/dFRdR0tEdojImYGuoz2ISKyI/E1ECkXk1UDX01hn2Sa6shaDSkSiReTPIrJTRIpFZKWIfL+jivM3EXleRFRELmx0/UPu9VcHqC4RkW0isi4Qj9/e3Ne5yg2SfBH5p4iMas/HUNVdqpqgqrWtLPeyqn6vPR+7o7mv5/8Fug4fXAr0AXqo6mXHujIROVVEPO72VCwiG0Xkmjbed5D72Y6ouy5YtonmfpyIyI9E5BWv2kvcvx0ick8gavWq7YhtUUQ+EZEKrzo3Nrr9x26mlIrIWyLSvbXHaa1FFQFkA6cAycBvgcUiMsiXJ3M0vDckP9sE/LTR484EtnbQ4zdlBtAbGCIime298g58bb3NVdUEIBXYDzzfeAE3oG13dOgZCGxS1Rpf79jCtrrH3Z6SgF8Dz4jIyGOoMZj9AHjf63KK+9wvBf5LRM4KTFktusX94ZigqvXvi4iMBZ4CrsT58VIGPN7q2lTVpz9gNXCJ+/9Tgd3A7ThfPrnANV7L9gDeAYqAJcD/Al80s95BgALXAruAz+rW32i5HcCZ7v9/DywGXgSKgbVAhg/P5XngfmAf0M297jzgA+AL4GqvZX8GrAcOAR8CA71uexgn0IuAZcDJXrf5XCPwLPAy8AbwqHtdf6Ac6O613CTgABDZhhoVuBnYDGxvQ92xwAvuutYDd3m/F249rwN5wHbgV628zv/ndfkHQIn7/0+A+4Av3ec3DBgF/BPIBzYCMxvV9QCwEyh036dYr+0nwl3uamCb+5pvB67wuv4Lr/WdCGS568oCTvS67ROcbfZLdz3/AHo28xy7Ae+6r8ch9/+pbV0Xzgd3J3AQ+E+8tvPWXs9Gt10HbHFfu3eA/u71AjyE8zktAr4Dxrm3nQusc+vKAe7wWt95wEqgAPgKGO91293u8sXu+3RGE/X8D1AFVAMlOJ/vMJwfvTvdel4Ekpv7Hmhinady5PfCfuAy9/9hwD04PzYP4nz+uru37XLXX+L+ndDENqHAL3A+KwXAY4C4tw0F/u2u9wDO5zSl0ffTHTjfk4XAIiCmtdcTeAnw4HwGSoC7vJ7LPqAnjbZx9/YlwJ2tfU+18v4/7z7H99z38ltgqNc6m/w8Ate772uVW/PfvLb1nzezff4BeMXr8lD3/oktfie29UvdXWkfoAIY5bXB1AD3ApE4G3wZh7/0F7obSTwwDmejbi2oXnSXj6VtQVXhPm448EfgGx+ez/PA/wFPAze61y0GfoRXUAEX4nz4R+O0Mn8LfOW1np/ghHIETmjvxd04fa0RiHM3pHOBS3A+DFHubf8GrvNadh7wZBtrVHdj6w7EtqHu2cCnOF/AqTgfvN1eH55lwO+AKGAITiic3doXK5AAvAJ87rVR7wLGunUk44TnNe7lujAe4y7/mHufAe7reSIQjdeHGGf7KQJGuvfpB4x1/3817jbovhaHcEIiwn3fD+HsoqqrbSswAmd7/ASY3cxz7OG+X3FAIvAq8JbX7c2uCxiD80Gf4T6XB3E+Vz4FFXC6+1pNdtfzCO4XPXC2+56l4HxpjQb6ubfl4v5Icd/vye7/J+F8sU11X+urcD5/0cBI932qC8JBeH25Narr98BfGn2ZbsHZbhJwfpC91Nz3QBPrO5WG2+IFOF/yk9zrbgW+wdluo3F+wf+10fq9v+zrtwmvz8q77muVjvPj4xz3tmHAWe56e+H8oJ7f6PtpCc4Pue44ofGL1l7Pxt9tXuubBnzdVO3ubWXAD1v7Dmjl/X8eJ3iPd+/3MrDQvS2elj+Pz9NoW8TZtvPc5b4ETvW67W3g7kbLlwBTWvyu9uFLPRL4CHiq0QZT3uhN3+++gOE4aTvK67Y/0HpQDWlqg2y0IXgH1Udet40Byn14Ts/jBNV04Gv3TdyH80XiHVQfANd63S/M3UAGNrPeQ8CEo6kRJzzy3I0iBudXWd2G+HPg3+7/xd2AZrSlRve1Pb2V18O77gbB4z523ZfDVGBXo/v+Bniuhde5AudX5F6cX/pDvTbqe72WnYUbYl7XPQX8t/ucyutqbGb7qQuqApzgiG203NUcDqorgSWNbv/a633/BPit1203AX9v47Y1ETjU6MPb5LpwAn+h123xOL8yfQ2qP+PsYq27nIDzGRyEE2KbcD6bYY3utwu4AUhqdP0TwP82um4jTlfAMJzP+pm4LfoWXovf0zCo/gXc5HV5pFtnBE18DzSxvlNxgqkAqARqgf/wun09Xq07nB8qjdffWlBN97q8GLinmVouAlZ4Xd4B/MTr8lwO/5hs9vX0um/joPpf4L8abeMFOJ8DxdkjVNfaa/Y7oJX3/3lggdflc4ENrX0em9sWcb4fEnHC/CqcVlrd5/1fuMHttXwOXmHW1F+b+gPcfoOXcD48tzS6+aA23PdchvMB6cXhPq46O9vwcNmtL9LA3kaPHSMiEe5InrrOvA9aWoGqfuHW+5/Au6pa3miRgcDDIlIgIgU4TWDB+VWPiNwhIuvdUU0FOK2Cnq3V2Ew5VwGLVbVGVStwdq9d5d72OnCCiPTD+fXtAT5vS42uBq9tK3X3b7S89/8HAv3rHsu97//DaXE3535VTVHVvqp6gap69wE2XvfURuu+Aujr1hZDK/2HqlqK8wH7BZArIu81M3ijP0dukztp+Jo1fu8SmnpMEYkTkafcTuIinF/aKY1GITa3rgavtVv/weaeXwsaPB9VLXHXM0BV/w08itMi3S8iT4tIkrvoJThfTjtF5FMROcG9fiBwe6P3Ig2nFbUF+A+cENovIgtFpP/R1On+P4KG209r3wN7VDUFp4/qTzhfxHUGAm961bweJ8xa2j4ba/K9EpE+7nPNcd/nv9Dws97sfWnh9WyhjnNp2D+F+3gJOHtBTsVpRNStv8nvgFbe/9Zqbu7z2CRV/VZVi1W1UlVfwGlVneveXILznnlLwgmzZrUaVCIiOL/U+uD0TVW3dh9XHs7uizSv69LbcD/1+n8pzq6UulrCcQKl9ZU4I3nqOvPaMlLxLzhv/ItN3JYN3OB+0db9xarqVyJyMk7/zUycXZ4pOK0gaUud3kQkFecD9xMR2Ssie3E6TM8VkZ6qeginb2MW8GOcX+F1r1ezNXq/LF6P1VrduTi7Tup4v4/ZOP1c3o+VqKrncnS83/Ns4NNG605Q1RtxdiVU4OzXbnmFqh+q6lk4v6Y3AM80sdgenA+it3ScX3i+uh2nZTBVVZNwfkhA27aDXLxeXxGJw9mV6KsGz0dE4t315ACo6p9UdQpOq34EcKd7fZaqXogzgOctnBYEOO/FfY3eizhV/at7v1dUdbr7mArMOZo6cV7zGpy9GXWUNlDVSpy+suNE5CKvur/fqO4YVc1p63pb8Ad3Hce57/NPaPtnvcXXs3FtItIXZ/td3nhFqlqrqg/ifB5u8lp/s98Bzb3/bai5uc/jETU3Qzn8Gq0FJng9xyE4La9NLa2gLS2qJ3D2Z57fREuj+cqcYcJvAL93f22O4XDLoK024bQ+fiAikTj7XKN9XEdb/Qln3/NnTdz2JPAbd8QKIpIsInXDbBNxPmR5QISI/I4jfzG01ZU4z3kkzq6jiTgb1G6c/hNw+nd+ihNgr7Sxxqa0Vvdid33dRGQADVvSS4BiEblbnGNkwkVkXDuNUHwXGCEiV4pIpPuXKSKjVdWDM9DkQRHp7z7uCSLSYJtwf/Ve6H5RV+L8ivM08Vjvu4/1Y7cVPgvnQ/zuUdSdiLM7psAdbvvfPtz3NeA8EZkuIlE4fb6tfTbDRSTG6y8K+CtwjYhMdF+TPwDfquoO9zWc6n6OSnG+4DwiEuXufUh2f4QWcfi1egb4hXs/EZF497OYKCIjReR093Eq3Ofe1GvclL8CvxaRwSKS4Na5SI9iVCCAqlbhDLD5nXvVk8B9IjIQQER6yeFDUPLcOocczWPhvM8lQKH7uWjLl32dZl9P9/Z9jer6Ps7u4ZbCYDZwl4jE0MJ3QHPvfxtqbvbz2FTNIpIiIme722SEiFyB86Pt7+4iLwPni8jJ7ufzXuANVT36FpX7Rt+A84W5Vw7vSruiDU8QnC+3BJxm5fPAc228HwCqWojza2EBzq/CUpwv7Xanqvmq+q+mNgpVfRPn1+JCt7m/BmcjAmdkzd9xAmYnzgbg6+7LOlcBj6vqXu8/nA2wLuTfAYYDe1V1VRtrbEprdd+L81pvx+mbfA3nS7/uR8h5ONvFdpyWzgKcXYfHxN1gvwdcjvPLe6/7vOrC6A6cEUtZOLs25nDkdhwG3ObePx+nT+XGRsugqgfd53E7zi6yu4DzVPXAUZQ+H6dv8wBOR/7fW168QR1rcUZkvoLTujpE69v5PTjhUPf3b1X9CPgvnF3EuTgtz8vd5ZNwvigPcXh04Tz3tiuBHe528wucXTuo6lKcUYSPuvfbgtOfA877Mdt9vntxWmO/aeNTfhanK+EznO2nAvhlG+/b0jrTReR8nNGs7wD/EJFinPdjqvucynBHmbq7sqb5+Dj/gzNYpRBnlNwbbb1jK68nOAOtfuvWdQdHDktvynvuuq5r5Tugpfe/pZpb+zz+GRjj1vwWzm7I/+PwYIpfAhep6iZ3fWtxtrGXcfo4EzncImyWtBzWxjhE5EbgclU9JdC1GBPqxOnD3oszqKQo0PUEmh1caZokIv1E5CQRCRPnQMrbgTcDXZcxXUR3nNF+XT6kwFpUphnubt/3gME4w2EXAr9x+wOMMabDWFAZY4wJarbrzxhjTFCzoDLGGBPUAjGLdrvo2bOnDho0KNBlGGNMp7Js2bIDqtqmiROCRacNqkGDBrF06dJAl2GMMZ2KiLRlKrugYrv+jDHGBDULKmOMMUHNgsoYY0xQs6AyxhgT1CyojDHGBDULKmOMMUGtywXVil2H2HGgNNBlGGOMaaNOexzV0fr939axKruAaUO6c3lmOueM60tMZHjrdzTGGBMQnXZS2oyMDD2aA373Flbw2rJsFi/dza78MpJiIrho0gBmZqQxbsAxn/fPGGOCmogsU9WMQNfhiy4XVHU8HuWb7QdZlJXNB2v2UlXjYdyAJGZlpHHBxAEkx0a2Y7XGGBMcLKg60LEGlbfCsmreXpXDwiXZrMstIjoijHOP68fMjDSmDemOiLTL4xhjTKBZUHWg9gwqb2tyClmYtYu3V+6huKKGgT3imJmRxqVTUumTFNPuj2eMMR3JgqoD+Suo6pRX1fLBmlwWZWXz7fZ8wsOE00b2YmZGGqeN6k1keJcbMGmMCQEWVB3I30HlbfuBUhYvzea1ZbvJK66kV2I0l0xOZVZmGoN7xndIDcYY0x4sqDpQRwZVnZpaDx9vzGNRVjYfb9xPrUc5fnB3ZmWkce5x/YiNsmHuxpjgZkHVgQIRVN72F1Xw2vLdLM7KZsfBMhKjI7hgYn9mZaZx3IBkG4BhjAlKFlQdKNBBVUdV+XZ7Pouzsnnvu1wqazyM7pfErIxULpo0gJS4qECXaIwx9SyoOlCwBJW3wvJq3lm1h8VZ2XyXU0hURBjnjO3L5ZlpTBvSg7Awa2UZYwLLgqq5BxE5B3gYCAcWqOrsJpaZCfweUGCVqv64pXUedVCtfhXCwmDMRRDmvz6ltXsKWZyVzZsrciiqqCGteywzp6RxaUYq/ZJj/fa4xhjTEguqph5AJBzYBJwF7AaygB+p6jqvZYYDi4HTVfWQiPRW1f0trfeog+qF82H7Z9BzBJx8B4y7BML9N+VhRXUtH67dy8Il2Xy97SBhAqeM6MWszDTOGN3HhrkbYzqUBVVTDyByAvB7VT3bvfwbAFX9o9cyc4FNqrqgres96qDy1MK6t+Gz+2H/Wug+BKbfBhMuh3D/Tpu082Apry7dzavLstlXVEnPhCgunpzKzIw0hvVO8OtjG2MMWFA1/QAilwLnqOrP3ctXAlNV9RavZd7CaXWdhLN78Peq+veW1nvMfVQeD2x8Dz6dC3tXQ0o6TP81TLwCIqKPfr1tUFPr4bPNeSxcks2/N+ynxqNkDOzGrMw0fjC+H3FRXW5Se2NMB7GgauoB2hZU7wLVwEwgFfgMOE5VCxqt63rgeoD09PQpO3fuPPYCVWHzP5zAylkKSQPgpFth8k8h0v99SXnFlbyxfDeLsrLZdqCUhOgIzp/gDHOfkGrD3I0x7cuCqqkHaNuuvyeBb1X1Offyv4B7VDWrufW2+6g/Vdj2MXw6D3Z9BQl94MRfQcY1EOX/2SdUlaU7D7FwSTbvf5dLeXUtI/skMiszjR9OGkC3eBvmbow5dhZUTT2ASATObr0zgBycwRQ/VtW1XsucgzPA4ioR6QmsACaq6sHm1uvX4ek7voBP5ziDLuJ6wAm3wPHXQXSifx6vkeKKav62KpdFWbtYtbuQqPAwzhrbh8sz0zhpaE8b5m6MOWoWVM09iMi5wHyc/qdnVfU+EbkXWKqq74izf+sB4BygFrhPVRe2tM4OOY5q17fw2VzY8hHEpMC0m2DqDRCb4t/H9bI+t4hFWdm8tTKHgrJqBqTEMjMjjcsyUumfYsPcjTG+CemgEpHeOIMd+gPlwBqcoPH4r7zmdegBvznLnF2Cmz6A6CQnrKbdBHHdO+bxcYa5/2PdPhZnZfPFlgOIwIzhzjD3M0f3ISrChrkbY1oXkkElIqcB9wDdcXbJ7QdigBHAUOA14AFVLfJvqQ0FZGaK3NXw2TxY/w5EJUDmtXDCLyGhV4eWkZ1fxqtLs3l12W5yCyvoHh/FxZMGMCszjeF9Omb3pDGmcwrVoJoHPKKqu5q4LQI4DwhX1df9U2LTAjqF0v71znFYa16HiBhnwMWJv4Kkfh1aRq1H+WxzHouzsvlo/T6qa5VJ6SlcnpnGeeP7Ex9tw9yNMQ2FZFAFq6CY6+/AZvj8AVi9GMIinCHt0/8DklM7vpSSSt5cnsOipdls2V9CXFQ454/vz8zMNCanp9gwd2MMEOJBJSJ9gD8AA1T1HBEZA5ygqn/2Z4HNCYqgqpO/Hb54EFa+AghM/DGcfBt0G9Thpagqy3cdYlFWNu+uzqWsqpbhvRPqh7n3SPDvwczGmOAW6kH1AfAc8J+qOsHd7bdCVY/zZ4HNCaqgqlOQDV/Oh+UvOlM1TbgcTr4degwNSDkllTW8u2oPi5Zms2JXAZHhwllj+jArM53pw3oSbsPcjelyQj2oslQ1U0RWqOok97qVqjrRrxU2IyiDqk7RHvjyT7DsOaitcia+PfkO6D0qYCVt2lfMoqxs3li+m0Nl1fRPjuHSjDQum5JKWve4gNVljOlYoR5UnwCXAP9U1ckiMg2Yo6qn+LG+ZgV1UNUp2Q9fPQJZf4bqMhhzAcy4E/oGpBEKQGVNLR+t28+ipdl8vjkPgOnDejIzI43vje1DdIT/Tn1ijAm8UA+qycAjwDicY6h6AZeq6mr/lde8ThFUdUoPwjePwbdPQ1UxjPwBnHIn9J8U0LJ2HyrjtWW7eXXpbnIKyukWF8lF7jD3UX2TAlqbMcY/QjqooH44+khAgI2qWu2vwlrTqYKqTvkh+PYp+OZxqCiEYWfBKXdB2vEBLavWo3y55QCLsrL5x7q9VNcqE9Lqhrn3IzHGv6c/McZ0nJAOKhG5uImrC4HvWjvJoT90yqCqU1EES56Grx+D8nwYfIoTWIOmB7oy8kureHNFDouydrFpXwmxkeH8YHw/Ls9MY8rAbjbM3ZhOLtSD6j3gBOBj96pTgWXAYOBeVX3JHwU2p1MHVZ3KElj6rNOPVbofBp7k9GENORUCHAiqysrsAhZlZfO3VXsoraplSK94ZmWkcfHkVHol2jB3YzqjUA+qD4Gfquo+93If4EXgR8BnqjrOb1U2ISSCqk51OSx7wRnaXpwLqZlwyt0w7MyABxZAaWUN763OZdHSbJbtPEREmHDG6N5cnpnOjBG9bJi7MZ1IqAfVOlUd43VZgLWqOsZ7yHpHCamgqlNdASv/Al/Mh8JsZ7DFjDth5LlBEVgAW/bXDXPP4WBpFX2TYrgsI5WZGWk2zN2YTiDUg+pxIB141b3qEmA3cCfwrqqe5pcKmxGSQVWnpgpWL3SmZzq0A/qMgxl3wOgLISw4ZkmvqvHwr/X7WLQ0m8825eFROHFoD2ZlpnH22L7ERNowd2OCUagHleCE00nuVV8Cr2uAJgsM6aCqU1sDa15zZmw/uAV6jXIOHB53MYQFTxDsKSjntWW7Wbw0m92HykmOjeSHkwYwMyONMf1tmLsxwSSkgyrYdImgquOphbVvOjO2562H7kOdqZnGz4Tw4Bk67vEoX209yKKl2Xy4Zi9VtR6OG5DMrMw0LpjYnyQb5m5MwIV0ULnD0+cAvXGOoxJAVTUgP5m7VFDV8Xhgw7vOWYf3fgcpA2H6r2HiFRARFejqGjhUWsVbK3NYlJXNhr3FxESGce5x/ZiVkcbxg7vbMHdjAiTUg2oLcL6qrvdvSW3TJYOqjips+jt8Ohf2LIekVOf0IpOuhMiYQFfXgKqyenchi5Zm887KPZRU1jCkZzyXZaRxyZQB9E4MrnqNCXWhHlRfqupJrS/ZMbp0UNVRha3/gk/nQfY3kNAXTvoVTLkGooJvBF5ZVQ3vf7eXxVnZLNmRT3iYcPqo3szKSOPUkb2ICA+OgSLGhLJQD6qHgb7AW0Bl3fWq+oZ/SmuZBZUXVdjxudPC2vE5xPWEE38JmddCdHCemn5rXgmLl2bz+rLdHCipondiNJdOcYa5D+oZH+jyjAlZoR5UzzVxtarqz9q3pLaxoGrGzq+dPqyt/4bYbjDtZph6PcQkB7qyJlXXevj3hv0szsrm44378ShMG9KdWZlpfH9cPxvmbkw7C+mgCjYWVK3YvdQZ1r7p7xCdDFNvgGk3Qlz3QFfWrL2FFby+fDeLsrLZlV9GYkwEF010ZnMfNyA4g9aYziakg0pEYoBrgbFAfQ+4taiC3J6VTmBteBeiEuD46+CEWyC+Z6Ara5bHo3yz/SCLs7J5f81eqmo8jO2fxKzMNC6cMIDkOBvmbszRCvWgehXYAPwYuBe4Alivqrf6r7zmWVD5aN9a5zistW9CZCxk/Mzpx0rsG+jKWlRYVs3bq3JYuCSbdblFREeE8f1xfZmVmc60ITbM3RhfhXpQrVDVSSKyWlXHi0gk8LmqTvNviU2zoDpKeRudqZm+exXCImHKVXDSf0DygEBX1qo1OYUszNrF2yv3UFxRw8AecczMSOPSKan0SbJh7sa0RagH1RJVPV5EPgNuAvYCS1R1iD8LbI4F1TE6uBW+eBBWLQQJcw4anv5r6DYw0JW1qryqlg/W5LIoK5tvt+cTJnDayN7MykzjtFG9ibRh7sY0K9SD6ufA68BxwPNAAvBfqvqU36prgQVVOzm00zm9yIq/gHpg/OVw8m3QY2igK2uT7QdKWbw0m9eW7SavuJKeCXXD3FMZ0ish0OUZE3RCNqhEJAy4VFUX+7+ktrGgameFOfDlw7D8BaitguMucybA7TUi0JW1SU2th0825rHQHeZe61GOH+QMcz/3uH7ERtkwd2MghIMKQESWBtOTs6Dyk+J98NWfnDMPV5fD2Iucc2L1GRvoytpsf1EFry3fzeKsbHYcLCMxOoLzJ/bn8sw0jhuQbAMwTJcW6kE1GzgALAJK665X1Xz/lNYyCyo/Kz0AXz8GS56GqhIYdZ4TWP0nBrqyNlNVvt2ez+KsbN77LpfKGg+j+iZyeWYaF00aQEpccE3ka0xHCPWg2t7E1WqDKUJcWT58+yR88yRUFsLws+GUuyC1U23nFJZX886qPSzOyua7nEKiIsI4Z2xfZmWmccKQHoSFWSvLdA0hHVTBxoKqg1UUOq2rrx+D8kMw5DQnsAaeGOjKfLZ2TyGLs7J5c0UORRU1pHWPZeaUNC7NSKVfcmygyzPGr0I6qNzjpm4EZrhXfQI8parV/imtZRZUAVJZDFl/hq8fhdI8GHSys0tw8AzoZH0/FdW1fLh2L4uysvlq60HCBE4Z0YtZmWmcPqoPURE2zN2EnlAPqgVAJPCCe9WVQK2q/txPtbXIgirAqspg2fPOSMGSvZA2FWbcBcPO6HSBBbDzYCmvLt3Nq8uy2VdUSY/4KC5xZ3Mf1tuGuZvQEepBtUpVJ7R2XUexoAoS1RWw4iX4Yj4U7Yb+k51dgiPO6ZSBVVPr4bPNeSxcks2/N+ynxqNkDOzGzMw0zhvfj7ioiECXaMwxCfWgWg5cpqpb3ctDgNdUdbIf62uWBVWQqamCVa/A5w9CwU7oe5yzS3DU+RDWOXeh5RVX8oY7m/u2A6UkREdw/oR+zMxIY2Jaig1zN51SqAfVGcBzwDZAgIHANar6sf/Ka54FVZCqrXbmEfzsfsjfCr1Gw4w7YOwPIaxzHnSrqizdeYhFWdm8tzqX8upaRvZJZGZmGj+cNIDu8TbM3XQeIR1UACISDYx0L25U1cqWlvcnC6og56mFNW/A5/dD3gboMRxOvt2Z8SK88+4+K66o5m+rclmUtYtVuwuJCg/jrLF9mJWRxvRhPW2Yuwl6IR9UwcSCqpPweGD9O845sfatgW6DYPptMOFHENG5WyLrc4tYlJXNWytzKCirZkBKLJdlpHLRxAEM7BFnuwZNULKgau5BRM4BHgbCgQWqOrvR7VcD84Ac96pHVXVBS+u0oOpkPB7Y9AF8OhdyV0JyGpx0K0y6EiI79yk6Kqpr+ce6fSzOyuaLLQcASIqJYEz/JMb2T2as++/QXvFE2MzuJs64h8EAACAASURBVMAsqJp6AJFwYBNwFrAbyAJ+pKrrvJa5GshQ1Vvaul4Lqk5KFbZ85ATW7iWQ2M8JrMlXQVRcoKs7Ztn5ZXy6KY91uUWs3VPEhtwiKms8AERFhDGqbyJj+ycxxg2w0X2TbMJc06G6VFCJSD8gv7V+KhE5Afi9qp7tXv4NgKr+0WuZq7Gg6lpUYfun8Ok82PkFxPdyzjiccS1Eh85xSzW1HrYdKGXtnkLW5hTVB1hhuXOcfJjAkF4JjOmXVN/yGts/iW42QMP4SWcMqmPp1X4JGCoir6vqHS0sNwDI9rq8G5jaxHKXiMgMnNbXr1U1u4llTKgQgSGnOn87voTP5sI/f+ccj3XCzXD89RCTFNga20FEeBgj+iQyok8iP5zkXKeq5BSUs3aPE1rr9hSydEc+76zaU3+//skxjOmf7O4+dP4GpMRav5fpko5p1584n5oxqrq2hWUuBc6pm8FCRK4Epnq3nkSkB1CiqpUicgMwS1VPb2Jd1wPXA6Snp0/ZuXPnUdduglD2EmfQxeZ/QEwyTL0Rpv0CYrsFurIOkV9axbo9RU7ra4/T+tqWV4LH/YimxEXWt7zq+r+G9LR+L+Obztii8nV4+nRguKo+JyI9gURVbWpWde/7tLrrr9Hy4Ti7FJNbWq/t+gthe1Y4uwQ3vgdRiTD1eph2M8T3CHRlHa6sqoYNe4vrW15r9xSxYW8xVW6/V3REGKPqwsv9d5T1e5kWhHRQich/AxnASFUdISL9gVdV9aRW7heBszvvDJxRfVnAj71bYSLST1Vz3f//ELhbVae1tF4Lqi5g7xqnhbXubYiMg8yfwQm/hMQ+ga4soGpqPWzNK61vea3dU8i6PUUUVdQATr/X0F4J9X1edbsP7fxbBkI/qFYCk4DlqjrJvW61qo5vw33PBebjDE9/VlXvE5F7gaWq+o6I/BG4AKgB8oEbVXVDS+u0oOpC9m+Azx+ANa9BeBRMudoZKZjUP9CVBQ1VZfeh8gYtr3W5ReQWVtQvMyAl1qvPywmw/skx1u/VxYR6UC1R1eNFZLmqThaReODrtgSVP1hQdUEHtzqBtWqhMx3TpCth+n9ASnqgKwtaB0sq60ca1rW+th8ope5j3y0ussHxXmP6JTGkVwLhNsNGyAr1oLoDGI5zPNQfgZ8Br6jqI/4rr3kWVF3YoR3wxUOw4mVAnVkuTr4NugfkZNOdTmml0++1rn7XYREb9xZTVev0e8VEhjGqb8Ph8iP7JhITaf1eoSCkgwpARM4CvoczKe2HqvpPfxXWGgsqQ+FuZzj78hfBUwPjZzrzCfYcHujKOp3qWg9b80pYm3O45bUut4hit98rPEwY2iv+cMurfxJj+yWTHBcZ4MqNr0I+qIKJBZWpV5QLXz0CS5+FmgoYdzGcfAf0GRPoyjq1w/1ehQ12He4rOnyM/4CU2AYtr7EDkuibZP1ewSykg0pEioG6haNwzvZbqqoBOSrTgsocoSQPvn4UljwD1aUw+nznrMP9AtKNGrIOlFS6x3sdHnG4/eDhfq/u8VFHHO81uGe89XsFiZAOqgZ3cn4uXQhMU9V72r2qNrCgMs0qy4dvHodvn4LKIhjxfTjlThgwJdCVhazSyhrW59bNtFHE2txCNu0tqe/3io0MZ1S/xAatrxF9rN8rELpMUNXfWWRF3VD1jmZBZVpVXgBLnoavH4OKAhh6BpxyF6S3eIieaSdVNR627C+p7+9au6eI9XuKKK483O81vLczz+EYryHzybHW7+VPIR1UInKx18UwnIN/T1HVE/xRWGssqEybVRRB1gJnt2DZQRh0MpxyNwya7sw5aDqMx6NkHyo73PJy+7/2Fx/u90rrHsvYfskNjvnqkxRt/V7tJNSD6jmvizXADuAZVd3vh7paZUFlfFZVCkufg6/+BCX7IP0EmHEnDD3dAivA8oorD89x6B6svP1Aaf3tPeKjGrS6xvZPYnCPeDuj8lEI6aAKNhZU5qhVl8Pyl+DL+VCUAwMynMAacbYFVhApqev3yjk808amfcVU1zrfWXFR4YxuMM9hMiP6JhAdYf1eLQnJoBKRRzg82u8Iqvqr9i6qLSyozDGrqYSVL8PnD0HhLug73unDGvkDCLMZyYNRVY2HzfuLD7e83AArcfu9IsKEYb0TGhzvNaZ/Ekkx1u9VJ1SD6qqWblfVF9q1ojayoDLtprYaVi9ypmfK3wa9x8CMO2DMRc5UTSaoeTzKrvyy+uHydcd8HSg53O+V3j2u/rxedbsPeyd2zX6vkAyqI+4gkgSoqhb7p6S2saAy7a62Bta+4czYfmAT9BzhHDg87hIIP5ZzjJpA2F9c0WDQxro9Rew4WFZ/e8+EKMbUHajsBtjA7nEh3+8V0kElIhnAc0AizhRKBcDPVHWZ/8prngWV8RtPrXNqkc/uh/1rodtgZ2qmCZdDuO1C6syKK6pZn1vcoOW1eV8xNe7ZKeO9+r3qWl7D+4RWv1eoB9Vq4GZV/dy9PB143GZPNyHL44GN78NncyF3FSSnO7O1T/oJREQHujrTTipratm8r6TBcPn1uUWUVtUCEBkuDOud2KDlNbpfIomdtN8r1IPqiIN760754ZfKWmFBZTqMKmz+B3w6F3KWQmJ/J7Am/xQiYwNdnfEDj0fZmV/WoOW1bk8hB0qq6pcZ2COOxien7J0YE8Cq2yYkg0pE6oLop0As8FecUYCzgApVvc2vFTbDgsp0OFXY9jF8Og92fQXxveGkX0HGzyAqPtDVGT9TVfd4r4aDNnblH+736pUYXT/PYd3Iw/R26PdSVXJLc9lSsIWtBVv5yeifEHmUu6FDNag+buFmVdXT27ektrGgMgG14wv4dA5s/wziesAJN0PmdRATkDmaTQAVVVSzfk/Dk1Nu2V9S3++VEB3B6H6JDVpew3snEhVx5CEQqsq+sn1sLdhaH0pbC7aytXArpdWHD4B+68K3GJoy9KjqDcmgClYWVCYo7PrW6cPa8hHEpMC0m+D46yCue6ArMwFUUe32e+Uebnmtzy2irL7fC4b09dC/VyEJiQfwROwlvzqbnUXbKa4+PKC6R0wPhqUMY2jKUIamDK3/f3J08lHXFpJBJSI/AV7WZhYUkaFAP1X9wg/1NcuCygSVnGXOLsFNHziXoxIhsQ8k9IGE3pDQ1/k3sa/X5T5Oa8wOLg5ZqsrBioNsLdjK5kNbWLl3AxvyN5NbtoMqPdxC8tTEo1W9iZdUUuMHM6bncE5IG8vUgen0SmzfgTuhGlS34px2fpn7lwfEAMOAU4ADwD2qutm/pTZkQWWCUu5qp3VVsh9K9jr/Frv/VjVx6KGEu8HVOMyaCDgbuBHU8ivyG+yyq/u3oLKgfpnk6GSGJg9t0EpKCk9jz8Hww+f4yi0kO7+8/j69E6MbnNurrt/raA9WDsmgAhCRcOB04CSgH1AOrAc+UNVdfq2wGRZUptOpKnUmw/UOr5K9R15Xuh/Uc+T9o5PcAOvTQmutD8R2t1aaHxVWFh4RRlsKtpBfkV+/TGJk4hG764alDKNnbM82BUxhefXhA5VznYOWN+8vodbt9/rnr2cwvE/iUdUfskEVjCyoTMjy1DqnI2kpzOpaa1UlR94/LMIZkdhSmNX9RQb/cOpAKaoqarKFdKD8QP0y8ZHxh8PIq6XUO653u0/PVFFdy6Z9zjyHl01JJSL86H6MdMagsnlhjAk2YV67A1tTWeKG2L6mw6woB3KWQ2keTc4tHZ3sFWiNW2tef3HdQ3Zm+ZKqErYWbm0YSoe2sL/88BmMYiNiGZo8lJP6n9SghdQ3vm+HzRcYExnO+NQUxqemdMjjBRMLKmM6s+gE569HK0OVa2ucVlqDfrNGAZezzPl/ddmR9w+LdMOzuV2P7nXxvYO2lVZWXdYwjAqdf/eW7q1fJiY8hiEpQ5jWf1qD3Xb94vsRJrY7NVAsqIzpCsIjnCBJ7NP6spXFzYdZ8V4o3O3M0FF6gCZbaTEpzbfMvK+L7eaXVlp5TTnbCrcdcSxSTklO/TJRYVEMSRnClD5TGuy2G5A4wAIpCLU5qNzRf88BxcACYBLOaL9/+Kk2Y0wgRCc6f21ppZXmHQ6xur60Yq9w253lXK4pP/L+YZFNB1r95b6Hd4E2MbdiRU0F2wu3NwijLQVbyCnJQd0AjQyLZFDyIMb3Gs/Fwy+ubyWlJqQSbqdw6TR8aVH9TFUfFpGzgW7AlcBLgAWVMV1ReAQk9XP+WqJ6uJXWVJiV7IOCXZC9BMoOHHH3KmB7Yg+2JnRjS0wsW8PD2EIVu2vL8LiBFCHhDEpMZ2yPsVww7IL6XXbpielEhNmOo87Ol3ewro1+LvCSqq6VrnjWMWOMb0ScqaVikqDnsGYXq66tZkfBFrbuW8GWA+vYWriVLSV7yK46RC0KVBGuVaRXwcjKCn5QWcnQqiqGVVeTXl1DJNsh/JvDfWf1B1c30VqL7w0RUR33Gphj4ktQLRORfwCDgd+ISCLQxMEexhjTvGpPNdlF2fW77DYXbGZrwVZ2Fe2iRp1TyodJGOmJ6QztO4nveQ1qGJQ0iKjwKLeVVnRky8y7tZa/DXZ97QwiaUps92bCzHv2kN4QkxyyIx47C1+C6lpgIrBNVctEpDtwjX/KMsZ0djWeGrKLs484FmlH0Q5qPE4gCUJqYipDU4ZyRvoZ9X1Ig5IHER3ewtRBIk6AxCRDrxEtF1Jb7e52bCLM6v52fe1cV1t55P0jYloIM+/+td52Yk0/8SWoTgBWqmqpO//fZOBh/5RljOksaj215JTkHNFC2lG4gyrP4fM3DUgYwLCUYcxInVHfQhqcPJjYCD9PDRUeCckDnL+WqEJFYfNhVrIPDm6FnV9BeX7T64jr0cS8jk0EXHSStdJ84EtQPQFMEJEJwO04I/9exJnvzxgT4jzqIack54hh39sKt1Hp1RLpH9+foSnOwbF1LaTByYOJi4wLYPVtIAKxKc5fr5EtL1tT5Ux11VSg1V0+uMX5t7bqyPtHxDYRZk1MXBzfyxm00sX58grUqKqKyIXAo6r6ZxG51l+FGWMCw/skfd677LYXbqfca5h5n7g+DEsZRmbfzAaTrMZHdoGTSEZEQXKq89cSVagoaCLM9h7eHXlgs3N+s/JDTaxAnFZa436zE26BhF5+eWrByJegKhaR3+AMSz9ZRMIA2yFrTCdVd5K+xnPZbS3YSlmN11lrY3sxLGUYlwy/pEEgJUYd3aSoXYqIc2BzbDfoParlZWsqW+9Ly9vk/Dv1ho6pP0j4ElSzgB/jHE+1V0TSgXn+KcsY015UlbzyPKeFdGiLM+y7YAvbCrZRUn14Utu6k/RdNOyidjtJn/FBRDSkpDl/LemkE4kfizYHlRtOLwOZInIesERVX/RfacYYX9SdpK+pU1AUe50Lq1t0N4Z1G8Z5Q85rMMFqSkzXm+y0U+qCgzB8mUJpJk4L6hOcg38fEZE7VfU1P9VmjGlGfkU+Ww417EPaWriVwsrC+mXqTtL3/UHfb9BC6hHbI4CVG+M7X3b9/SeQqar7AUSkF/ARYEFljJ8UVBQcEUZbC7Y2eZK+swae1aCF1COmR4edgsIYf/IlqMLqQsp1ELBpho1pB4WVhUcMaNhSsIWDFYdnVag7Sd+paaf6/SR9xgQTX4Lq7yLyIfBX9/Is4IP2L8mY0FVcVdwgiOr+39RJ+qYPmM7wbsPrW0h94vpYIJkuyZfBFHeKyMXAdPeqp1X1zbbcV0TOwZnFIhxYoKqzm1nuEpxdiZmqaueZN52W90n6vHfd7SvbV7+MnaTPmLbx6ZBnVX0DeKPusojsUtX0lu4jIuHAY8BZwG4gS0TeUdV1jZZLBG4FvvWlJmOCRVl1GQu+W8B7295jT+me+uujw6MZnDyYjL4ZDEsZVh9IAxLsJH3GtMWxzs3Rlv0QxwNbVHUbgIgsBC4E1jVa7n+BOcCdx1iTMR3Kox7e2/YeDy17iLzyPE5NO5VLRlxiJ+kzpp0ca1C15cizAUC21+XdwFTvBURkMpCmqu+JSLNBJSLXA9cDpKe32JAzpkN8l/cds7NmszpvNcf1PI6HTnuICb0mBLosY0JKq0ElIrc1dxOQcKwFuFMxPQhc3dqyqvo08DRARkZG1zs82wSNvLI85i+fzztb36FnbE/um34f5w05z3blGeMHbWlRtTShV1tO85EDeM8Jkupe573+ccAn7oimvsA7InKBDagwwaaqtoqX1r3E06ufptpTzbXjruW68dd1jYlYjQmQVoNKVf/nGB8jCxguIoNxAupynDkD69ZfCPSsuywinwB3WEiZYKKqfJL9CfOWziO7OJvT0k7jzow7SUtqZV42Y8wx8/uJTlS1RkRuAT7EGZ7+rKquFZF7gaWq+o6/azDmWGwt2MqcJXP4OvdrhiYP5amznuLE/icGuixjuowOOSOXqr4PvN/out81s+ypHVGTMa0prCzkiVVPsHDDQuIi47jn+HuYOXImkWF2dhtjOpKdOtKYRmo9tby++XUeWfEIRVVFXDbiMm6eeDPdYroFujRjuqSjDir3TL97VdUO0DUhI2tvFrOXzGbToU1k9s3k7sy7Gdm9ldOSG2P86lhaVFOB40QkQlW/314FGRMIOSU5PLD0Af6585/0j+/Pg6c+yJnpZ9rcesYEgaMOKlX9f+1ZiDGBUFZdxrNrnuX5tc8TJmHcMvEWrhp7FTERMYEuzRjj8uXEib2Bk4D+QDmwBmfUnsdPtRnjN6rK+9vf58FlD7K/bD/nDj6XX0/5NX3j+wa6NGNMI22ZmeI04B6gO7AC2A/EABcBQ0XkNeABVS3yZ6HGtJe1B9cyZ8kcVuxfwZgeY7j/lPuZ1HtSoMsyxjSjLS2qc4HrVHVX4xtEJAI4D2dm9NfbuTZj2tWB8gM8suIR3tz8Jt1iunHvifdy4bALbdojY4JcW2amaHaSWFWtAd5q14qMaWfVtdW8suEVnlz1JBW1FVw19ipuGH8DCVHHPFWlMaYD+NJHFQ1cAgzyvp+q3tv+ZRnTPj7b/Rlzs+ays2gnM1JncGfGnQxKHhTosowxPvBl1N/bQCGwDKj0TznGtI9thduYlzWPL3K+YFDSIB4/43FOTj050GUZY46CL0GVqqrn+K0SY9pBUVURT616ilfWv0JsRCx3ZtzJj0b/yKY9MqYT8yWovhKR41T1O79VY8xRqvXU8uaWN3lkxSMcqjjExcMv5peTfkmP2B6BLs0Yc4x8CarpwNUish1n158Aqqrj/VKZMW20bN8y5iyZw/r89UzuPZknz3yS0T1GB7osY0w78SWobJokE1RyS3J5aNlDfLDjA/rG92XejHmcPehsm/bImBDT5qBS1Z0iMh0YrqrPiUgv2uFU9Mb4qrymnOfXPM+za55FUW6ccCPXjLuG2IjYQJdmjPEDX4an/zeQAYwEngMigb/gTKtkjN+pKh/u/JAHlz5Ibmku5ww6h9um3Ea/hH6BLs0Y40e+7Pr7ITAJWA6gqntEJNEvVRnTyIb8DcxeMptl+5Yxqvso/jD9D2T0zQh0WcaYDuBLUFWpqoqIAohIvJ9qMqZefkU+j6x4hNc3vU5KdAq/O+F3XDzsYsLDwgNdmjGmg/gSVItF5CkgRUSuA34GPOOfskxXV+2pZuGGhTyx8gnKa8q5YvQV3DjxRpKikgJdmjGmg/kymOJ+ETkLKMLpp/qdqv7Tb5WZLuvLnC+ZkzWH7YXbOan/SdyVeRdDUoYEuixjTID4euLETTjHTn0kInEikqiqxf4ozHQ9O4t2Mi9rHp/u/pT0xHQePf1RZqTOsOHmxnRxvoz6uw64Hue8VEOBAcCTwBn+Kc10FSVVJTy9+mleWv8S0eHR3D7ldq4YfQWR4TbtkTHGtxbVzcDxwLcAqrrZPeuvMUfFox7e3vI2Dy9/mPyKfC4adhG/mvwresb2DHRpxpgg4ktQVapqVd1uGPekieqXqkzIW7l/JbOXzGbtwbVM6DWBx854jLE9xwa6LGNMEPIlqD4Vkf8HxLqDKm4C/uafskyo2lu6l/nL5/PetvfoHdubP578R34w+AfWD2WMaZYvQXU38HPgO+AG4H1ggT+KMqGnsraSF9a+wILvFlDrqeX68ddz7bhriYuMC3Rpxpgg16agEpFwYK2qjsKOnTI+UFX+tetf3L/0fnJKcjhr4FncNuU2UhNTA12aMaaTaFNQqWqtiGwUkXRV3eXvokxo2Ji/kblZc1mydwnDuw1nwfcWMLXf1ECXZYzpZHzZ9dcNWCsiS4DSuitV9YJ2r8p0agUVBTy68lFe3fQqiVGJ/Hbqb7lkxCVEhPl62J4xxvgWVP/ltypMSKjx1LB442IeW/kYpdWlXD7ycm6aeBPJ0cmBLs0Y04n5MoXSp/4sxHRuX+/5mrlZc9lSsIVp/aZxd+bdDOs2LNBlGWNCgC8zUxRz5HFThcBS4HZV3daehZnOIbsom/uX3s+/s/9NakIqD5/2MKelnWbDzY0x7caXXX/zgd3AK4AAl+NMpbQceBY4tb2LM8GrtLqUBd8t4IW1LxARFsGtk2/lyjFXEh0eHejSjDEhxpegukBVJ3hdflpEVqrq3e6BwKYL8KiHd7e9y/xl88krz+OCoRdw6+Rb6R1ns2kZY/zDl6AqE5GZwGvu5UuBCvf/NpVSF7A6bzVzlsxh9YHVHNfzOB467SEm9JrQ+h2NMeYY+BJUVwAPA4/jBNM3wE9EJBa4xQ+1mSCRV5bH/OXzeWfrO/SM7cl90+/jvCHnESZhgS7NGNMF+DLqbxtwfjM3f9E+5ZhgUlVbxYvrXuSZ1c9Q7anm2nHXct3464iPjA90acaYLqTVoBKR3wKPq2p+M7efDsSp6rvtXZwJDFXl4+yPuX/p/WQXZ3Na2mncmXEnaUlpgS7NGNMFtaVF9R3wNxGpwBnhlwfEAMOBicBHwB/8VqHpUFsLtjJnyRy+zv2aoclDeeqspzix/4mBLssY04W1GlSq+jbwtogMB04C+gFFwF+A61W1vLV1iMg5OP1b4cACVZ3d6PZf4JyYsRYocde7zsfnYo5BYWUhT6x6goUbFhIXGcc9x9/DzJEziQyzs+waYwLLlz6qzcBmXx/AnXn9MeAsnOOwskTknUZB9IqqPukufwHwIHCOr49lfFfjqeH1Ta/z6MpHKaoq4rIRl3HzxJvpFtMt0KUZYwzg26i/o3U8sKVu5goRWQhcCNQHlaoWeS0fjw137xBZe7OYvWQ2mw5tIrNvJndn3s3I7iMDXZYxxjTQEUE1AMj2urwbOOJcDyJyM3AbEAWc3tSKROR64HqA9PT0di+0q8gpyeGBpQ/wz53/pH98fx489UHOTD/Tpj0yxgQlX+b666mqB/xViKo+BjwmIj8Gfgtc1cQyTwNPA2RkZFiry0dl1WX8ec2feX7N84SHhXPLxFu4auxVxETEBLo0Y4xpVluGp5+PM5dfjYjUAjNV9SsfHiMH8B7XnOpe15yFwBM+rN+0QlV5f/v7PLjsQfaX7efcwefy6ym/pm9830CXZowxrWpLi+o+4GRV3SAiU4G5wCk+PEYWMFxEBuME1OXAj70XEJHh7mANgB9wFIM2TNPWHlzL7G9nszJvJWN6jOH+U+5nUu9JgS7LGGParC1BVaOqGwBU9VsRSfTlAVS1RkRuAT7EGZ7+rKquFZF7gaWq+g5wi4icCVQDh2hit5/xzYHyA/xp+Z94a8tbdIvpxr0n3suFwy60aY+MMZ1OW4Kqt4jc1txlVX2wtRWo6vvA+42u+53X/29tQx2mDaprq3l5/cs8ufpJKmsruWrsVdww/gYSohICXZoxxhyVtgTVM0BiM5dtQEOQUFU+z/mcuVlz2Vm0k1NST+GOjDsYlDwo0KUZY8wxacvMFP/T3G0iktm+5Zijsa1wG3Oz5vJlzpcMShrEE2c+wfQB0wNdljHGtAufj6MSkTHAj9y/AiCjvYsybVNUVcSTq57kr+v/SmxELHdm3MmPRv/Ipj0yxoSUNgWViAzicDhVAwOBDFXd4a/CTPNqPbW8ueVNHlnxCIcqDnHx8Iv55aRf0iO2R6BLM8aYdteW46i+BpJwjm+6RFU3i8h2C6nAWLZvGXOWzGF9/nom957Mk2c+yegeowNdljHG+E1bWlT7cKZB6gP0wjnGyQZRdLDcklweXPYgf9/xd/rG92XejHmcPehsm/bIGBPy2jKY4iIRSQYuBn7vnu4jRUSOV9Ulfq+wiyuvKef5Nc/z7JpnUZQbJ9zINeOuITYiNtClGWNMh2hTH5WqFgLPAc+JSB9gJvCQiKSrqp321Q9UlQ93fsiDSx8ktzSXcwadw21TbqNfQr9Al2aMMR3K51F/qroPeAR4REQGtn9JZv3B9cxeMpvl+5czqvso/jD9D2T0tcGVxpiuyZfZ0zOA/8QZ8ed9v/HtXVRXlV+RzyMrHuH1Ta+TEp3C7074HRcPu5jwsPBAl2aMMQHjS4vqZeBO4DvA459yuqZqTzULNyzkiZVPUF5Tzk/G/IRfTPgFSVFJgS7NGGMCzpegynMnkDXt6IucL5ibNZfthds5qf9J3JV5F0NShgS6LGOMCRq+BNV/i8gC4F9AZd2VqvpGu1fVBews2sm8rHl8uvtT0hPTefT0R5mROsOGmxtjTCO+BNU1wCggksO7/hSwoPJBSVUJT69+mpfWv0R0eDS3T7mdK0ZfQWS4TXtkjDFN8SWoMlV1pN8qCXEe9fD2lreZv3w+hyoOcdGwi/jV5F/RM7ZnoEszxpig5ktQfSUiY1R1nd+qCVEr96/kj0v+yLqD65jYayKPn/E4Y3uODXRZxhjTKfgSVNOAlSKyHaePSgBVVRue3oy9pXt5aNlDvL/9fXrH9Wb2ybM5d/C51g9ljDE+8CWozvFbFSGmoqaCF9a+wJ/X/JlaTy3Xj7+ea8ddS1xkXKBLM8aYTqfNQaWqO/1ZSChQVT7a9REPLH2AnJIczhp4C47tdgAAFRBJREFUFrdNuY3UxNRAl2aMMZ2Wz1MomaZtzN/I3Ky5LNm7hOHdhrPgewuY2m9qoMsyxphOz4LqGBVUFPDoykd5ddOrJEYl8tupv+WSEZcQEWYvrTHGtAf7Nj1K1Z5qFm9czOMrH6e0upTLR17OTRNvIjk6OdClGWNMSLGgOgpf7/maOUvmsLVwK9P6TePuzLsZ1m1YoMsyxpiQZEHlg+yibOYtncfH2R+TmpDKw6c9zGlpp9lwc2OM8SMLqjYorS7lmdXP8OK6F4kIi+DWybdy5ZgriQ6PDnRpxhgT8iyoWuBRD+9ue5f5y+aTV57HBUMv4NbJt9I7rnegSzPGmC7DgqoZq/NWM2fJHFYfWM1xPY9j/mnzGd/LJuEwxpiOZkHVSF5ZHvOXz+edre/QM7Yn902/j/OGnEeYhAW6NGOM6ZIsqFyVtZW8tO4lnln9DNWeaq4ddy3Xjb+O+Mj4QJdmjDFdWpcPKlXl4+yPmZc1j90luzkt7TTuzLiTtKS0QJdmjDGGLh5UWw5tYU7WHL7J/YahyUN56qynOLH/iYEuyxhjjJcuGVSFlYU8vvJxFm1cRFxkHPccfw+zRs6yaY+MMSYIdalvZlVl8cbFPLryUYqqirhsxGXcPPFmusV0C3RpxhhjmtGlgkpEWLpvKcO7DefuzLsZ2X1koEsyxhjTii4VVAD3nnQvMeExNu2RMcZ0El0uqGIjYgNdgjHGGB/YUazGGGOCmgWVMcaYoGZBZYwxJqh1SFCJyDkislFEtoj8//buPMqOsszj+PeXCIkQJqwiwkAIwjAgW2BAIMQgjmEZNokDiBwiqOCoIzKjowdllOG44YKIjCyHCYhKZHMQoxCWEBBJCJCVTQioMEgQAQ1CFHjmj/dpUmnv7b7d9L1dnfw+59zTb721PfVW3fvct6q6rj7VYPwpku6VtEDSjZK26ERcZmZWf21PVJKGA98GDgC2A46WtF23ye4BdouIHYErgK+0Oy4zMxsaOtGj2h14KCKWRMSfgcuAQ6sTRMTNEfGnHLwD2KwDcZmZ2RDQiUS1KfCbyvBjWdfMCcBPG42Q9EFJcyXNfeqppwYwRDMzq6ta/R+VpPcCuwFvazQ+Is4Hzs9pn5L0q36uakPgd/2ct50cV984rr5xXH2zqsY15O4B6ESiehyo/mbGZlm3EknvAE4F3hYRy3tbaERs1N+AJM2NiN36O3+7OK6+cVx947j6xnHVRydO/d0JbC1pS0lrAkcB11QnkLQLcB5wSEQs7UBMZmY2RLQ9UUXES8BHgOuA+4AfRsRiSadLOiQnOxMYBVwuaZ6ka5oszszMVjMduUYVEdOB6d3qTquU39GJOCrO7/D6WuW4+sZx9Y3j6hvHVROKiMGOwczMrCk/QsnMzGrNicrMzGptlUtULTxXcISkaTl+tqQxlXGfzvoHJE3qcFxNn3co6eW8yWTAbzRpIa4p+T9rXet/f2XccZJ+ma/jOhzXNyoxPSjp2cq4trSXpIskLZW0qMl4STo7Y14gaVxlXDvbqre4jsl4Fkq6XdJOlXGPZv08SXM7HNdESc9V9tVplXE97v82x/WJSkyL8nhaP8e1s73+VtLN+TmwWNLHGkwzKMfYoIuIVeYFDAceBsYCawLzge26TfMvwHeyfBQwLcvb5fQjgC1zOcM7GNe+wFpZ/lBXXDm8bBDbawpwToN51weW5N/1srxep+LqNv1HgYs60F4TgHHAoibjD6Q8VUXAW4HZ7W6rFuPaq2t9lGduzq6MexTYcJDaayJw7Wvd/wMdV7dpDwZu6lB7bQKMy/I6wIMN3o+DcowN9mtV61H1+lzBHL44y1cA+0lS1l8WEcsj4hHgoVxeR+KKwXneYSvt1cwkYEZE/D4ingFmAPsPUlxHAz8YoHU3FRGzgN/3MMmhwCVR3AGsK2kT2ttWvcYVEbfneqGDz9Jsob2aeS3H5UDH1ZFjCyAinoiIu7P8R8q/83R/3NygHGODbVVLVK08V/DVaaL8j9dzwAYtztvOuKq6P+9wpMozDu+QdNgAxdSXuI7I0wxXSOp6ykgt2itPkW4J3FSpbld79aZZ3O1sq77qfmwFcL2kuyR9cBDi2VPSfEk/lbR91tWivSStRfmwv7JS3ZH2UrkksQswu9uooXCMDbhaPevPmj7vcIuIeFzSWOAmSQsj4uEOhfRj4AcRsVzSiZTe6Ns7tO5WHAVcEREvV+oGs71qS9K+lEQ1vlI9PtvqDcAMSfdnj6MT7qbsq2WSDgR+BGzdoXW34mDg5xFR7X21vb0kjaIkx5Mj4g8DueyhalXrUbXyXMFXp5H0OmA08HSL87YzrurzDg+JyvMOI+Lx/LsEmEn5ptWRuCLi6UosFwK7tjpvO+OqOIpup2ba2F69aRZ3O9uqJZJ2pOy/QyPi6a76SlstBa5m4E539yoi/hARy7I8HVhD0obUoL1ST8dWW9pL0hqUJPW9iLiqwSS1PcbaarAvkg3ki9JDXEI5FdR1EXb7btN8mJVvpvhhlrdn5ZspljBwN1O0EtculAvIW3erXw8YkeUNgV8yQBeWW4xrk0r5cOCOLK8PPJLxrZfl9TsVV063LeXitjrRXrnMMTS/OeAgVr7QPafdbdViXJtTrrnu1a1+bWCdSvl2YP8OxvXGrn1H+cD/dbZdS/u/XXHl+NGU61hrd6q9ctsvAc7qYZpBO8YG8zXoAQz4BpW7Yh6kfOifmnWnU3opACOBy/ONOwcYW5n31JzvAeCADsd1A/AkMC9f12T9XsDCfLMuBE7ocFxfBBbn+m8Gtq3Me3y240PA+zoZVw5/DvhSt/na1l6Ub9dPAH+hXAM4ATgJOCnHi/Jr1g/nunfrUFv1FteFwDOVY2tu1o/Ndpqf+/jUDsf1kcqxdQeVRNpo/3cqrpxmCuXmqup87W6v8ZRrYAsq++rAOhxjg/3yI5TMzKzWVrVrVGZmtopxojIzs1pzojIzs1pzojIzs1pzojIzs1pzojIbBPnk8GsHOw6zocCJyszMas2JyqwHkt4raU7+/tB5koZLWpa/h7VY5bfDNsppd84H4S6QdLWk9bL+zZJuyIev3i1pq1z8qHzQ7/2SvpdP8TezbpyozJqQ9PfAkcDeEbEz8DJwDOXxOXMjYnvgFuA/c5ZLgP+IiB0pTw3oqv8e8O2I2Iny5Iwnsn4X4GTKb6GNBfZu+0aZDUF+erpZc/tRHsJ7Z3Z2Xg8sBV4BpuU0lwJXSRoNrBsRt2T9xcDlktYBNo2IqwEi4kWAXN6ciHgsh+dRnj93W/s3y2xocaIya07AxRHx6ZUqpc92m66/zyFbXim/jN+PZg351J9ZczcCk/O3h5C0fv5Q4zBgck7zHuC2iHgOeEbSPll/LHBLlF9qfazrBxwljcgf5DOzFvkbnFkTEXGvpM9QftF1GOVp2x8Gngd2z3FLKdexAI4DvpOJaAnwvqw/FjhP0um5jHd3cDPMhjw/Pd2sjyQti4hRgx2H2erCp/7MzKzW3KMyM7Nac4/KzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqzYnKzMxqbbVNVJKW1SCGMZIWDfAyz5L0uKRhlbopkp6SNE/SvZI+UBl3mKTTBjKGvpD0VUlvb2G6wyRt14/lHyLpU/2Yrx375gpJY7Pc8PiTNFXS5H4ufwdJU19DiM2W+2pbSNpN0tlNpntU0oY9LGdA33N5rE8YyGVaPa22iWpVlMnpcOA3wNu6jZ4WETsDE4EvSNo46z8JnNuHdbxuAEKt+hbQSiI5DGiYqHqKKSKuiYgv9TO2ASNpe2B4RCxp1zoiYiGwmaTN27iOuRHxr+1afqskbQC8NSJm9WGe4W0MydrIiapC0laSfibpLkm3Stq2Un+HpIWSzqh+M5T0CUl3Slog6fNZN0bSfZIukLRY0vWSXp/jdpU0X9J84MMtxrVM0pm5rBsk7S5ppqQlkg6pTDoRWAz8N3B0o2VFxFLgYWALSdsAyyPidz1tp6SJ2R7XAPdKGp7xdG33if1tj4j4FbCBpDf2sP17AYcAZ2avcKvc/rMkzQU+JulgSbMl3ZNttHHOO0XSOVmeKulsSbdn27XUc5E0UtL/ZLvcI2nfrN9e0pyMaYGkrSWtLeknuY8XSToyF3MM8L/dlvuNbI8bJW3UYL2v9lCyJzMzy2tLuijXfY+kQyuz/Rg4qpftuUzSQZXhqZIm5366VdLd+dqrwbwTJV2b5Q1yXy6WdCGg3lsTVJyZ7bOwq40kbSJpVrbnIkn75LE2tTLtx3MxRwA/qyxzv2yLhdk2Iypt+GVJdwPvlvSBPD7nS7pS0lqVNvirY0PSMEnnSrpf0gxJ0yvjdpV0i8rnxXWSNmll+60fImK1fAHLGtTdCGyd5T2Am7J8LXB0lk/qmhd4J3A+5Q06LKebAIwBXgJ2zul+CLw3ywuACVk+E1iU5TcB05vEGsABWb4auB5YA9gJmFeZ7gLgWOBvgMeBNbJ+CnBOlscCS4H1gfcBX6vM32w7JwLPA1vm8AeBz2R5BDAX2LI/7VGJ+4he9tdUYHJleCZwbmV4PUBZfn/XdnXb9qnA5RnbdsBDPaxvTGXf/BtwUZa3BX4NjKT0Bo/J+jWB11M+QC+oLGd0/r0F2KHbPu2a97RuMU7O8qPAhlneDZiZ5S+w4nhaF3gQWDuH9wZ+3EtbHg5cXIn7Nxn7WsDIrN8amNugLSYC12b5bOC0LB+U27Rhb++5bKMZwHBg42zPTbKdT81phgPrALsCMyrLWDf/XgwcnOWRuQ3b5PAlwMmVNvxkZf4NKuUzgI/2dGwAk4HpWf9G4JmsWwO4Hdgopzuy6xjxa+BfA30aZ8iSNArYC7hcevWL4Yj8uyfl1BPA94GvZvmd+bonh0dR3uC/Bh6JiHlZfxcwRtK6lDda1+mK7wIHAETE/wEHNgnvz6z49riQ0gv6i6SFlA8RJK2Z858SEX+UNBuYREkWAEdKGg8sB06MiN/nN8CnKutptp0AcyLikcp271jpkYzO7e5Te1SWvZSSqPtqWqW8GTAtt2lN4JHGs/CjiHiF0jPcuMk03Y2nJCUi4n5JvwK2AX4BnCppM+CqiPhl7pOvSfoy5QP91lxG97Z+pRL/pcBVLcYCpY0PkfTvOTwS2By4j9ba8qfAN7PXsT8wKyJekDQaOEfSzsDLuY09mQC8CyAifiLpmRbjHw/8ICJeBp6UdAvwD8CdwEWS1qDsp3mSlgBjJX0L+AnlSxqs3J5/Rzm+HszhiylnK87K4epx8hZJZ1AS/Cjgusq4RsfGeODyrP+tpJsr63wLMCM/L4YDT7S4/dZHTlQrDAOejXIdp1UCvhgR561UKY2hJIQuL1O+sfbXXyK/tlE+4JYDRMQrWnF9ZhLlzbcw3zhrAS+wIlFNi4iPdFvuC5Qk04rnK2VRvolW3+RImkT/2mNkxtJX1Zi+BXw9Iq6RNBH4XJN5qnG0dKqqmYj4fn4hOAiYLunEiLhJ0jjKl4YzJN0YEadTtm9kT4trUPcSK07PV+cVpQf6QIN5em3LiHgxTyNOovQELstRHweepPTUhwEv9rScgRYRs1RujjgImCrp6xFxiaSdMtaTgH8Gjqf39qyqHidTgcMiYr6kKZQeYpe+HBsCFkfEni3GYK+Br1GliPgD8Iikd8Or59F3ytF3UE5XwMrn/68Djs/eGJI2lfSGHtbxLPBs9mygXLcYKEcD74+IMRExhnIq7h+7zsE3cR/w5spws+3s7jrgQ/nNF0nbSFqbPrZHxTZA111lX5R0eINp/kg5FdTMaMrpToDjWlhnX9xK7iuV63qbAw+o3MG3JCLOplx/2lHSm4A/RcSllFO743IZ3dt6GOUUEsB7gNsarPdRyqkvWLFfoLTzR5XfSCTtUhlXbcvdJV3SZJumUU797sOK3vpo4InsPRxL6SX0ZFbGjqQDKKdfW3ErpYc/XOXa3ARgjqQtgCcj4gLgQmCcyjW6YRFxJfAZGrfnA5QzFl3Dx1JOtTayDvBEHrutvP9+DhyR16o2ZkViewDYSNKeAJLWULlhxtpgdU5Ua0l6rPI6hXLgnqByo8NioOsi9cnAKZIWUN4czwFExPWUU2S/yFM+V9DzhymUD4dvS5pH5VubpDdJmt6fDclktD/l1AgZ2/OUD7+De5h1FrBL1wceTbazgQuBe4G7VW5bPg94XX/aIz8w3ky5zgWwA/DbBpNeBnwiL5hv1WD85yinbe8CftfTOvvhXGBYbtM0YEpELKd8u1+U+/ItlGsjO1A+dOcB/0m5DgJl30ysLPN5YPdsv7cDpzdY7+cpp+jmUnqhXf6Lco1kgaTFOdxlX1YcB5vTvHd1PeXO0Bsi4s+V7Twuj/9tWbkn0sjngQkZw7sop3hbcTXlWu184CbKNaTfUtpnvqR7KD29bwKbAjOzPS8FPp3LeLU9I+JFyvvq8txHrwDfabLuzwKzKQno/hZivRJ4jHK8XwrcDTyXbTYZ+HK21zzKpQNrA604o2TNZCJ4ISJC0lGUGw4O7W2+oUDSNykX328YjO3M3tO4iPhsDl8XEZPauc7BoHKX483A3nltph3rGEHpSYyPiJcknQl8NyIWtGN9g03SbcA/5ZmKdq5nVEQsU7klfg5lHzb6MmVt4kTVAkn7AOdQekDPAsdHxEODG9XAyNMZe+S1nY5vZ55qndHuD5s6yGt490VEqz2Pvi5/a2DTiJjZjuXXjaQ9KF+s2pqI83reupSbdL4SEVPbuT77a05UtlqTtAPl7suq5RGxx2DEM9Rlr+PGBqP2i4inOx2PrRqcqMzMrNZW55spzMxsCHCiMjOzWnOiMjOzWnOiMjOzWnOiMjOzWvt/fLZgLPdmaqYAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7avoRYCJtvHW"
      },
      "source": [
        "# Finding:\n",
        "# resetting to baseline also resets lr_find"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 354
        },
        "id": "ytq4iB0qp3nK",
        "outputId": "b0659556-a191-4020-fa73-34715a0b7f33"
      },
      "source": [
        "# 3rd run\n",
        "model_type_ret = models.mmdet.retinanet #\n",
        "backbone_r50 = model_type_ret.backbones.resnet50_fpn_1x(pretrained=True)#\n",
        "model_ret = model_type_ret.model(backbone=backbone_r50(pretrained=True), #\n",
        "                      num_classes=len(parser.class_map))\n",
        "\n",
        "train_dl_ret = model_type_ret.train_dl(train_ds, batch_size=16, num_workers=4, shuffle=True)#\n",
        "valid_dl_ret = model_type_ret.valid_dl(valid_ds, batch_size=16, num_workers=4, shuffle=False)#\n",
        "\n",
        "learn_ret = model_type_ret.fastai.learner(dls=[train_dl_ret, valid_dl_ret], model=model_ret, metrics=metrics)#\n",
        "\n",
        "learn_ret.lr_find()"
      ],
      "execution_count": 42,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/mmcv/cnn/utils/weight_init.py:119: UserWarning: init_cfg without layer key, if you do not define override key either, this init_cfg will do nothing\n",
            "  'init_cfg without layer key, if you do not define override'\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "Use load_from_local loader\n",
            "The model and loaded state dict do not match exactly\n",
            "\n",
            "size mismatch for bbox_head.retina_cls.weight: copying a param with shape torch.Size([720, 256, 3, 3]) from checkpoint, the shape in current model is torch.Size([333, 256, 3, 3]).\n",
            "size mismatch for bbox_head.retina_cls.bias: copying a param with shape torch.Size([720]) from checkpoint, the shape in current model is torch.Size([333]).\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "error",
          "ename": "NameError",
          "evalue": "ignored",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-42-a18a6abc9a83>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m     10\u001b[0m \u001b[0mlearn_ret\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mmodel_type_ret\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfastai\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mlearner\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdls\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mtrain_dl_ret\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mvalid_dl_ret\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mmodel\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mmodel_ret\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mmetrics\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mmetrics\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;31m#\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     11\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 12\u001b[0;31m \u001b[0mlearn\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mlr_find\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
            "\u001b[0;31mNameError\u001b[0m: name 'learn' is not defined"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bER3g4BYp6BI"
      },
      "source": [
        "learn_ret.fine_tune(3, , freeze_epochs=1) #"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "LoZnsAmMp8Et"
      },
      "source": [
        " plot_metrics(learn_ret, '3rd run- Mean Average Precision and Losses for Retinanet/Resnet50', 0, 0.01)"
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}